Files
SmolLM2-MagpieUltra-8k/trainer_state.json

4426 lines
101 KiB
JSON
Raw Permalink Normal View History

{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.0,
"eval_steps": 500,
"global_step": 3125,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0016,
"grad_norm": 65.89872163766714,
"learning_rate": 4.792332268370606e-06,
"loss": 3.7957,
"step": 5
},
{
"epoch": 0.0032,
"grad_norm": 29.212265099142172,
"learning_rate": 9.584664536741212e-06,
"loss": 3.3022,
"step": 10
},
{
"epoch": 0.0048,
"grad_norm": 10.895903204228713,
"learning_rate": 1.4376996805111818e-05,
"loss": 2.333,
"step": 15
},
{
"epoch": 0.0064,
"grad_norm": 4.219156823970311,
"learning_rate": 1.9169329073482425e-05,
"loss": 1.6667,
"step": 20
},
{
"epoch": 0.008,
"grad_norm": 1.5171803020477856,
"learning_rate": 2.3961661341853032e-05,
"loss": 1.2733,
"step": 25
},
{
"epoch": 0.0096,
"grad_norm": 0.8329780378455843,
"learning_rate": 2.8753993610223637e-05,
"loss": 1.1109,
"step": 30
},
{
"epoch": 0.0112,
"grad_norm": 1.0434654832706145,
"learning_rate": 3.3546325878594245e-05,
"loss": 1.0074,
"step": 35
},
{
"epoch": 0.0128,
"grad_norm": 0.3300553023184043,
"learning_rate": 3.833865814696485e-05,
"loss": 1.0076,
"step": 40
},
{
"epoch": 0.0144,
"grad_norm": 0.2340031237394517,
"learning_rate": 4.313099041533546e-05,
"loss": 0.9492,
"step": 45
},
{
"epoch": 0.016,
"grad_norm": 0.2146710034764919,
"learning_rate": 4.7923322683706065e-05,
"loss": 0.9139,
"step": 50
},
{
"epoch": 0.0176,
"grad_norm": 0.189627899314042,
"learning_rate": 5.2715654952076676e-05,
"loss": 0.9043,
"step": 55
},
{
"epoch": 0.0192,
"grad_norm": 0.17145418468919713,
"learning_rate": 5.7507987220447274e-05,
"loss": 0.9286,
"step": 60
},
{
"epoch": 0.0208,
"grad_norm": 0.15072219389318597,
"learning_rate": 6.230031948881788e-05,
"loss": 0.8869,
"step": 65
},
{
"epoch": 0.0224,
"grad_norm": 0.1487582889759295,
"learning_rate": 6.709265175718849e-05,
"loss": 0.8893,
"step": 70
},
{
"epoch": 0.024,
"grad_norm": 0.15216037450481035,
"learning_rate": 7.18849840255591e-05,
"loss": 0.898,
"step": 75
},
{
"epoch": 0.0256,
"grad_norm": 0.14117550078296717,
"learning_rate": 7.66773162939297e-05,
"loss": 0.8824,
"step": 80
},
{
"epoch": 0.0272,
"grad_norm": 0.12582794694884158,
"learning_rate": 8.14696485623003e-05,
"loss": 0.8736,
"step": 85
},
{
"epoch": 0.0288,
"grad_norm": 0.1288847171575301,
"learning_rate": 8.626198083067092e-05,
"loss": 0.8658,
"step": 90
},
{
"epoch": 0.0304,
"grad_norm": 0.09710460790442074,
"learning_rate": 9.105431309904153e-05,
"loss": 0.8421,
"step": 95
},
{
"epoch": 0.032,
"grad_norm": 0.09519507016763937,
"learning_rate": 9.584664536741213e-05,
"loss": 0.8456,
"step": 100
},
{
"epoch": 0.0336,
"grad_norm": 0.09817656864973161,
"learning_rate": 0.00010063897763578275,
"loss": 0.8345,
"step": 105
},
{
"epoch": 0.0352,
"grad_norm": 0.07324878049182121,
"learning_rate": 0.00010543130990415335,
"loss": 0.8337,
"step": 110
},
{
"epoch": 0.0368,
"grad_norm": 0.08391462156230026,
"learning_rate": 0.00011022364217252396,
"loss": 0.8563,
"step": 115
},
{
"epoch": 0.0384,
"grad_norm": 0.06815110154684177,
"learning_rate": 0.00011501597444089455,
"loss": 0.8442,
"step": 120
},
{
"epoch": 0.04,
"grad_norm": 0.07029410799138884,
"learning_rate": 0.00011980830670926518,
"loss": 0.8389,
"step": 125
},
{
"epoch": 0.0416,
"grad_norm": 0.0695009528747182,
"learning_rate": 0.00012460063897763577,
"loss": 0.839,
"step": 130
},
{
"epoch": 0.0432,
"grad_norm": 0.06007797832529693,
"learning_rate": 0.00012939297124600637,
"loss": 0.8491,
"step": 135
},
{
"epoch": 0.0448,
"grad_norm": 0.06895549748289434,
"learning_rate": 0.00013418530351437698,
"loss": 0.8158,
"step": 140
},
{
"epoch": 0.0464,
"grad_norm": 0.06663703619650603,
"learning_rate": 0.0001389776357827476,
"loss": 0.8378,
"step": 145
},
{
"epoch": 0.048,
"grad_norm": 0.05800521787720269,
"learning_rate": 0.0001437699680511182,
"loss": 0.8347,
"step": 150
},
{
"epoch": 0.0496,
"grad_norm": 0.06821420625192964,
"learning_rate": 0.0001485623003194888,
"loss": 0.8326,
"step": 155
},
{
"epoch": 0.0512,
"grad_norm": 0.07637965717808766,
"learning_rate": 0.0001533546325878594,
"loss": 0.826,
"step": 160
},
{
"epoch": 0.0528,
"grad_norm": 0.06023982285545024,
"learning_rate": 0.00015814696485623,
"loss": 0.8247,
"step": 165
},
{
"epoch": 0.0544,
"grad_norm": 0.06451880405747862,
"learning_rate": 0.0001629392971246006,
"loss": 0.8549,
"step": 170
},
{
"epoch": 0.056,
"grad_norm": 0.056614991679051825,
"learning_rate": 0.00016773162939297124,
"loss": 0.8209,
"step": 175
},
{
"epoch": 0.0576,
"grad_norm": 0.05986673427938732,
"learning_rate": 0.00017252396166134184,
"loss": 0.8293,
"step": 180
},
{
"epoch": 0.0592,
"grad_norm": 0.06762202063357754,
"learning_rate": 0.00017731629392971245,
"loss": 0.8153,
"step": 185
},
{
"epoch": 0.0608,
"grad_norm": 0.058024285850389316,
"learning_rate": 0.00018210862619808305,
"loss": 0.7969,
"step": 190
},
{
"epoch": 0.0624,
"grad_norm": 0.06951587147435814,
"learning_rate": 0.00018690095846645365,
"loss": 0.8287,
"step": 195
},
{
"epoch": 0.064,
"grad_norm": 0.059967590804059515,
"learning_rate": 0.00019169329073482426,
"loss": 0.8184,
"step": 200
},
{
"epoch": 0.0656,
"grad_norm": 0.06888699815960402,
"learning_rate": 0.00019648562300319486,
"loss": 0.8448,
"step": 205
},
{
"epoch": 0.0672,
"grad_norm": 0.06331950295195966,
"learning_rate": 0.0002012779552715655,
"loss": 0.8341,
"step": 210
},
{
"epoch": 0.0688,
"grad_norm": 0.0682428678938053,
"learning_rate": 0.0002060702875399361,
"loss": 0.8385,
"step": 215
},
{
"epoch": 0.0704,
"grad_norm": 0.06942911129366544,
"learning_rate": 0.0002108626198083067,
"loss": 0.8041,
"step": 220
},
{
"epoch": 0.072,
"grad_norm": 0.06516484454506838,
"learning_rate": 0.0002156549520766773,
"loss": 0.8157,
"step": 225
},
{
"epoch": 0.0736,
"grad_norm": 0.0684316359036024,
"learning_rate": 0.0002204472843450479,
"loss": 0.805,
"step": 230
},
{
"epoch": 0.0752,
"grad_norm": 0.05851180156988952,
"learning_rate": 0.00022523961661341852,
"loss": 0.8208,
"step": 235
},
{
"epoch": 0.0768,
"grad_norm": 0.06721932048821487,
"learning_rate": 0.0002300319488817891,
"loss": 0.8234,
"step": 240
},
{
"epoch": 0.0784,
"grad_norm": 0.08197284968960065,
"learning_rate": 0.0002348242811501597,
"loss": 0.8359,
"step": 245
},
{
"epoch": 0.08,
"grad_norm": 0.05972037966198098,
"learning_rate": 0.00023961661341853036,
"loss": 0.831,
"step": 250
},
{
"epoch": 0.0816,
"grad_norm": 0.07027569430100532,
"learning_rate": 0.00024440894568690096,
"loss": 0.8359,
"step": 255
},
{
"epoch": 0.0832,
"grad_norm": 0.061373618615669416,
"learning_rate": 0.00024920127795527154,
"loss": 0.824,
"step": 260
},
{
"epoch": 0.0848,
"grad_norm": 0.06472732495649915,
"learning_rate": 0.00025399361022364217,
"loss": 0.8128,
"step": 265
},
{
"epoch": 0.0864,
"grad_norm": 0.06371406940567984,
"learning_rate": 0.00025878594249201275,
"loss": 0.8129,
"step": 270
},
{
"epoch": 0.088,
"grad_norm": 0.05573045766785217,
"learning_rate": 0.0002635782747603834,
"loss": 0.7984,
"step": 275
},
{
"epoch": 0.0896,
"grad_norm": 0.05708848530630277,
"learning_rate": 0.00026837060702875396,
"loss": 0.8258,
"step": 280
},
{
"epoch": 0.0912,
"grad_norm": 0.05553109724821234,
"learning_rate": 0.00027316293929712453,
"loss": 0.8044,
"step": 285
},
{
"epoch": 0.0928,
"grad_norm": 0.07317565452474585,
"learning_rate": 0.0002779552715654952,
"loss": 0.818,
"step": 290
},
{
"epoch": 0.0944,
"grad_norm": 0.06554274004318197,
"learning_rate": 0.0002827476038338658,
"loss": 0.8341,
"step": 295
},
{
"epoch": 0.096,
"grad_norm": 0.0589873509019811,
"learning_rate": 0.0002875399361022364,
"loss": 0.812,
"step": 300
},
{
"epoch": 0.0976,
"grad_norm": 0.07444838770982509,
"learning_rate": 0.000292332268370607,
"loss": 0.8219,
"step": 305
},
{
"epoch": 0.0992,
"grad_norm": 0.055086133002943516,
"learning_rate": 0.0002971246006389776,
"loss": 0.8363,
"step": 310
},
{
"epoch": 0.1008,
"grad_norm": 0.05244546721814248,
"learning_rate": 0.00029999962555314543,
"loss": 0.8158,
"step": 315
},
{
"epoch": 0.1024,
"grad_norm": 0.048906209160856015,
"learning_rate": 0.00029999541304750175,
"loss": 0.8005,
"step": 320
},
{
"epoch": 0.104,
"grad_norm": 0.05270153527283544,
"learning_rate": 0.00029998652010953066,
"loss": 0.8266,
"step": 325
},
{
"epoch": 0.1056,
"grad_norm": 0.0651045392890373,
"learning_rate": 0.000299972947016726,
"loss": 0.8204,
"step": 330
},
{
"epoch": 0.1072,
"grad_norm": 0.06037453132506936,
"learning_rate": 0.00029995469419262033,
"loss": 0.8275,
"step": 335
},
{
"epoch": 0.1088,
"grad_norm": 0.062334910502221307,
"learning_rate": 0.00029993176220677174,
"loss": 0.8121,
"step": 340
},
{
"epoch": 0.1104,
"grad_norm": 0.05436618680254399,
"learning_rate": 0.0002999041517747462,
"loss": 0.8016,
"step": 345
},
{
"epoch": 0.112,
"grad_norm": 0.046877341620779,
"learning_rate": 0.0002998718637580951,
"loss": 0.8192,
"step": 350
},
{
"epoch": 0.1136,
"grad_norm": 0.055756481803914806,
"learning_rate": 0.00029983489916432846,
"loss": 0.8133,
"step": 355
},
{
"epoch": 0.1152,
"grad_norm": 0.06734132796776768,
"learning_rate": 0.00029979325914688344,
"loss": 0.8092,
"step": 360
},
{
"epoch": 0.1168,
"grad_norm": 0.058298020977884744,
"learning_rate": 0.0002997469450050883,
"loss": 0.8194,
"step": 365
},
{
"epoch": 0.1184,
"grad_norm": 0.05630128783858887,
"learning_rate": 0.00029969595818412183,
"loss": 0.8091,
"step": 370
},
{
"epoch": 0.12,
"grad_norm": 0.055035530859369304,
"learning_rate": 0.0002996403002749686,
"loss": 0.8057,
"step": 375
},
{
"epoch": 0.1216,
"grad_norm": 0.05085006615854443,
"learning_rate": 0.0002995799730143689,
"loss": 0.8252,
"step": 380
},
{
"epoch": 0.1232,
"grad_norm": 0.05448546233902133,
"learning_rate": 0.0002995149782847646,
"loss": 0.7952,
"step": 385
},
{
"epoch": 0.1248,
"grad_norm": 0.05848704335233154,
"learning_rate": 0.0002994453181142407,
"loss": 0.8147,
"step": 390
},
{
"epoch": 0.1264,
"grad_norm": 0.05446939822521251,
"learning_rate": 0.00029937099467646167,
"loss": 0.8024,
"step": 395
},
{
"epoch": 0.128,
"grad_norm": 0.04770976632099585,
"learning_rate": 0.000299292010290604,
"loss": 0.8043,
"step": 400
},
{
"epoch": 0.1296,
"grad_norm": 0.05839860714064522,
"learning_rate": 0.0002992083674212835,
"loss": 0.8116,
"step": 405
},
{
"epoch": 0.1312,
"grad_norm": 0.049952793750330717,
"learning_rate": 0.0002991200686784785,
"loss": 0.7844,
"step": 410
},
{
"epoch": 0.1328,
"grad_norm": 0.07123910874954166,
"learning_rate": 0.0002990271168174487,
"loss": 0.8151,
"step": 415
},
{
"epoch": 0.1344,
"grad_norm": 0.05425468599745498,
"learning_rate": 0.00029892951473864865,
"loss": 0.8004,
"step": 420
},
{
"epoch": 0.136,
"grad_norm": 0.05714217590019245,
"learning_rate": 0.0002988272654876376,
"loss": 0.8067,
"step": 425
},
{
"epoch": 0.1376,
"grad_norm": 0.1253122019127607,
"learning_rate": 0.00029872037225498453,
"loss": 0.8135,
"step": 430
},
{
"epoch": 0.1392,
"grad_norm": 0.05522711987740126,
"learning_rate": 0.0002986088383761684,
"loss": 0.8113,
"step": 435
},
{
"epoch": 0.1408,
"grad_norm": 0.055953856909272716,
"learning_rate": 0.0002984926673314741,
"loss": 0.8005,
"step": 440
},
{
"epoch": 0.1424,
"grad_norm": 0.04955428496122915,
"learning_rate": 0.0002983718627458838,
"loss": 0.7894,
"step": 445
},
{
"epoch": 0.144,
"grad_norm": 0.047611164737695784,
"learning_rate": 0.0002982464283889642,
"loss": 0.7985,
"step": 450
},
{
"epoch": 0.1456,
"grad_norm": 0.04813634006358177,
"learning_rate": 0.0002981163681747483,
"loss": 0.7943,
"step": 455
},
{
"epoch": 0.1472,
"grad_norm": 0.04962556443270237,
"learning_rate": 0.0002979816861616138,
"loss": 0.8024,
"step": 460
},
{
"epoch": 0.1488,
"grad_norm": 0.05247098602006731,
"learning_rate": 0.00029784238655215626,
"loss": 0.8091,
"step": 465
},
{
"epoch": 0.1504,
"grad_norm": 0.052649745678383535,
"learning_rate": 0.0002976984736930578,
"loss": 0.8025,
"step": 470
},
{
"epoch": 0.152,
"grad_norm": 0.05330966657762999,
"learning_rate": 0.0002975499520749518,
"loss": 0.8,
"step": 475
},
{
"epoch": 0.1536,
"grad_norm": 0.04821750597458573,
"learning_rate": 0.00029739682633228245,
"loss": 0.8091,
"step": 480
},
{
"epoch": 0.1552,
"grad_norm": 0.04498787942068754,
"learning_rate": 0.0002972391012431605,
"loss": 0.8069,
"step": 485
},
{
"epoch": 0.1568,
"grad_norm": 0.04465216728155639,
"learning_rate": 0.00029707678172921366,
"loss": 0.7911,
"step": 490
},
{
"epoch": 0.1584,
"grad_norm": 0.05168404633277035,
"learning_rate": 0.0002969098728554336,
"loss": 0.7818,
"step": 495
},
{
"epoch": 0.16,
"grad_norm": 0.05321123357599694,
"learning_rate": 0.00029673837983001754,
"loss": 0.7887,
"step": 500
},
{
"epoch": 0.1616,
"grad_norm": 0.0553022357371466,
"learning_rate": 0.0002965623080042057,
"loss": 0.7798,
"step": 505
},
{
"epoch": 0.1632,
"grad_norm": 0.046594635452508006,
"learning_rate": 0.00029638166287211453,
"loss": 0.7795,
"step": 510
},
{
"epoch": 0.1648,
"grad_norm": 0.04619283115819418,
"learning_rate": 0.00029619645007056527,
"loss": 0.7824,
"step": 515
},
{
"epoch": 0.1664,
"grad_norm": 0.05935860026842771,
"learning_rate": 0.0002960066753789077,
"loss": 0.7938,
"step": 520
},
{
"epoch": 0.168,
"grad_norm": 0.04820460354727657,
"learning_rate": 0.00029581234471884045,
"loss": 0.7923,
"step": 525
},
{
"epoch": 0.1696,
"grad_norm": 0.04740546556763041,
"learning_rate": 0.00029561346415422554,
"loss": 0.773,
"step": 530
},
{
"epoch": 0.1712,
"grad_norm": 0.04749533059447172,
"learning_rate": 0.0002954100398908995,
"loss": 0.7968,
"step": 535
},
{
"epoch": 0.1728,
"grad_norm": 0.044893537989822555,
"learning_rate": 0.00029520207827647997,
"loss": 0.788,
"step": 540
},
{
"epoch": 0.1744,
"grad_norm": 0.056228050008934445,
"learning_rate": 0.0002949895858001669,
"loss": 0.7904,
"step": 545
},
{
"epoch": 0.176,
"grad_norm": 0.05863386920589539,
"learning_rate": 0.0002947725690925409,
"loss": 0.7978,
"step": 550
},
{
"epoch": 0.1776,
"grad_norm": 0.04997976221137418,
"learning_rate": 0.00029455103492535573,
"loss": 0.7929,
"step": 555
},
{
"epoch": 0.1792,
"grad_norm": 0.04840293365417223,
"learning_rate": 0.00029432499021132737,
"loss": 0.7907,
"step": 560
},
{
"epoch": 0.1808,
"grad_norm": 0.05601975290390555,
"learning_rate": 0.00029409444200391807,
"loss": 0.7859,
"step": 565
},
{
"epoch": 0.1824,
"grad_norm": 0.04753883644223113,
"learning_rate": 0.0002938593974971163,
"loss": 0.7908,
"step": 570
},
{
"epoch": 0.184,
"grad_norm": 0.047947841969773995,
"learning_rate": 0.00029361986402521255,
"loss": 0.7873,
"step": 575
},
{
"epoch": 0.1856,
"grad_norm": 0.054142174907416185,
"learning_rate": 0.00029337584906256994,
"loss": 0.7983,
"step": 580
},
{
"epoch": 0.1872,
"grad_norm": 0.054755120438311614,
"learning_rate": 0.0002931273602233915,
"loss": 0.7883,
"step": 585
},
{
"epoch": 0.1888,
"grad_norm": 0.04519348960924246,
"learning_rate": 0.0002928744052614823,
"loss": 0.7983,
"step": 590
},
{
"epoch": 0.1904,
"grad_norm": 0.044447268636241266,
"learning_rate": 0.00029261699207000757,
"loss": 0.7859,
"step": 595
},
{
"epoch": 0.192,
"grad_norm": 0.04364085507828995,
"learning_rate": 0.0002923551286812465,
"loss": 0.7556,
"step": 600
},
{
"epoch": 0.1936,
"grad_norm": 0.043353167391000154,
"learning_rate": 0.00029208882326634125,
"loss": 0.7692,
"step": 605
},
{
"epoch": 0.1952,
"grad_norm": 0.04633749372582479,
"learning_rate": 0.0002918180841350427,
"loss": 0.7679,
"step": 610
},
{
"epoch": 0.1968,
"grad_norm": 0.04839266146845783,
"learning_rate": 0.00029154291973545007,
"loss": 0.7686,
"step": 615
},
{
"epoch": 0.1984,
"grad_norm": 0.04676710759702116,
"learning_rate": 0.0002912633386537485,
"loss": 0.7909,
"step": 620
},
{
"epoch": 0.2,
"grad_norm": 0.051805169020436165,
"learning_rate": 0.00029097934961394025,
"loss": 0.7901,
"step": 625
},
{
"epoch": 0.2016,
"grad_norm": 0.04804749627680198,
"learning_rate": 0.0002906909614775729,
"loss": 0.7902,
"step": 630
},
{
"epoch": 0.2032,
"grad_norm": 0.05854711786733177,
"learning_rate": 0.00029039818324346254,
"loss": 0.795,
"step": 635
},
{
"epoch": 0.2048,
"grad_norm": 0.05333084863219475,
"learning_rate": 0.00029010102404741344,
"loss": 0.7802,
"step": 640
},
{
"epoch": 0.2064,
"grad_norm": 0.049081747767870276,
"learning_rate": 0.00028979949316193244,
"loss": 0.7794,
"step": 645
},
{
"epoch": 0.208,
"grad_norm": 0.05322940137804775,
"learning_rate": 0.00028949359999593996,
"loss": 0.7857,
"step": 650
},
{
"epoch": 0.2096,
"grad_norm": 0.04818612569385237,
"learning_rate": 0.0002891833540944764,
"loss": 0.8009,
"step": 655
},
{
"epoch": 0.2112,
"grad_norm": 0.04737304027832465,
"learning_rate": 0.000288868765138404,
"loss": 0.7934,
"step": 660
},
{
"epoch": 0.2128,
"grad_norm": 0.05584356580287081,
"learning_rate": 0.00028854984294410507,
"loss": 0.7781,
"step": 665
},
{
"epoch": 0.2144,
"grad_norm": 0.04283263855888774,
"learning_rate": 0.00028822659746317566,
"loss": 0.7672,
"step": 670
},
{
"epoch": 0.216,
"grad_norm": 0.043050424067810095,
"learning_rate": 0.00028789903878211477,
"loss": 0.8039,
"step": 675
},
{
"epoch": 0.2176,
"grad_norm": 0.04625087744694306,
"learning_rate": 0.00028756717712201,
"loss": 0.779,
"step": 680
},
{
"epoch": 0.2192,
"grad_norm": 0.05167588972499982,
"learning_rate": 0.00028723102283821823,
"loss": 0.7771,
"step": 685
},
{
"epoch": 0.2208,
"grad_norm": 0.04996238700800662,
"learning_rate": 0.0002868905864200428,
"loss": 0.7668,
"step": 690
},
{
"epoch": 0.2224,
"grad_norm": 0.049907457774237184,
"learning_rate": 0.0002865458784904059,
"loss": 0.7973,
"step": 695
},
{
"epoch": 0.224,
"grad_norm": 0.046459584215081846,
"learning_rate": 0.0002861969098055174,
"loss": 0.7598,
"step": 700
},
{
"epoch": 0.2256,
"grad_norm": 0.043878162763446554,
"learning_rate": 0.0002858436912545391,
"loss": 0.7684,
"step": 705
},
{
"epoch": 0.2272,
"grad_norm": 0.04684577416688988,
"learning_rate": 0.0002854862338592448,
"loss": 0.8022,
"step": 710
},
{
"epoch": 0.2288,
"grad_norm": 0.048864969061598686,
"learning_rate": 0.0002851245487736766,
"loss": 0.7919,
"step": 715
},
{
"epoch": 0.2304,
"grad_norm": 0.04633290517789117,
"learning_rate": 0.0002847586472837968,
"loss": 0.7844,
"step": 720
},
{
"epoch": 0.232,
"grad_norm": 0.049443592292022394,
"learning_rate": 0.00028438854080713557,
"loss": 0.79,
"step": 725
},
{
"epoch": 0.2336,
"grad_norm": 0.04491274060576765,
"learning_rate": 0.0002840142408924348,
"loss": 0.7863,
"step": 730
},
{
"epoch": 0.2352,
"grad_norm": 0.04360119247087781,
"learning_rate": 0.00028363575921928793,
"loss": 0.7983,
"step": 735
},
{
"epoch": 0.2368,
"grad_norm": 0.04411227163396601,
"learning_rate": 0.000283253107597775,
"loss": 0.7951,
"step": 740
},
{
"epoch": 0.2384,
"grad_norm": 0.04483270657736783,
"learning_rate": 0.0002828662979680947,
"loss": 0.7645,
"step": 745
},
{
"epoch": 0.24,
"grad_norm": 0.04464727990792154,
"learning_rate": 0.0002824753424001914,
"loss": 0.7952,
"step": 750
},
{
"epoch": 0.2416,
"grad_norm": 0.04228910734476722,
"learning_rate": 0.00028208025309337865,
"loss": 0.7809,
"step": 755
},
{
"epoch": 0.2432,
"grad_norm": 0.04660577036489219,
"learning_rate": 0.00028168104237595863,
"loss": 0.7571,
"step": 760
},
{
"epoch": 0.2448,
"grad_norm": 0.040295740530661546,
"learning_rate": 0.0002812777227048371,
"loss": 0.7574,
"step": 765
},
{
"epoch": 0.2464,
"grad_norm": 0.04609531144523526,
"learning_rate": 0.00028087030666513525,
"loss": 0.7884,
"step": 770
},
{
"epoch": 0.248,
"grad_norm": 0.0451935469029109,
"learning_rate": 0.0002804588069697964,
"loss": 0.7566,
"step": 775
},
{
"epoch": 0.2496,
"grad_norm": 0.04420421636010786,
"learning_rate": 0.00028004323645918974,
"loss": 0.7744,
"step": 780
},
{
"epoch": 0.2512,
"grad_norm": 0.04241174672641038,
"learning_rate": 0.00027962360810070955,
"loss": 0.7749,
"step": 785
},
{
"epoch": 0.2528,
"grad_norm": 0.0501739530739705,
"learning_rate": 0.00027919993498837047,
"loss": 0.7803,
"step": 790
},
{
"epoch": 0.2544,
"grad_norm": 0.0396318248988297,
"learning_rate": 0.00027877223034239896,
"loss": 0.771,
"step": 795
},
{
"epoch": 0.256,
"grad_norm": 0.04376685578343337,
"learning_rate": 0.00027834050750882074,
"loss": 0.7773,
"step": 800
},
{
"epoch": 0.2576,
"grad_norm": 0.04488045699495738,
"learning_rate": 0.0002779047799590445,
"loss": 0.7617,
"step": 805
},
{
"epoch": 0.2592,
"grad_norm": 0.04268108741438631,
"learning_rate": 0.00027746506128944147,
"loss": 0.7676,
"step": 810
},
{
"epoch": 0.2608,
"grad_norm": 0.044050818383708,
"learning_rate": 0.00027702136522092107,
"loss": 0.7742,
"step": 815
},
{
"epoch": 0.2624,
"grad_norm": 0.038564649454174094,
"learning_rate": 0.0002765737055985028,
"loss": 0.7969,
"step": 820
},
{
"epoch": 0.264,
"grad_norm": 0.04830433217959732,
"learning_rate": 0.00027612209639088427,
"loss": 0.751,
"step": 825
},
{
"epoch": 0.2656,
"grad_norm": 0.04519542288097644,
"learning_rate": 0.0002756665516900053,
"loss": 0.7621,
"step": 830
},
{
"epoch": 0.2672,
"grad_norm": 0.04064036493734512,
"learning_rate": 0.00027520708571060823,
"loss": 0.749,
"step": 835
},
{
"epoch": 0.2688,
"grad_norm": 0.04264802524163404,
"learning_rate": 0.0002747437127897943,
"loss": 0.7451,
"step": 840
},
{
"epoch": 0.2704,
"grad_norm": 0.046380711961248426,
"learning_rate": 0.0002742764473865763,
"loss": 0.7891,
"step": 845
},
{
"epoch": 0.272,
"grad_norm": 0.0411595744721063,
"learning_rate": 0.0002738053040814274,
"loss": 0.7551,
"step": 850
},
{
"epoch": 0.2736,
"grad_norm": 0.04174213578074948,
"learning_rate": 0.00027333029757582624,
"loss": 0.765,
"step": 855
},
{
"epoch": 0.2752,
"grad_norm": 0.041690426051977744,
"learning_rate": 0.00027285144269179816,
"loss": 0.7644,
"step": 860
},
{
"epoch": 0.2768,
"grad_norm": 0.042446196407229984,
"learning_rate": 0.0002723687543714525,
"loss": 0.784,
"step": 865
},
{
"epoch": 0.2784,
"grad_norm": 0.04106468767139888,
"learning_rate": 0.0002718822476765167,
"loss": 0.7565,
"step": 870
},
{
"epoch": 0.28,
"grad_norm": 0.04194569704278691,
"learning_rate": 0.00027139193778786603,
"loss": 0.7784,
"step": 875
},
{
"epoch": 0.2816,
"grad_norm": 0.04357814776830591,
"learning_rate": 0.0002708978400050501,
"loss": 0.7656,
"step": 880
},
{
"epoch": 0.2832,
"grad_norm": 0.05711679659881714,
"learning_rate": 0.0002703999697458152,
"loss": 0.7825,
"step": 885
},
{
"epoch": 0.2848,
"grad_norm": 0.0415912954827306,
"learning_rate": 0.0002698983425456235,
"loss": 0.7508,
"step": 890
},
{
"epoch": 0.2864,
"grad_norm": 0.04242258243133465,
"learning_rate": 0.00026939297405716807,
"loss": 0.7695,
"step": 895
},
{
"epoch": 0.288,
"grad_norm": 0.04501147994762728,
"learning_rate": 0.00026888388004988456,
"loss": 0.7419,
"step": 900
},
{
"epoch": 0.2896,
"grad_norm": 0.04743269896364638,
"learning_rate": 0.00026837107640945905,
"loss": 0.7658,
"step": 905
},
{
"epoch": 0.2912,
"grad_norm": 0.04452453652650223,
"learning_rate": 0.00026785457913733237,
"loss": 0.7615,
"step": 910
},
{
"epoch": 0.2928,
"grad_norm": 0.048856154299445204,
"learning_rate": 0.000267334404350201,
"loss": 0.7677,
"step": 915
},
{
"epoch": 0.2944,
"grad_norm": 0.03748901610551578,
"learning_rate": 0.00026681056827951386,
"loss": 0.7351,
"step": 920
},
{
"epoch": 0.296,
"grad_norm": 0.037646901686802126,
"learning_rate": 0.000266283087270966,
"loss": 0.7699,
"step": 925
},
{
"epoch": 0.2976,
"grad_norm": 0.04600811299738042,
"learning_rate": 0.00026575197778398843,
"loss": 0.7676,
"step": 930
},
{
"epoch": 0.2992,
"grad_norm": 0.05059509042956706,
"learning_rate": 0.0002652172563912348,
"loss": 0.7703,
"step": 935
},
{
"epoch": 0.3008,
"grad_norm": 0.04721913512257634,
"learning_rate": 0.00026467893977806387,
"loss": 0.7554,
"step": 940
},
{
"epoch": 0.3024,
"grad_norm": 0.04053439407377295,
"learning_rate": 0.0002641370447420192,
"loss": 0.7625,
"step": 945
},
{
"epoch": 0.304,
"grad_norm": 0.04465667290108694,
"learning_rate": 0.00026359158819230484,
"loss": 0.7665,
"step": 950
},
{
"epoch": 0.3056,
"grad_norm": 0.04103396429204654,
"learning_rate": 0.00026304258714925766,
"loss": 0.7755,
"step": 955
},
{
"epoch": 0.3072,
"grad_norm": 0.04085184907696068,
"learning_rate": 0.00026249005874381655,
"loss": 0.7613,
"step": 960
},
{
"epoch": 0.3088,
"grad_norm": 0.04436558524954724,
"learning_rate": 0.0002619340202169873,
"loss": 0.7611,
"step": 965
},
{
"epoch": 0.3104,
"grad_norm": 0.038850261454907435,
"learning_rate": 0.00026137448891930514,
"loss": 0.753,
"step": 970
},
{
"epoch": 0.312,
"grad_norm": 0.04437039419534914,
"learning_rate": 0.0002608114823102932,
"loss": 0.771,
"step": 975
},
{
"epoch": 0.3136,
"grad_norm": 0.04211467359387969,
"learning_rate": 0.0002602450179579176,
"loss": 0.7354,
"step": 980
},
{
"epoch": 0.3152,
"grad_norm": 0.04353742644114259,
"learning_rate": 0.0002596751135380392,
"loss": 0.7556,
"step": 985
},
{
"epoch": 0.3168,
"grad_norm": 0.040596688235953624,
"learning_rate": 0.00025910178683386247,
"loss": 0.7631,
"step": 990
},
{
"epoch": 0.3184,
"grad_norm": 0.03710704491539419,
"learning_rate": 0.0002585250557353799,
"loss": 0.7341,
"step": 995
},
{
"epoch": 0.32,
"grad_norm": 0.03852594895485715,
"learning_rate": 0.0002579449382388144,
"loss": 0.7456,
"step": 1000
},
{
"epoch": 0.3216,
"grad_norm": 0.03785627709199648,
"learning_rate": 0.00025736145244605746,
"loss": 0.7675,
"step": 1005
},
{
"epoch": 0.3232,
"grad_norm": 0.04162677723935767,
"learning_rate": 0.00025677461656410417,
"loss": 0.7516,
"step": 1010
},
{
"epoch": 0.3248,
"grad_norm": 0.04771538458909519,
"learning_rate": 0.00025618444890448543,
"loss": 0.7715,
"step": 1015
},
{
"epoch": 0.3264,
"grad_norm": 0.04299346898090266,
"learning_rate": 0.0002555909678826964,
"loss": 0.7762,
"step": 1020
},
{
"epoch": 0.328,
"grad_norm": 0.03864310543566412,
"learning_rate": 0.0002549941920176217,
"loss": 0.7355,
"step": 1025
},
{
"epoch": 0.3296,
"grad_norm": 0.039238228908675614,
"learning_rate": 0.00025439413993095784,
"loss": 0.7375,
"step": 1030
},
{
"epoch": 0.3312,
"grad_norm": 0.03947808720119218,
"learning_rate": 0.00025379083034663194,
"loss": 0.7469,
"step": 1035
},
{
"epoch": 0.3328,
"grad_norm": 0.04437116579538836,
"learning_rate": 0.0002531842820902175,
"loss": 0.7407,
"step": 1040
},
{
"epoch": 0.3344,
"grad_norm": 0.04156624023402751,
"learning_rate": 0.0002525745140883471,
"loss": 0.7636,
"step": 1045
},
{
"epoch": 0.336,
"grad_norm": 0.040237270074472614,
"learning_rate": 0.0002519615453681218,
"loss": 0.7556,
"step": 1050
},
{
"epoch": 0.3376,
"grad_norm": 0.04242608855098961,
"learning_rate": 0.00025134539505651716,
"loss": 0.7568,
"step": 1055
},
{
"epoch": 0.3392,
"grad_norm": 0.042429052835812414,
"learning_rate": 0.00025072608237978664,
"loss": 0.7487,
"step": 1060
},
{
"epoch": 0.3408,
"grad_norm": 0.04033545351922627,
"learning_rate": 0.00025010362666286185,
"loss": 0.7648,
"step": 1065
},
{
"epoch": 0.3424,
"grad_norm": 0.04303405340953674,
"learning_rate": 0.00024947804732874896,
"loss": 0.7593,
"step": 1070
},
{
"epoch": 0.344,
"grad_norm": 0.04731076259564207,
"learning_rate": 0.0002488493638979233,
"loss": 0.7486,
"step": 1075
},
{
"epoch": 0.3456,
"grad_norm": 0.04140947659846873,
"learning_rate": 0.00024821759598771964,
"loss": 0.7501,
"step": 1080
},
{
"epoch": 0.3472,
"grad_norm": 0.04687571061010034,
"learning_rate": 0.0002475827633117206,
"loss": 0.7568,
"step": 1085
},
{
"epoch": 0.3488,
"grad_norm": 0.042820580386593164,
"learning_rate": 0.00024694488567914106,
"loss": 0.7429,
"step": 1090
},
{
"epoch": 0.3504,
"grad_norm": 0.04032466634205255,
"learning_rate": 0.0002463039829942104,
"loss": 0.7535,
"step": 1095
},
{
"epoch": 0.352,
"grad_norm": 0.041362809714592476,
"learning_rate": 0.0002456600752555511,
"loss": 0.7583,
"step": 1100
},
{
"epoch": 0.3536,
"grad_norm": 0.04182236949981564,
"learning_rate": 0.0002450131825555548,
"loss": 0.7471,
"step": 1105
},
{
"epoch": 0.3552,
"grad_norm": 0.04849110153177951,
"learning_rate": 0.00024436332507975553,
"loss": 0.7408,
"step": 1110
},
{
"epoch": 0.3568,
"grad_norm": 0.04658865068956746,
"learning_rate": 0.00024371052310619967,
"loss": 0.7261,
"step": 1115
},
{
"epoch": 0.3584,
"grad_norm": 0.05051533767940744,
"learning_rate": 0.00024305479700481317,
"loss": 0.7364,
"step": 1120
},
{
"epoch": 0.36,
"grad_norm": 0.04276350726652445,
"learning_rate": 0.000242396167236766,
"loss": 0.7563,
"step": 1125
},
{
"epoch": 0.3616,
"grad_norm": 0.041420301812633155,
"learning_rate": 0.0002417346543538337,
"loss": 0.746,
"step": 1130
},
{
"epoch": 0.3632,
"grad_norm": 0.03822477826395941,
"learning_rate": 0.00024107027899775596,
"loss": 0.7601,
"step": 1135
},
{
"epoch": 0.3648,
"grad_norm": 0.03620070650500286,
"learning_rate": 0.00024040306189959283,
"loss": 0.7495,
"step": 1140
},
{
"epoch": 0.3664,
"grad_norm": 0.0427410166828309,
"learning_rate": 0.0002397330238790774,
"loss": 0.7431,
"step": 1145
},
{
"epoch": 0.368,
"grad_norm": 0.041891699350651304,
"learning_rate": 0.00023906018584396641,
"loss": 0.7589,
"step": 1150
},
{
"epoch": 0.3696,
"grad_norm": 0.04122486139728472,
"learning_rate": 0.00023838456878938786,
"loss": 0.7583,
"step": 1155
},
{
"epoch": 0.3712,
"grad_norm": 0.048970925700072804,
"learning_rate": 0.00023770619379718573,
"loss": 0.7626,
"step": 1160
},
{
"epoch": 0.3728,
"grad_norm": 0.04399464813611991,
"learning_rate": 0.00023702508203526235,
"loss": 0.7496,
"step": 1165
},
{
"epoch": 0.3744,
"grad_norm": 0.03871505770708521,
"learning_rate": 0.0002363412547569177,
"loss": 0.7424,
"step": 1170
},
{
"epoch": 0.376,
"grad_norm": 0.040796136255471696,
"learning_rate": 0.00023565473330018622,
"loss": 0.7359,
"step": 1175
},
{
"epoch": 0.3776,
"grad_norm": 0.04504615230274341,
"learning_rate": 0.00023496553908717115,
"loss": 0.7595,
"step": 1180
},
{
"epoch": 0.3792,
"grad_norm": 0.0409999978371454,
"learning_rate": 0.00023427369362337598,
"loss": 0.7466,
"step": 1185
},
{
"epoch": 0.3808,
"grad_norm": 0.03896861451931133,
"learning_rate": 0.00023357921849703335,
"loss": 0.7379,
"step": 1190
},
{
"epoch": 0.3824,
"grad_norm": 0.03805021975699206,
"learning_rate": 0.0002328821353784315,
"loss": 0.729,
"step": 1195
},
{
"epoch": 0.384,
"grad_norm": 0.03488320531224215,
"learning_rate": 0.000232182466019238,
"loss": 0.7186,
"step": 1200
},
{
"epoch": 0.3856,
"grad_norm": 0.040000324888927304,
"learning_rate": 0.00023148023225182102,
"loss": 0.7496,
"step": 1205
},
{
"epoch": 0.3872,
"grad_norm": 0.043739059629126366,
"learning_rate": 0.00023077545598856815,
"loss": 0.7415,
"step": 1210
},
{
"epoch": 0.3888,
"grad_norm": 0.0410141998156519,
"learning_rate": 0.00023006815922120262,
"loss": 0.7159,
"step": 1215
},
{
"epoch": 0.3904,
"grad_norm": 0.038108075126545675,
"learning_rate": 0.00022935836402009706,
"loss": 0.7684,
"step": 1220
},
{
"epoch": 0.392,
"grad_norm": 0.038848513464745,
"learning_rate": 0.00022864609253358474,
"loss": 0.7387,
"step": 1225
},
{
"epoch": 0.3936,
"grad_norm": 0.041501009662772,
"learning_rate": 0.00022793136698726863,
"loss": 0.7398,
"step": 1230
},
{
"epoch": 0.3952,
"grad_norm": 0.038213284503260184,
"learning_rate": 0.00022721420968332766,
"loss": 0.7472,
"step": 1235
},
{
"epoch": 0.3968,
"grad_norm": 0.04311819403546097,
"learning_rate": 0.00022649464299982116,
"loss": 0.7489,
"step": 1240
},
{
"epoch": 0.3984,
"grad_norm": 0.03770617577871306,
"learning_rate": 0.00022577268938999006,
"loss": 0.719,
"step": 1245
},
{
"epoch": 0.4,
"grad_norm": 0.03813545763715225,
"learning_rate": 0.00022504837138155666,
"loss": 0.7306,
"step": 1250
},
{
"epoch": 0.4016,
"grad_norm": 0.038457549045430986,
"learning_rate": 0.0002243217115760217,
"loss": 0.7457,
"step": 1255
},
{
"epoch": 0.4032,
"grad_norm": 0.04001321536490191,
"learning_rate": 0.0002235927326479588,
"loss": 0.7426,
"step": 1260
},
{
"epoch": 0.4048,
"grad_norm": 0.041024710335847044,
"learning_rate": 0.00022286145734430713,
"loss": 0.7197,
"step": 1265
},
{
"epoch": 0.4064,
"grad_norm": 0.03918303755025132,
"learning_rate": 0.00022212790848366164,
"loss": 0.7254,
"step": 1270
},
{
"epoch": 0.408,
"grad_norm": 0.036956610920742726,
"learning_rate": 0.00022139210895556104,
"loss": 0.7528,
"step": 1275
},
{
"epoch": 0.4096,
"grad_norm": 0.039136082628621616,
"learning_rate": 0.00022065408171977326,
"loss": 0.7495,
"step": 1280
},
{
"epoch": 0.4112,
"grad_norm": 0.03966283782035079,
"learning_rate": 0.00021991384980557958,
"loss": 0.7439,
"step": 1285
},
{
"epoch": 0.4128,
"grad_norm": 0.03978160138541518,
"learning_rate": 0.0002191714363110554,
"loss": 0.756,
"step": 1290
},
{
"epoch": 0.4144,
"grad_norm": 0.039080256705117104,
"learning_rate": 0.00021842686440235002,
"loss": 0.7509,
"step": 1295
},
{
"epoch": 0.416,
"grad_norm": 0.04155956423561961,
"learning_rate": 0.00021768015731296345,
"loss": 0.7452,
"step": 1300
},
{
"epoch": 0.4176,
"grad_norm": 0.03760072649313962,
"learning_rate": 0.00021693133834302145,
"loss": 0.7289,
"step": 1305
},
{
"epoch": 0.4192,
"grad_norm": 0.04656999356384314,
"learning_rate": 0.00021618043085854872,
"loss": 0.7374,
"step": 1310
},
{
"epoch": 0.4208,
"grad_norm": 0.04133245269084771,
"learning_rate": 0.00021542745829073958,
"loss": 0.7361,
"step": 1315
},
{
"epoch": 0.4224,
"grad_norm": 0.037588854626515254,
"learning_rate": 0.00021467244413522673,
"loss": 0.7391,
"step": 1320
},
{
"epoch": 0.424,
"grad_norm": 0.03805822254306633,
"learning_rate": 0.00021391541195134843,
"loss": 0.7431,
"step": 1325
},
{
"epoch": 0.4256,
"grad_norm": 0.040957247951686995,
"learning_rate": 0.0002131563853614131,
"loss": 0.7424,
"step": 1330
},
{
"epoch": 0.4272,
"grad_norm": 0.04027062941812723,
"learning_rate": 0.00021239538804996216,
"loss": 0.7513,
"step": 1335
},
{
"epoch": 0.4288,
"grad_norm": 0.038733573038441466,
"learning_rate": 0.00021163244376303132,
"loss": 0.7602,
"step": 1340
},
{
"epoch": 0.4304,
"grad_norm": 0.039921899700948164,
"learning_rate": 0.0002108675763074092,
"loss": 0.7436,
"step": 1345
},
{
"epoch": 0.432,
"grad_norm": 0.036728064643141406,
"learning_rate": 0.00021010080954989482,
"loss": 0.7411,
"step": 1350
},
{
"epoch": 0.4336,
"grad_norm": 0.035499942850762016,
"learning_rate": 0.00020933216741655258,
"loss": 0.7348,
"step": 1355
},
{
"epoch": 0.4352,
"grad_norm": 0.03812443979332148,
"learning_rate": 0.00020856167389196584,
"loss": 0.7266,
"step": 1360
},
{
"epoch": 0.4368,
"grad_norm": 0.03898025685932717,
"learning_rate": 0.0002077893530184885,
"loss": 0.7271,
"step": 1365
},
{
"epoch": 0.4384,
"grad_norm": 0.04031802844377227,
"learning_rate": 0.0002070152288954947,
"loss": 0.7376,
"step": 1370
},
{
"epoch": 0.44,
"grad_norm": 0.04429882701998814,
"learning_rate": 0.00020623932567862693,
"loss": 0.7254,
"step": 1375
},
{
"epoch": 0.4416,
"grad_norm": 0.04317464133226769,
"learning_rate": 0.0002054616675790423,
"loss": 0.7509,
"step": 1380
},
{
"epoch": 0.4432,
"grad_norm": 0.037580872841089984,
"learning_rate": 0.0002046822788626568,
"loss": 0.7371,
"step": 1385
},
{
"epoch": 0.4448,
"grad_norm": 0.0376095279868885,
"learning_rate": 0.00020390118384938842,
"loss": 0.7335,
"step": 1390
},
{
"epoch": 0.4464,
"grad_norm": 0.04061633520605329,
"learning_rate": 0.00020311840691239822,
"loss": 0.7392,
"step": 1395
},
{
"epoch": 0.448,
"grad_norm": 0.041717279079417066,
"learning_rate": 0.0002023339724773297,
"loss": 0.7349,
"step": 1400
},
{
"epoch": 0.4496,
"grad_norm": 0.039293898353158246,
"learning_rate": 0.00020154790502154653,
"loss": 0.7279,
"step": 1405
},
{
"epoch": 0.4512,
"grad_norm": 0.03992264637190769,
"learning_rate": 0.00020076022907336902,
"loss": 0.7443,
"step": 1410
},
{
"epoch": 0.4528,
"grad_norm": 0.04383939318985396,
"learning_rate": 0.00019997096921130862,
"loss": 0.7203,
"step": 1415
},
{
"epoch": 0.4544,
"grad_norm": 0.043820833475301676,
"learning_rate": 0.00019918015006330087,
"loss": 0.7409,
"step": 1420
},
{
"epoch": 0.456,
"grad_norm": 0.040400610701221557,
"learning_rate": 0.0001983877963059372,
"loss": 0.737,
"step": 1425
},
{
"epoch": 0.4576,
"grad_norm": 0.040567222551813824,
"learning_rate": 0.00019759393266369443,
"loss": 0.7429,
"step": 1430
},
{
"epoch": 0.4592,
"grad_norm": 0.03972604845474063,
"learning_rate": 0.00019679858390816385,
"loss": 0.7348,
"step": 1435
},
{
"epoch": 0.4608,
"grad_norm": 0.043501993298391314,
"learning_rate": 0.000196001774857278,
"loss": 0.7257,
"step": 1440
},
{
"epoch": 0.4624,
"grad_norm": 0.040776034820328655,
"learning_rate": 0.00019520353037453598,
"loss": 0.725,
"step": 1445
},
{
"epoch": 0.464,
"grad_norm": 0.041230798349950214,
"learning_rate": 0.00019440387536822807,
"loss": 0.7545,
"step": 1450
},
{
"epoch": 0.4656,
"grad_norm": 0.039961315015310875,
"learning_rate": 0.00019360283479065833,
"loss": 0.7319,
"step": 1455
},
{
"epoch": 0.4672,
"grad_norm": 0.04254792018485242,
"learning_rate": 0.00019280043363736579,
"loss": 0.7309,
"step": 1460
},
{
"epoch": 0.4688,
"grad_norm": 0.04048553943500062,
"learning_rate": 0.0001919966969463448,
"loss": 0.727,
"step": 1465
},
{
"epoch": 0.4704,
"grad_norm": 0.04316845941777185,
"learning_rate": 0.0001911916497972635,
"loss": 0.7175,
"step": 1470
},
{
"epoch": 0.472,
"grad_norm": 0.03783372136381564,
"learning_rate": 0.0001903853173106815,
"loss": 0.7289,
"step": 1475
},
{
"epoch": 0.4736,
"grad_norm": 0.0389157416747743,
"learning_rate": 0.00018957772464726574,
"loss": 0.7295,
"step": 1480
},
{
"epoch": 0.4752,
"grad_norm": 0.04219066065752722,
"learning_rate": 0.00018876889700700556,
"loss": 0.7184,
"step": 1485
},
{
"epoch": 0.4768,
"grad_norm": 0.04161411422302911,
"learning_rate": 0.00018795885962842628,
"loss": 0.737,
"step": 1490
},
{
"epoch": 0.4784,
"grad_norm": 0.04035309724428451,
"learning_rate": 0.0001871476377878018,
"loss": 0.7349,
"step": 1495
},
{
"epoch": 0.48,
"grad_norm": 0.04184638257355627,
"learning_rate": 0.00018633525679836568,
"loss": 0.7487,
"step": 1500
},
{
"epoch": 0.4816,
"grad_norm": 0.04004974821302725,
"learning_rate": 0.00018552174200952133,
"loss": 0.7285,
"step": 1505
},
{
"epoch": 0.4832,
"grad_norm": 0.04171223830308486,
"learning_rate": 0.00018470711880605122,
"loss": 0.7255,
"step": 1510
},
{
"epoch": 0.4848,
"grad_norm": 0.04250897175819996,
"learning_rate": 0.00018389141260732444,
"loss": 0.7271,
"step": 1515
},
{
"epoch": 0.4864,
"grad_norm": 0.04076185254200827,
"learning_rate": 0.00018307464886650377,
"loss": 0.7165,
"step": 1520
},
{
"epoch": 0.488,
"grad_norm": 0.039216538966245114,
"learning_rate": 0.00018225685306975134,
"loss": 0.7245,
"step": 1525
},
{
"epoch": 0.4896,
"grad_norm": 0.040994427440341694,
"learning_rate": 0.00018143805073543343,
"loss": 0.7223,
"step": 1530
},
{
"epoch": 0.4912,
"grad_norm": 0.0394145288759038,
"learning_rate": 0.00018061826741332405,
"loss": 0.7179,
"step": 1535
},
{
"epoch": 0.4928,
"grad_norm": 0.040009604176984924,
"learning_rate": 0.0001797975286838079,
"loss": 0.7191,
"step": 1540
},
{
"epoch": 0.4944,
"grad_norm": 0.03726486414212371,
"learning_rate": 0.00017897586015708198,
"loss": 0.7272,
"step": 1545
},
{
"epoch": 0.496,
"grad_norm": 0.03976279936238677,
"learning_rate": 0.0001781532874723566,
"loss": 0.7411,
"step": 1550
},
{
"epoch": 0.4976,
"grad_norm": 0.038599359368600886,
"learning_rate": 0.0001773298362970552,
"loss": 0.7204,
"step": 1555
},
{
"epoch": 0.4992,
"grad_norm": 0.0401664051064554,
"learning_rate": 0.00017650553232601354,
"loss": 0.7353,
"step": 1560
},
{
"epoch": 0.5008,
"grad_norm": 0.03738264613063052,
"learning_rate": 0.00017568040128067782,
"loss": 0.7448,
"step": 1565
},
{
"epoch": 0.5024,
"grad_norm": 0.03853467145191168,
"learning_rate": 0.00017485446890830225,
"loss": 0.723,
"step": 1570
},
{
"epoch": 0.504,
"grad_norm": 0.0375595639198087,
"learning_rate": 0.00017402776098114539,
"loss": 0.7154,
"step": 1575
},
{
"epoch": 0.5056,
"grad_norm": 0.03909937065737528,
"learning_rate": 0.00017320030329566615,
"loss": 0.7165,
"step": 1580
},
{
"epoch": 0.5072,
"grad_norm": 0.03652376753762799,
"learning_rate": 0.0001723721216717188,
"loss": 0.7204,
"step": 1585
},
{
"epoch": 0.5088,
"grad_norm": 0.03602599991489169,
"learning_rate": 0.00017154324195174718,
"loss": 0.7153,
"step": 1590
},
{
"epoch": 0.5104,
"grad_norm": 0.036925105258512454,
"learning_rate": 0.00017071368999997848,
"loss": 0.7155,
"step": 1595
},
{
"epoch": 0.512,
"grad_norm": 0.03908920466891324,
"learning_rate": 0.00016988349170161607,
"loss": 0.7238,
"step": 1600
},
{
"epoch": 0.5136,
"grad_norm": 0.03410674141609763,
"learning_rate": 0.0001690526729620318,
"loss": 0.7364,
"step": 1605
},
{
"epoch": 0.5152,
"grad_norm": 0.03669328777188668,
"learning_rate": 0.00016822125970595772,
"loss": 0.7263,
"step": 1610
},
{
"epoch": 0.5168,
"grad_norm": 0.03888960042867729,
"learning_rate": 0.00016738927787667706,
"loss": 0.732,
"step": 1615
},
{
"epoch": 0.5184,
"grad_norm": 0.037017055980232354,
"learning_rate": 0.00016655675343521463,
"loss": 0.7181,
"step": 1620
},
{
"epoch": 0.52,
"grad_norm": 0.038248175503987805,
"learning_rate": 0.00016572371235952697,
"loss": 0.7217,
"step": 1625
},
{
"epoch": 0.5216,
"grad_norm": 0.03575187831459462,
"learning_rate": 0.0001648901806436914,
"loss": 0.7281,
"step": 1630
},
{
"epoch": 0.5232,
"grad_norm": 0.03632394793973946,
"learning_rate": 0.00016405618429709531,
"loss": 0.706,
"step": 1635
},
{
"epoch": 0.5248,
"grad_norm": 0.03997997427640186,
"learning_rate": 0.00016322174934362422,
"loss": 0.7348,
"step": 1640
},
{
"epoch": 0.5264,
"grad_norm": 0.04021229815953381,
"learning_rate": 0.00016238690182084986,
"loss": 0.7015,
"step": 1645
},
{
"epoch": 0.528,
"grad_norm": 0.0366309746554456,
"learning_rate": 0.0001615516677792177,
"loss": 0.7255,
"step": 1650
},
{
"epoch": 0.5296,
"grad_norm": 0.037279314257418296,
"learning_rate": 0.00016071607328123425,
"loss": 0.7223,
"step": 1655
},
{
"epoch": 0.5312,
"grad_norm": 0.03759198074622222,
"learning_rate": 0.0001598801444006534,
"loss": 0.7387,
"step": 1660
},
{
"epoch": 0.5328,
"grad_norm": 0.03518096068292559,
"learning_rate": 0.0001590439072216632,
"loss": 0.7005,
"step": 1665
},
{
"epoch": 0.5344,
"grad_norm": 0.03896613837402476,
"learning_rate": 0.0001582073878380718,
"loss": 0.7151,
"step": 1670
},
{
"epoch": 0.536,
"grad_norm": 0.03907662349310434,
"learning_rate": 0.00015737061235249309,
"loss": 0.6996,
"step": 1675
},
{
"epoch": 0.5376,
"grad_norm": 0.03911161698811334,
"learning_rate": 0.0001565336068755325,
"loss": 0.7315,
"step": 1680
},
{
"epoch": 0.5392,
"grad_norm": 0.038117748491007554,
"learning_rate": 0.00015569639752497196,
"loss": 0.7069,
"step": 1685
},
{
"epoch": 0.5408,
"grad_norm": 0.03868499072903773,
"learning_rate": 0.00015485901042495504,
"loss": 0.7256,
"step": 1690
},
{
"epoch": 0.5424,
"grad_norm": 0.037768151116837435,
"learning_rate": 0.00015402147170517184,
"loss": 0.7129,
"step": 1695
},
{
"epoch": 0.544,
"grad_norm": 0.035146221464533924,
"learning_rate": 0.00015318380750004352,
"loss": 0.7236,
"step": 1700
},
{
"epoch": 0.5456,
"grad_norm": 0.03720287204643548,
"learning_rate": 0.00015234604394790687,
"loss": 0.7326,
"step": 1705
},
{
"epoch": 0.5472,
"grad_norm": 0.038686413117410404,
"learning_rate": 0.00015150820719019874,
"loss": 0.7214,
"step": 1710
},
{
"epoch": 0.5488,
"grad_norm": 0.03994154647602964,
"learning_rate": 0.00015067032337064018,
"loss": 0.7188,
"step": 1715
},
{
"epoch": 0.5504,
"grad_norm": 0.04447078918782078,
"learning_rate": 0.00014983241863442086,
"loss": 0.7316,
"step": 1720
},
{
"epoch": 0.552,
"grad_norm": 0.03756715443440511,
"learning_rate": 0.00014899451912738307,
"loss": 0.7087,
"step": 1725
},
{
"epoch": 0.5536,
"grad_norm": 0.035466288303780846,
"learning_rate": 0.00014815665099520588,
"loss": 0.6948,
"step": 1730
},
{
"epoch": 0.5552,
"grad_norm": 0.03901434209246453,
"learning_rate": 0.00014731884038258946,
"loss": 0.706,
"step": 1735
},
{
"epoch": 0.5568,
"grad_norm": 0.04049446829396525,
"learning_rate": 0.00014648111343243907,
"loss": 0.7326,
"step": 1740
},
{
"epoch": 0.5584,
"grad_norm": 0.038166521181569386,
"learning_rate": 0.00014564349628504937,
"loss": 0.7073,
"step": 1745
},
{
"epoch": 0.56,
"grad_norm": 0.03550094230425758,
"learning_rate": 0.00014480601507728885,
"loss": 0.6926,
"step": 1750
},
{
"epoch": 0.5616,
"grad_norm": 0.037454018664029386,
"learning_rate": 0.00014396869594178403,
"loss": 0.7098,
"step": 1755
},
{
"epoch": 0.5632,
"grad_norm": 0.038152055665675956,
"learning_rate": 0.0001431315650061042,
"loss": 0.7227,
"step": 1760
},
{
"epoch": 0.5648,
"grad_norm": 0.037743995463609896,
"learning_rate": 0.00014229464839194618,
"loss": 0.715,
"step": 1765
},
{
"epoch": 0.5664,
"grad_norm": 0.034892097661775624,
"learning_rate": 0.00014145797221431903,
"loss": 0.7198,
"step": 1770
},
{
"epoch": 0.568,
"grad_norm": 0.04264265788852482,
"learning_rate": 0.0001406215625807293,
"loss": 0.7128,
"step": 1775
},
{
"epoch": 0.5696,
"grad_norm": 0.038119174371547215,
"learning_rate": 0.00013978544559036638,
"loss": 0.71,
"step": 1780
},
{
"epoch": 0.5712,
"grad_norm": 0.03674005453515878,
"learning_rate": 0.00013894964733328797,
"loss": 0.7152,
"step": 1785
},
{
"epoch": 0.5728,
"grad_norm": 0.03538257887586784,
"learning_rate": 0.00013811419388960622,
"loss": 0.7237,
"step": 1790
},
{
"epoch": 0.5744,
"grad_norm": 0.03868691442507021,
"learning_rate": 0.00013727911132867365,
"loss": 0.7056,
"step": 1795
},
{
"epoch": 0.576,
"grad_norm": 0.038482050801233125,
"learning_rate": 0.00013644442570826982,
"loss": 0.7194,
"step": 1800
},
{
"epoch": 0.5776,
"grad_norm": 0.037890697003074354,
"learning_rate": 0.0001356101630737883,
"loss": 0.7089,
"step": 1805
},
{
"epoch": 0.5792,
"grad_norm": 0.04132795310535581,
"learning_rate": 0.0001347763494574239,
"loss": 0.7236,
"step": 1810
},
{
"epoch": 0.5808,
"grad_norm": 0.03630261250379611,
"learning_rate": 0.0001339430108773602,
"loss": 0.7152,
"step": 1815
},
{
"epoch": 0.5824,
"grad_norm": 0.03898346570156072,
"learning_rate": 0.00013311017333695796,
"loss": 0.7205,
"step": 1820
},
{
"epoch": 0.584,
"grad_norm": 0.03487145367369247,
"learning_rate": 0.00013227786282394349,
"loss": 0.7,
"step": 1825
},
{
"epoch": 0.5856,
"grad_norm": 0.038999878622698395,
"learning_rate": 0.00013144610530959784,
"loss": 0.7124,
"step": 1830
},
{
"epoch": 0.5872,
"grad_norm": 0.03835161731919042,
"learning_rate": 0.00013061492674794646,
"loss": 0.7147,
"step": 1835
},
{
"epoch": 0.5888,
"grad_norm": 0.03730043338102035,
"learning_rate": 0.00012978435307494916,
"loss": 0.7062,
"step": 1840
},
{
"epoch": 0.5904,
"grad_norm": 0.036977276929533315,
"learning_rate": 0.00012895441020769095,
"loss": 0.7038,
"step": 1845
},
{
"epoch": 0.592,
"grad_norm": 0.038939333496150315,
"learning_rate": 0.00012812512404357328,
"loss": 0.7256,
"step": 1850
},
{
"epoch": 0.5936,
"grad_norm": 0.037225774423761265,
"learning_rate": 0.0001272965204595059,
"loss": 0.7183,
"step": 1855
},
{
"epoch": 0.5952,
"grad_norm": 0.04081105556123695,
"learning_rate": 0.00012646862531109945,
"loss": 0.7094,
"step": 1860
},
{
"epoch": 0.5968,
"grad_norm": 0.04181029136047441,
"learning_rate": 0.0001256414644318588,
"loss": 0.7137,
"step": 1865
},
{
"epoch": 0.5984,
"grad_norm": 0.03588648909549693,
"learning_rate": 0.00012481506363237654,
"loss": 0.7027,
"step": 1870
},
{
"epoch": 0.6,
"grad_norm": 0.036808997828812244,
"learning_rate": 0.00012398944869952816,
"loss": 0.723,
"step": 1875
},
{
"epoch": 0.6016,
"grad_norm": 0.03522573429032402,
"learning_rate": 0.00012316464539566683,
"loss": 0.708,
"step": 1880
},
{
"epoch": 0.6032,
"grad_norm": 0.03454246392118955,
"learning_rate": 0.0001223406794578199,
"loss": 0.6919,
"step": 1885
},
{
"epoch": 0.6048,
"grad_norm": 0.03473658287408358,
"learning_rate": 0.00012151757659688571,
"loss": 0.7072,
"step": 1890
},
{
"epoch": 0.6064,
"grad_norm": 0.03431214376771132,
"learning_rate": 0.00012069536249683134,
"loss": 0.7089,
"step": 1895
},
{
"epoch": 0.608,
"grad_norm": 0.036810476493388276,
"learning_rate": 0.00011987406281389094,
"loss": 0.6957,
"step": 1900
},
{
"epoch": 0.6096,
"grad_norm": 0.039483568169593,
"learning_rate": 0.00011905370317576543,
"loss": 0.6988,
"step": 1905
},
{
"epoch": 0.6112,
"grad_norm": 0.033934505615374644,
"learning_rate": 0.00011823430918082282,
"loss": 0.7087,
"step": 1910
},
{
"epoch": 0.6128,
"grad_norm": 0.04018209525941744,
"learning_rate": 0.00011741590639729918,
"loss": 0.7092,
"step": 1915
},
{
"epoch": 0.6144,
"grad_norm": 0.040385686633241026,
"learning_rate": 0.00011659852036250113,
"loss": 0.7077,
"step": 1920
},
{
"epoch": 0.616,
"grad_norm": 0.039665974606226624,
"learning_rate": 0.00011578217658200874,
"loss": 0.7048,
"step": 1925
},
{
"epoch": 0.6176,
"grad_norm": 0.03354649474007928,
"learning_rate": 0.00011496690052887973,
"loss": 0.717,
"step": 1930
},
{
"epoch": 0.6192,
"grad_norm": 0.03701722021814875,
"learning_rate": 0.00011415271764285474,
"loss": 0.7077,
"step": 1935
},
{
"epoch": 0.6208,
"grad_norm": 0.0381990564552986,
"learning_rate": 0.00011333965332956332,
"loss": 0.7195,
"step": 1940
},
{
"epoch": 0.6224,
"grad_norm": 0.03570431209088912,
"learning_rate": 0.00011252773295973121,
"loss": 0.6873,
"step": 1945
},
{
"epoch": 0.624,
"grad_norm": 0.03944416803051141,
"learning_rate": 0.00011171698186838887,
"loss": 0.6891,
"step": 1950
},
{
"epoch": 0.6256,
"grad_norm": 0.04099611704369037,
"learning_rate": 0.00011090742535408063,
"loss": 0.7055,
"step": 1955
},
{
"epoch": 0.6272,
"grad_norm": 0.04133111071932114,
"learning_rate": 0.00011009908867807556,
"loss": 0.7048,
"step": 1960
},
{
"epoch": 0.6288,
"grad_norm": 0.03831577950670325,
"learning_rate": 0.00010929199706357905,
"loss": 0.6927,
"step": 1965
},
{
"epoch": 0.6304,
"grad_norm": 0.03551375777130708,
"learning_rate": 0.0001084861756949457,
"loss": 0.689,
"step": 1970
},
{
"epoch": 0.632,
"grad_norm": 0.03393922854857721,
"learning_rate": 0.00010768164971689374,
"loss": 0.6899,
"step": 1975
},
{
"epoch": 0.6336,
"grad_norm": 0.03750670368784519,
"learning_rate": 0.00010687844423372018,
"loss": 0.7013,
"step": 1980
},
{
"epoch": 0.6352,
"grad_norm": 0.03806186758532188,
"learning_rate": 0.00010607658430851744,
"loss": 0.7038,
"step": 1985
},
{
"epoch": 0.6368,
"grad_norm": 0.04134318018265958,
"learning_rate": 0.00010527609496239142,
"loss": 0.7049,
"step": 1990
},
{
"epoch": 0.6384,
"grad_norm": 0.03550182496466848,
"learning_rate": 0.00010447700117368078,
"loss": 0.7047,
"step": 1995
},
{
"epoch": 0.64,
"grad_norm": 0.037739871885332836,
"learning_rate": 0.00010367932787717727,
"loss": 0.7388,
"step": 2000
},
{
"epoch": 0.6416,
"grad_norm": 0.039812095343439186,
"learning_rate": 0.00010288309996334801,
"loss": 0.7086,
"step": 2005
},
{
"epoch": 0.6432,
"grad_norm": 0.03962726915267911,
"learning_rate": 0.00010208834227755847,
"loss": 0.6995,
"step": 2010
},
{
"epoch": 0.6448,
"grad_norm": 0.03703057191589663,
"learning_rate": 0.00010129507961929748,
"loss": 0.6996,
"step": 2015
},
{
"epoch": 0.6464,
"grad_norm": 0.03724460564299106,
"learning_rate": 0.00010050333674140326,
"loss": 0.7305,
"step": 2020
},
{
"epoch": 0.648,
"grad_norm": 0.039996173756348674,
"learning_rate": 9.971313834929099e-05,
"loss": 0.7082,
"step": 2025
},
{
"epoch": 0.6496,
"grad_norm": 0.036464486333798785,
"learning_rate": 9.892450910018203e-05,
"loss": 0.7043,
"step": 2030
},
{
"epoch": 0.6512,
"grad_norm": 0.03667379715726693,
"learning_rate": 9.813747360233443e-05,
"loss": 0.6782,
"step": 2035
},
{
"epoch": 0.6528,
"grad_norm": 0.03543386252761958,
"learning_rate": 9.73520564142751e-05,
"loss": 0.6894,
"step": 2040
},
{
"epoch": 0.6544,
"grad_norm": 0.03794120210301895,
"learning_rate": 9.656828204403352e-05,
"loss": 0.6988,
"step": 2045
},
{
"epoch": 0.656,
"grad_norm": 0.03829948035849497,
"learning_rate": 9.578617494837682e-05,
"loss": 0.7053,
"step": 2050
},
{
"epoch": 0.6576,
"grad_norm": 0.03832452621376072,
"learning_rate": 9.500575953204684e-05,
"loss": 0.7021,
"step": 2055
},
{
"epoch": 0.6592,
"grad_norm": 0.04287718638930358,
"learning_rate": 9.422706014699863e-05,
"loss": 0.7146,
"step": 2060
},
{
"epoch": 0.6608,
"grad_norm": 0.03894226172707719,
"learning_rate": 9.345010109164037e-05,
"loss": 0.6902,
"step": 2065
},
{
"epoch": 0.6624,
"grad_norm": 0.03938303692148145,
"learning_rate": 9.267490661007529e-05,
"loss": 0.694,
"step": 2070
},
{
"epoch": 0.664,
"grad_norm": 0.034936262311856585,
"learning_rate": 9.190150089134512e-05,
"loss": 0.6967,
"step": 2075
},
{
"epoch": 0.6656,
"grad_norm": 0.03616855319427236,
"learning_rate": 9.112990806867543e-05,
"loss": 0.703,
"step": 2080
},
{
"epoch": 0.6672,
"grad_norm": 0.03979096069450952,
"learning_rate": 9.036015221872238e-05,
"loss": 0.7299,
"step": 2085
},
{
"epoch": 0.6688,
"grad_norm": 0.039915619737043856,
"learning_rate": 8.959225736082158e-05,
"loss": 0.6726,
"step": 2090
},
{
"epoch": 0.6704,
"grad_norm": 0.03729325756021141,
"learning_rate": 8.882624745623856e-05,
"loss": 0.6973,
"step": 2095
},
{
"epoch": 0.672,
"grad_norm": 0.034404029229393965,
"learning_rate": 8.806214640742099e-05,
"loss": 0.7098,
"step": 2100
},
{
"epoch": 0.6736,
"grad_norm": 0.03650525795905925,
"learning_rate": 8.729997805725303e-05,
"loss": 0.6881,
"step": 2105
},
{
"epoch": 0.6752,
"grad_norm": 0.036096936020264495,
"learning_rate": 8.653976618831118e-05,
"loss": 0.6999,
"step": 2110
},
{
"epoch": 0.6768,
"grad_norm": 0.03778412130869877,
"learning_rate": 8.578153452212221e-05,
"loss": 0.711,
"step": 2115
},
{
"epoch": 0.6784,
"grad_norm": 0.04047592184151115,
"learning_rate": 8.502530671842288e-05,
"loss": 0.6997,
"step": 2120
},
{
"epoch": 0.68,
"grad_norm": 0.035782672640941406,
"learning_rate": 8.427110637442201e-05,
"loss": 0.7078,
"step": 2125
},
{
"epoch": 0.6816,
"grad_norm": 0.03576553726755442,
"learning_rate": 8.351895702406366e-05,
"loss": 0.7095,
"step": 2130
},
{
"epoch": 0.6832,
"grad_norm": 0.03413584162755607,
"learning_rate": 8.276888213729307e-05,
"loss": 0.6815,
"step": 2135
},
{
"epoch": 0.6848,
"grad_norm": 0.03566358486055788,
"learning_rate": 8.202090511932428e-05,
"loss": 0.7103,
"step": 2140
},
{
"epoch": 0.6864,
"grad_norm": 0.03456692741195639,
"learning_rate": 8.127504930990986e-05,
"loss": 0.7074,
"step": 2145
},
{
"epoch": 0.688,
"grad_norm": 0.03507058615307166,
"learning_rate": 8.053133798261233e-05,
"loss": 0.6985,
"step": 2150
},
{
"epoch": 0.6896,
"grad_norm": 0.035178970930575086,
"learning_rate": 7.978979434407842e-05,
"loss": 0.6745,
"step": 2155
},
{
"epoch": 0.6912,
"grad_norm": 0.032560705746558505,
"learning_rate": 7.905044153331441e-05,
"loss": 0.7063,
"step": 2160
},
{
"epoch": 0.6928,
"grad_norm": 0.0402934382923456,
"learning_rate": 7.831330262096444e-05,
"loss": 0.6964,
"step": 2165
},
{
"epoch": 0.6944,
"grad_norm": 0.03527387876715467,
"learning_rate": 7.757840060859053e-05,
"loss": 0.6861,
"step": 2170
},
{
"epoch": 0.696,
"grad_norm": 0.03758618705619355,
"learning_rate": 7.684575842795485e-05,
"loss": 0.7019,
"step": 2175
},
{
"epoch": 0.6976,
"grad_norm": 0.03399690957991901,
"learning_rate": 7.611539894030405e-05,
"loss": 0.6809,
"step": 2180
},
{
"epoch": 0.6992,
"grad_norm": 0.0350021795963823,
"learning_rate": 7.53873449356562e-05,
"loss": 0.6983,
"step": 2185
},
{
"epoch": 0.7008,
"grad_norm": 0.03679414632573209,
"learning_rate": 7.466161913208932e-05,
"loss": 0.7079,
"step": 2190
},
{
"epoch": 0.7024,
"grad_norm": 0.03983925157477276,
"learning_rate": 7.39382441750326e-05,
"loss": 0.6807,
"step": 2195
},
{
"epoch": 0.704,
"grad_norm": 0.035022468317805,
"learning_rate": 7.321724263655988e-05,
"loss": 0.688,
"step": 2200
},
{
"epoch": 0.7056,
"grad_norm": 0.03671129916435671,
"learning_rate": 7.249863701468514e-05,
"loss": 0.6955,
"step": 2205
},
{
"epoch": 0.7072,
"grad_norm": 0.03639449254399468,
"learning_rate": 7.178244973266072e-05,
"loss": 0.7208,
"step": 2210
},
{
"epoch": 0.7088,
"grad_norm": 0.03460495364590804,
"learning_rate": 7.106870313827734e-05,
"loss": 0.7035,
"step": 2215
},
{
"epoch": 0.7104,
"grad_norm": 0.03472980354804098,
"learning_rate": 7.035741950316698e-05,
"loss": 0.6829,
"step": 2220
},
{
"epoch": 0.712,
"grad_norm": 0.03510727461117341,
"learning_rate": 6.964862102210773e-05,
"loss": 0.6851,
"step": 2225
},
{
"epoch": 0.7136,
"grad_norm": 0.03546966441115835,
"learning_rate": 6.894232981233154e-05,
"loss": 0.702,
"step": 2230
},
{
"epoch": 0.7152,
"grad_norm": 0.03629653478705703,
"learning_rate": 6.823856791283366e-05,
"loss": 0.712,
"step": 2235
},
{
"epoch": 0.7168,
"grad_norm": 0.038692905227676086,
"learning_rate": 6.753735728368534e-05,
"loss": 0.6948,
"step": 2240
},
{
"epoch": 0.7184,
"grad_norm": 0.035518907504542584,
"learning_rate": 6.683871980534833e-05,
"loss": 0.707,
"step": 2245
},
{
"epoch": 0.72,
"grad_norm": 0.036175354430750345,
"learning_rate": 6.614267727799212e-05,
"loss": 0.6939,
"step": 2250
},
{
"epoch": 0.7216,
"grad_norm": 0.036533837452099954,
"learning_rate": 6.544925142081391e-05,
"loss": 0.7071,
"step": 2255
},
{
"epoch": 0.7232,
"grad_norm": 0.035541807748675644,
"learning_rate": 6.475846387136067e-05,
"loss": 0.6911,
"step": 2260
},
{
"epoch": 0.7248,
"grad_norm": 0.03727827236626591,
"learning_rate": 6.4070336184854e-05,
"loss": 0.6875,
"step": 2265
},
{
"epoch": 0.7264,
"grad_norm": 0.03858368267160788,
"learning_rate": 6.338488983351777e-05,
"loss": 0.7001,
"step": 2270
},
{
"epoch": 0.728,
"grad_norm": 0.036340556988103734,
"learning_rate": 6.270214620590773e-05,
"loss": 0.697,
"step": 2275
},
{
"epoch": 0.7296,
"grad_norm": 0.038413262159483236,
"learning_rate": 6.202212660624432e-05,
"loss": 0.7056,
"step": 2280
},
{
"epoch": 0.7312,
"grad_norm": 0.03831973777660665,
"learning_rate": 6.134485225374787e-05,
"loss": 0.6802,
"step": 2285
},
{
"epoch": 0.7328,
"grad_norm": 0.035761823589982554,
"learning_rate": 6.06703442819764e-05,
"loss": 0.6901,
"step": 2290
},
{
"epoch": 0.7344,
"grad_norm": 0.036461128986263136,
"learning_rate": 5.999862373816643e-05,
"loss": 0.6823,
"step": 2295
},
{
"epoch": 0.736,
"grad_norm": 0.03207206216632789,
"learning_rate": 5.932971158257583e-05,
"loss": 0.6931,
"step": 2300
},
{
"epoch": 0.7376,
"grad_norm": 0.036101447949103284,
"learning_rate": 5.866362868783008e-05,
"loss": 0.7031,
"step": 2305
},
{
"epoch": 0.7392,
"grad_norm": 0.0355932887214948,
"learning_rate": 5.800039583827083e-05,
"loss": 0.6828,
"step": 2310
},
{
"epoch": 0.7408,
"grad_norm": 0.03463469775440267,
"learning_rate": 5.7340033729307355e-05,
"loss": 0.688,
"step": 2315
},
{
"epoch": 0.7424,
"grad_norm": 0.03239707597454295,
"learning_rate": 5.668256296677093e-05,
"loss": 0.6874,
"step": 2320
},
{
"epoch": 0.744,
"grad_norm": 0.032819421746715025,
"learning_rate": 5.602800406627168e-05,
"loss": 0.6974,
"step": 2325
},
{
"epoch": 0.7456,
"grad_norm": 0.03481201938953441,
"learning_rate": 5.5376377452558405e-05,
"loss": 0.6841,
"step": 2330
},
{
"epoch": 0.7472,
"grad_norm": 0.032976952846685115,
"learning_rate": 5.472770345888134e-05,
"loss": 0.6835,
"step": 2335
},
{
"epoch": 0.7488,
"grad_norm": 0.03398456959274724,
"learning_rate": 5.4082002326357655e-05,
"loss": 0.6895,
"step": 2340
},
{
"epoch": 0.7504,
"grad_norm": 0.03414072023836646,
"learning_rate": 5.343929420333984e-05,
"loss": 0.6845,
"step": 2345
},
{
"epoch": 0.752,
"grad_norm": 0.038946896522945085,
"learning_rate": 5.2799599144786966e-05,
"loss": 0.7056,
"step": 2350
},
{
"epoch": 0.7536,
"grad_norm": 0.038611878374765096,
"learning_rate": 5.216293711163908e-05,
"loss": 0.6846,
"step": 2355
},
{
"epoch": 0.7552,
"grad_norm": 0.04022750003348445,
"learning_rate": 5.152932797019403e-05,
"loss": 0.6852,
"step": 2360
},
{
"epoch": 0.7568,
"grad_norm": 0.03462503063292095,
"learning_rate": 5.089879149148781e-05,
"loss": 0.6919,
"step": 2365
},
{
"epoch": 0.7584,
"grad_norm": 0.035157219428172494,
"learning_rate": 5.027134735067753e-05,
"loss": 0.6635,
"step": 2370
},
{
"epoch": 0.76,
"grad_norm": 0.03454241535056523,
"learning_rate": 4.964701512642746e-05,
"loss": 0.6836,
"step": 2375
},
{
"epoch": 0.7616,
"grad_norm": 0.03551890224974892,
"learning_rate": 4.9025814300298265e-05,
"loss": 0.689,
"step": 2380
},
{
"epoch": 0.7632,
"grad_norm": 0.03480443306167773,
"learning_rate": 4.840776425613886e-05,
"loss": 0.6818,
"step": 2385
},
{
"epoch": 0.7648,
"grad_norm": 0.04282065506147083,
"learning_rate": 4.779288427948168e-05,
"loss": 0.7078,
"step": 2390
},
{
"epoch": 0.7664,
"grad_norm": 0.04001890033275753,
"learning_rate": 4.718119355694091e-05,
"loss": 0.6972,
"step": 2395
},
{
"epoch": 0.768,
"grad_norm": 0.03615205351244455,
"learning_rate": 4.657271117561374e-05,
"loss": 0.6834,
"step": 2400
},
{
"epoch": 0.7696,
"grad_norm": 0.03741511789826621,
"learning_rate": 4.596745612248488e-05,
"loss": 0.6958,
"step": 2405
},
{
"epoch": 0.7712,
"grad_norm": 0.03785992286808968,
"learning_rate": 4.536544728383403e-05,
"loss": 0.6919,
"step": 2410
},
{
"epoch": 0.7728,
"grad_norm": 0.0342430298542495,
"learning_rate": 4.476670344464645e-05,
"loss": 0.6666,
"step": 2415
},
{
"epoch": 0.7744,
"grad_norm": 0.03435961039210759,
"learning_rate": 4.417124328802694e-05,
"loss": 0.6784,
"step": 2420
},
{
"epoch": 0.776,
"grad_norm": 0.03348688081049021,
"learning_rate": 4.357908539461678e-05,
"loss": 0.6807,
"step": 2425
},
{
"epoch": 0.7776,
"grad_norm": 0.03882741867355983,
"learning_rate": 4.299024824201402e-05,
"loss": 0.6935,
"step": 2430
},
{
"epoch": 0.7792,
"grad_norm": 0.03455997307328313,
"learning_rate": 4.240475020419675e-05,
"loss": 0.6678,
"step": 2435
},
{
"epoch": 0.7808,
"grad_norm": 0.03273696934206987,
"learning_rate": 4.1822609550950066e-05,
"loss": 0.6792,
"step": 2440
},
{
"epoch": 0.7824,
"grad_norm": 0.03561393891256124,
"learning_rate": 4.124384444729561e-05,
"loss": 0.7007,
"step": 2445
},
{
"epoch": 0.784,
"grad_norm": 0.0333436366605522,
"learning_rate": 4.0668472952924977e-05,
"loss": 0.7081,
"step": 2450
},
{
"epoch": 0.7856,
"grad_norm": 0.03400780896789157,
"learning_rate": 4.009651302163614e-05,
"loss": 0.6853,
"step": 2455
},
{
"epoch": 0.7872,
"grad_norm": 0.03441556129100297,
"learning_rate": 3.952798250077317e-05,
"loss": 0.6868,
"step": 2460
},
{
"epoch": 0.7888,
"grad_norm": 0.034964399629883444,
"learning_rate": 3.896289913066952e-05,
"loss": 0.7023,
"step": 2465
},
{
"epoch": 0.7904,
"grad_norm": 0.03287086622753301,
"learning_rate": 3.840128054409414e-05,
"loss": 0.6768,
"step": 2470
},
{
"epoch": 0.792,
"grad_norm": 0.03270391126339767,
"learning_rate": 3.784314426570156e-05,
"loss": 0.6722,
"step": 2475
},
{
"epoch": 0.7936,
"grad_norm": 0.035527816994049805,
"learning_rate": 3.728850771148486e-05,
"loss": 0.6918,
"step": 2480
},
{
"epoch": 0.7952,
"grad_norm": 0.03380039010403171,
"learning_rate": 3.67373881882323e-05,
"loss": 0.7045,
"step": 2485
},
{
"epoch": 0.7968,
"grad_norm": 0.03515245421032312,
"learning_rate": 3.618980289298736e-05,
"loss": 0.6815,
"step": 2490
},
{
"epoch": 0.7984,
"grad_norm": 0.033803992163129494,
"learning_rate": 3.5645768912512046e-05,
"loss": 0.671,
"step": 2495
},
{
"epoch": 0.8,
"grad_norm": 0.033620590297182014,
"learning_rate": 3.510530322275362e-05,
"loss": 0.7144,
"step": 2500
},
{
"epoch": 0.8016,
"grad_norm": 0.033339204297562666,
"learning_rate": 3.4568422688315025e-05,
"loss": 0.689,
"step": 2505
},
{
"epoch": 0.8032,
"grad_norm": 0.03421203186390974,
"learning_rate": 3.4035144061928635e-05,
"loss": 0.6863,
"step": 2510
},
{
"epoch": 0.8048,
"grad_norm": 0.03515095204294948,
"learning_rate": 3.350548398393345e-05,
"loss": 0.7134,
"step": 2515
},
{
"epoch": 0.8064,
"grad_norm": 0.03312271788466041,
"learning_rate": 3.297945898175584e-05,
"loss": 0.6816,
"step": 2520
},
{
"epoch": 0.808,
"grad_norm": 0.035691046358530064,
"learning_rate": 3.245708546939401e-05,
"loss": 0.6924,
"step": 2525
},
{
"epoch": 0.8096,
"grad_norm": 0.03660393192644267,
"learning_rate": 3.19383797469055e-05,
"loss": 0.6851,
"step": 2530
},
{
"epoch": 0.8112,
"grad_norm": 0.03357965660902685,
"learning_rate": 3.142335799989883e-05,
"loss": 0.6972,
"step": 2535
},
{
"epoch": 0.8128,
"grad_norm": 0.03542926472370797,
"learning_rate": 3.091203629902832e-05,
"loss": 0.6883,
"step": 2540
},
{
"epoch": 0.8144,
"grad_norm": 0.0347400277269967,
"learning_rate": 3.0404430599492636e-05,
"loss": 0.6911,
"step": 2545
},
{
"epoch": 0.816,
"grad_norm": 0.03187318572435314,
"learning_rate": 2.9900556740537026e-05,
"loss": 0.6914,
"step": 2550
},
{
"epoch": 0.8176,
"grad_norm": 0.03694045474832165,
"learning_rate": 2.9400430444958932e-05,
"loss": 0.6783,
"step": 2555
},
{
"epoch": 0.8192,
"grad_norm": 0.03742846319962199,
"learning_rate": 2.8904067318617453e-05,
"loss": 0.6811,
"step": 2560
},
{
"epoch": 0.8208,
"grad_norm": 0.03636117856176073,
"learning_rate": 2.841148284994635e-05,
"loss": 0.6751,
"step": 2565
},
{
"epoch": 0.8224,
"grad_norm": 0.03497151026361055,
"learning_rate": 2.792269240947076e-05,
"loss": 0.6828,
"step": 2570
},
{
"epoch": 0.824,
"grad_norm": 0.037020265060814364,
"learning_rate": 2.7437711249327655e-05,
"loss": 0.6807,
"step": 2575
},
{
"epoch": 0.8256,
"grad_norm": 0.03777818188790486,
"learning_rate": 2.6956554502789867e-05,
"loss": 0.6727,
"step": 2580
},
{
"epoch": 0.8272,
"grad_norm": 0.033108113036028036,
"learning_rate": 2.6479237183793712e-05,
"loss": 0.6874,
"step": 2585
},
{
"epoch": 0.8288,
"grad_norm": 0.03361982582759674,
"learning_rate": 2.6005774186470746e-05,
"loss": 0.6841,
"step": 2590
},
{
"epoch": 0.8304,
"grad_norm": 0.034688117279271555,
"learning_rate": 2.5536180284682857e-05,
"loss": 0.6768,
"step": 2595
},
{
"epoch": 0.832,
"grad_norm": 0.0355721114391825,
"learning_rate": 2.507047013156133e-05,
"loss": 0.695,
"step": 2600
},
{
"epoch": 0.8336,
"grad_norm": 0.03439958159407954,
"learning_rate": 2.4608658259049542e-05,
"loss": 0.6913,
"step": 2605
},
{
"epoch": 0.8352,
"grad_norm": 0.03567159049086682,
"learning_rate": 2.4150759077449698e-05,
"loss": 0.6803,
"step": 2610
},
{
"epoch": 0.8368,
"grad_norm": 0.03354275768498746,
"learning_rate": 2.369678687497289e-05,
"loss": 0.6868,
"step": 2615
},
{
"epoch": 0.8384,
"grad_norm": 0.0341529352716684,
"learning_rate": 2.3246755817293444e-05,
"loss": 0.6866,
"step": 2620
},
{
"epoch": 0.84,
"grad_norm": 0.03403159266100493,
"learning_rate": 2.2800679947106853e-05,
"loss": 0.681,
"step": 2625
},
{
"epoch": 0.8416,
"grad_norm": 0.03288210812033515,
"learning_rate": 2.2358573183691565e-05,
"loss": 0.6686,
"step": 2630
},
{
"epoch": 0.8432,
"grad_norm": 0.035249005655045906,
"learning_rate": 2.1920449322474742e-05,
"loss": 0.6888,
"step": 2635
},
{
"epoch": 0.8448,
"grad_norm": 0.03488362803403098,
"learning_rate": 2.1486322034601657e-05,
"loss": 0.671,
"step": 2640
},
{
"epoch": 0.8464,
"grad_norm": 0.03561191930191397,
"learning_rate": 2.105620486650914e-05,
"loss": 0.6804,
"step": 2645
},
{
"epoch": 0.848,
"grad_norm": 0.03456720687466403,
"learning_rate": 2.063011123950295e-05,
"loss": 0.6851,
"step": 2650
},
{
"epoch": 0.8496,
"grad_norm": 0.03432640424210853,
"learning_rate": 2.0208054449338878e-05,
"loss": 0.6705,
"step": 2655
},
{
"epoch": 0.8512,
"grad_norm": 0.03511000143477434,
"learning_rate": 1.979004766580797e-05,
"loss": 0.6746,
"step": 2660
},
{
"epoch": 0.8528,
"grad_norm": 0.03344677834063611,
"learning_rate": 1.9376103932325547e-05,
"loss": 0.6877,
"step": 2665
},
{
"epoch": 0.8544,
"grad_norm": 0.03435591498957771,
"learning_rate": 1.8966236165524102e-05,
"loss": 0.6893,
"step": 2670
},
{
"epoch": 0.856,
"grad_norm": 0.034662865159448594,
"learning_rate": 1.8560457154850395e-05,
"loss": 0.6766,
"step": 2675
},
{
"epoch": 0.8576,
"grad_norm": 0.03462276250146596,
"learning_rate": 1.8158779562166252e-05,
"loss": 0.6815,
"step": 2680
},
{
"epoch": 0.8592,
"grad_norm": 0.03367094557622228,
"learning_rate": 1.7761215921353576e-05,
"loss": 0.6817,
"step": 2685
},
{
"epoch": 0.8608,
"grad_norm": 0.03416564528986109,
"learning_rate": 1.7367778637923135e-05,
"loss": 0.6682,
"step": 2690
},
{
"epoch": 0.8624,
"grad_norm": 0.031887453735921126,
"learning_rate": 1.6978479988627592e-05,
"loss": 0.6725,
"step": 2695
},
{
"epoch": 0.864,
"grad_norm": 0.03311007741244363,
"learning_rate": 1.6593332121078295e-05,
"loss": 0.6839,
"step": 2700
},
{
"epoch": 0.8656,
"grad_norm": 0.03440719274136189,
"learning_rate": 1.6212347053366287e-05,
"loss": 0.6916,
"step": 2705
},
{
"epoch": 0.8672,
"grad_norm": 0.032512249567439665,
"learning_rate": 1.5835536673687278e-05,
"loss": 0.651,
"step": 2710
},
{
"epoch": 0.8688,
"grad_norm": 0.035144382060590305,
"learning_rate": 1.5462912739970663e-05,
"loss": 0.6684,
"step": 2715
},
{
"epoch": 0.8704,
"grad_norm": 0.037061202763024576,
"learning_rate": 1.5094486879512786e-05,
"loss": 0.6667,
"step": 2720
},
{
"epoch": 0.872,
"grad_norm": 0.03422207800786857,
"learning_rate": 1.4730270588613874e-05,
"loss": 0.6842,
"step": 2725
},
{
"epoch": 0.8736,
"grad_norm": 0.03383379241828916,
"learning_rate": 1.437027523221948e-05,
"loss": 0.6744,
"step": 2730
},
{
"epoch": 0.8752,
"grad_norm": 0.0377633250822453,
"learning_rate": 1.4014512043565812e-05,
"loss": 0.6724,
"step": 2735
},
{
"epoch": 0.8768,
"grad_norm": 0.03382097364001704,
"learning_rate": 1.3662992123829164e-05,
"loss": 0.7102,
"step": 2740
},
{
"epoch": 0.8784,
"grad_norm": 0.03482496157037288,
"learning_rate": 1.3315726441779629e-05,
"loss": 0.6792,
"step": 2745
},
{
"epoch": 0.88,
"grad_norm": 0.03437993429478452,
"learning_rate": 1.2972725833438752e-05,
"loss": 0.6765,
"step": 2750
},
{
"epoch": 0.8816,
"grad_norm": 0.03561497033864348,
"learning_rate": 1.2634001001741373e-05,
"loss": 0.6698,
"step": 2755
},
{
"epoch": 0.8832,
"grad_norm": 0.03593686290956888,
"learning_rate": 1.2299562516201739e-05,
"loss": 0.6878,
"step": 2760
},
{
"epoch": 0.8848,
"grad_norm": 0.03543056527199493,
"learning_rate": 1.1969420812583603e-05,
"loss": 0.6956,
"step": 2765
},
{
"epoch": 0.8864,
"grad_norm": 0.03279393333627028,
"learning_rate": 1.1643586192574684e-05,
"loss": 0.665,
"step": 2770
},
{
"epoch": 0.888,
"grad_norm": 0.03584649556457144,
"learning_rate": 1.1322068823465125e-05,
"loss": 0.6716,
"step": 2775
},
{
"epoch": 0.8896,
"grad_norm": 0.033357243313891595,
"learning_rate": 1.1004878737830363e-05,
"loss": 0.6625,
"step": 2780
},
{
"epoch": 0.8912,
"grad_norm": 0.032759236080115814,
"learning_rate": 1.06920258332179e-05,
"loss": 0.6795,
"step": 2785
},
{
"epoch": 0.8928,
"grad_norm": 0.03336069117498674,
"learning_rate": 1.0383519871838587e-05,
"loss": 0.6655,
"step": 2790
},
{
"epoch": 0.8944,
"grad_norm": 0.03299826661329797,
"learning_rate": 1.0079370480261984e-05,
"loss": 0.6687,
"step": 2795
},
{
"epoch": 0.896,
"grad_norm": 0.031209541292076762,
"learning_rate": 9.779587149115891e-06,
"loss": 0.68,
"step": 2800
},
{
"epoch": 0.8976,
"grad_norm": 0.035130456552499516,
"learning_rate": 9.484179232790389e-06,
"loss": 0.6609,
"step": 2805
},
{
"epoch": 0.8992,
"grad_norm": 0.035181507878859576,
"learning_rate": 9.193155949145736e-06,
"loss": 0.6666,
"step": 2810
},
{
"epoch": 0.9008,
"grad_norm": 0.03325417936643577,
"learning_rate": 8.906526379224865e-06,
"loss": 0.6961,
"step": 2815
},
{
"epoch": 0.9024,
"grad_norm": 0.03470125347520115,
"learning_rate": 8.624299466969964e-06,
"loss": 0.6765,
"step": 2820
},
{
"epoch": 0.904,
"grad_norm": 0.03107911564006217,
"learning_rate": 8.346484018943428e-06,
"loss": 0.673,
"step": 2825
},
{
"epoch": 0.9056,
"grad_norm": 0.03738582706757996,
"learning_rate": 8.073088704053082e-06,
"loss": 0.6816,
"step": 2830
},
{
"epoch": 0.9072,
"grad_norm": 0.034220413141167035,
"learning_rate": 7.80412205328153e-06,
"loss": 0.6792,
"step": 2835
},
{
"epoch": 0.9088,
"grad_norm": 0.033016368162411415,
"learning_rate": 7.539592459420219e-06,
"loss": 0.6641,
"step": 2840
},
{
"epoch": 0.9104,
"grad_norm": 0.0335933143687034,
"learning_rate": 7.279508176807258e-06,
"loss": 0.6759,
"step": 2845
},
{
"epoch": 0.912,
"grad_norm": 0.03250397034496483,
"learning_rate": 7.023877321070071e-06,
"loss": 0.6771,
"step": 2850
},
{
"epoch": 0.9136,
"grad_norm": 0.03419525821241761,
"learning_rate": 6.7727078688720515e-06,
"loss": 0.6883,
"step": 2855
},
{
"epoch": 0.9152,
"grad_norm": 0.0334179547568993,
"learning_rate": 6.526007657663656e-06,
"loss": 0.6689,
"step": 2860
},
{
"epoch": 0.9168,
"grad_norm": 0.033761022795765566,
"learning_rate": 6.283784385437962e-06,
"loss": 0.677,
"step": 2865
},
{
"epoch": 0.9184,
"grad_norm": 0.032636036075434506,
"learning_rate": 6.046045610490324e-06,
"loss": 0.6788,
"step": 2870
},
{
"epoch": 0.92,
"grad_norm": 0.03640168972043887,
"learning_rate": 5.8127987511825516e-06,
"loss": 0.6864,
"step": 2875
},
{
"epoch": 0.9216,
"grad_norm": 0.03297895881000032,
"learning_rate": 5.584051085711488e-06,
"loss": 0.6624,
"step": 2880
},
{
"epoch": 0.9232,
"grad_norm": 0.032448998874942715,
"learning_rate": 5.359809751881833e-06,
"loss": 0.6627,
"step": 2885
},
{
"epoch": 0.9248,
"grad_norm": 0.032596779472805294,
"learning_rate": 5.140081746883546e-06,
"loss": 0.6652,
"step": 2890
},
{
"epoch": 0.9264,
"grad_norm": 0.03442429036474717,
"learning_rate": 4.924873927073292e-06,
"loss": 0.67,
"step": 2895
},
{
"epoch": 0.928,
"grad_norm": 0.034485516453528195,
"learning_rate": 4.714193007760681e-06,
"loss": 0.6784,
"step": 2900
},
{
"epoch": 0.9296,
"grad_norm": 0.03309853498364676,
"learning_rate": 4.5080455629986465e-06,
"loss": 0.6814,
"step": 2905
},
{
"epoch": 0.9312,
"grad_norm": 0.033126906800632835,
"learning_rate": 4.306438025378317e-06,
"loss": 0.6829,
"step": 2910
},
{
"epoch": 0.9328,
"grad_norm": 0.033356739926376194,
"learning_rate": 4.109376685828336e-06,
"loss": 0.6882,
"step": 2915
},
{
"epoch": 0.9344,
"grad_norm": 0.0320316910992072,
"learning_rate": 3.916867693418474e-06,
"loss": 0.6724,
"step": 2920
},
{
"epoch": 0.936,
"grad_norm": 0.032605973162921166,
"learning_rate": 3.7289170551678626e-06,
"loss": 0.6775,
"step": 2925
},
{
"epoch": 0.9376,
"grad_norm": 0.033867274884899295,
"learning_rate": 3.545530635857463e-06,
"loss": 0.6792,
"step": 2930
},
{
"epoch": 0.9392,
"grad_norm": 0.035267733798289305,
"learning_rate": 3.366714157847078e-06,
"loss": 0.6732,
"step": 2935
},
{
"epoch": 0.9408,
"grad_norm": 0.03437896514874513,
"learning_rate": 3.192473200896828e-06,
"loss": 0.6855,
"step": 2940
},
{
"epoch": 0.9424,
"grad_norm": 0.034980736817982326,
"learning_rate": 3.0228132019930407e-06,
"loss": 0.6926,
"step": 2945
},
{
"epoch": 0.944,
"grad_norm": 0.03176635126743629,
"learning_rate": 2.8577394551785705e-06,
"loss": 0.6751,
"step": 2950
},
{
"epoch": 0.9456,
"grad_norm": 0.0325162512058197,
"learning_rate": 2.6972571113875798e-06,
"loss": 0.6773,
"step": 2955
},
{
"epoch": 0.9472,
"grad_norm": 0.032425959850668325,
"learning_rate": 2.5413711782848845e-06,
"loss": 0.6716,
"step": 2960
},
{
"epoch": 0.9488,
"grad_norm": 0.03372199529735856,
"learning_rate": 2.39008652010963e-06,
"loss": 0.6785,
"step": 2965
},
{
"epoch": 0.9504,
"grad_norm": 0.035790626790530945,
"learning_rate": 2.243407857523544e-06,
"loss": 0.6949,
"step": 2970
},
{
"epoch": 0.952,
"grad_norm": 0.03338215436619671,
"learning_rate": 2.1013397674636057e-06,
"loss": 0.6666,
"step": 2975
},
{
"epoch": 0.9536,
"grad_norm": 0.03485173137310071,
"learning_rate": 1.9638866829992606e-06,
"loss": 0.6758,
"step": 2980
},
{
"epoch": 0.9552,
"grad_norm": 0.03254224723678366,
"learning_rate": 1.8310528931940627e-06,
"loss": 0.6822,
"step": 2985
},
{
"epoch": 0.9568,
"grad_norm": 0.03309110315976776,
"learning_rate": 1.702842542971866e-06,
"loss": 0.6733,
"step": 2990
},
{
"epoch": 0.9584,
"grad_norm": 0.031592493593121926,
"learning_rate": 1.5792596329874451e-06,
"loss": 0.689,
"step": 2995
},
{
"epoch": 0.96,
"grad_norm": 0.03284447408135021,
"learning_rate": 1.4603080195017103e-06,
"loss": 0.6816,
"step": 3000
},
{
"epoch": 0.9616,
"grad_norm": 0.03371203971154065,
"learning_rate": 1.3459914142613381e-06,
"loss": 0.6736,
"step": 3005
},
{
"epoch": 0.9632,
"grad_norm": 0.035757427807579185,
"learning_rate": 1.2363133843829976e-06,
"loss": 0.6872,
"step": 3010
},
{
"epoch": 0.9648,
"grad_norm": 0.033383743233224024,
"learning_rate": 1.1312773522419716e-06,
"loss": 0.6769,
"step": 3015
},
{
"epoch": 0.9664,
"grad_norm": 0.034245949880416675,
"learning_rate": 1.0308865953654267e-06,
"loss": 0.6935,
"step": 3020
},
{
"epoch": 0.968,
"grad_norm": 0.03459041629757538,
"learning_rate": 9.351442463301273e-07,
"loss": 0.6677,
"step": 3025
},
{
"epoch": 0.9696,
"grad_norm": 0.0366654930646669,
"learning_rate": 8.440532926646315e-07,
"loss": 0.6869,
"step": 3030
},
{
"epoch": 0.9712,
"grad_norm": 0.03317086463770714,
"learning_rate": 7.576165767561815e-07,
"loss": 0.6901,
"step": 3035
},
{
"epoch": 0.9728,
"grad_norm": 0.03278628874901946,
"learning_rate": 6.758367957619082e-07,
"loss": 0.6699,
"step": 3040
},
{
"epoch": 0.9744,
"grad_norm": 0.033907781511625844,
"learning_rate": 5.98716501524732e-07,
"loss": 0.6672,
"step": 3045
},
{
"epoch": 0.976,
"grad_norm": 0.03326165151325172,
"learning_rate": 5.262581004936927e-07,
"loss": 0.677,
"step": 3050
},
{
"epoch": 0.9776,
"grad_norm": 0.034089530502770686,
"learning_rate": 4.5846385364886053e-07,
"loss": 0.6752,
"step": 3055
},
{
"epoch": 0.9792,
"grad_norm": 0.0313067667316255,
"learning_rate": 3.953358764308412e-07,
"loss": 0.6693,
"step": 3060
},
{
"epoch": 0.9808,
"grad_norm": 0.033836241370780035,
"learning_rate": 3.3687613867469653e-07,
"loss": 0.6859,
"step": 3065
},
{
"epoch": 0.9824,
"grad_norm": 0.03468937319358296,
"learning_rate": 2.830864645485265e-07,
"loss": 0.6813,
"step": 3070
},
{
"epoch": 0.984,
"grad_norm": 0.03339345202634523,
"learning_rate": 2.3396853249653124e-07,
"loss": 0.6825,
"step": 3075
},
{
"epoch": 0.9856,
"grad_norm": 0.03223757643782052,
"learning_rate": 1.895238751866035e-07,
"loss": 0.679,
"step": 3080
},
{
"epoch": 0.9872,
"grad_norm": 0.03278384204556813,
"learning_rate": 1.4975387946256635e-07,
"loss": 0.6727,
"step": 3085
},
{
"epoch": 0.9888,
"grad_norm": 0.03296114059220446,
"learning_rate": 1.1465978630085803e-07,
"loss": 0.6813,
"step": 3090
},
{
"epoch": 0.9904,
"grad_norm": 0.0312953732379823,
"learning_rate": 8.424269077182965e-08,
"loss": 0.6684,
"step": 3095
},
{
"epoch": 0.992,
"grad_norm": 0.03153039184869774,
"learning_rate": 5.8503542005555734e-08,
"loss": 0.6755,
"step": 3100
},
{
"epoch": 0.9936,
"grad_norm": 0.033138970245913064,
"learning_rate": 3.7443143162207976e-08,
"loss": 0.6981,
"step": 3105
},
{
"epoch": 0.9952,
"grad_norm": 0.03351683971290094,
"learning_rate": 2.1062151407025274e-08,
"loss": 0.6906,
"step": 3110
},
{
"epoch": 0.9968,
"grad_norm": 0.034446874139718094,
"learning_rate": 9.36107788981344e-09,
"loss": 0.6835,
"step": 3115
},
{
"epoch": 0.9984,
"grad_norm": 0.032846170943355714,
"learning_rate": 2.340287728941348e-09,
"loss": 0.6594,
"step": 3120
},
{
"epoch": 1.0,
"grad_norm": 0.03376153891544424,
"learning_rate": 0.0,
"loss": 0.6914,
"step": 3125
},
{
"epoch": 1.0,
"eval_loss": 0.6845724582672119,
"eval_runtime": 258.165,
"eval_samples_per_second": 120.435,
"eval_steps_per_second": 3.765,
"step": 3125
},
{
"epoch": 1.0,
"step": 3125,
"total_flos": 514082196226048.0,
"train_loss": 0.7505479205322265,
"train_runtime": 13621.5051,
"train_samples_per_second": 29.365,
"train_steps_per_second": 0.229
}
],
"logging_steps": 5,
"max_steps": 3125,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 500,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": false,
"should_training_stop": false
},
"attributes": {}
}
},
"total_flos": 514082196226048.0,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}