Files
Qwen2.5-1.5B-Open-R1-Distill/trainer_state.json
ModelHub XC 2076a8b8d6 初始化项目,由ModelHub XC社区提供模型
Model: wooferclaw/Qwen2.5-1.5B-Open-R1-Distill
Source: Original Platform
2026-06-22 14:34:52 +08:00

60075 lines
1.4 MiB

{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 5.0,
"eval_steps": 500,
"global_step": 42880,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0005830223880597014,
"grad_norm": 2.550195533598444,
"learning_rate": 5.827505827505828e-07,
"loss": 0.8852,
"step": 5
},
{
"epoch": 0.0011660447761194029,
"grad_norm": 2.2188855019798273,
"learning_rate": 1.1655011655011657e-06,
"loss": 0.8311,
"step": 10
},
{
"epoch": 0.0017490671641791045,
"grad_norm": 1.7967346771825998,
"learning_rate": 1.7482517482517485e-06,
"loss": 0.8113,
"step": 15
},
{
"epoch": 0.0023320895522388058,
"grad_norm": 1.618619190869765,
"learning_rate": 2.3310023310023313e-06,
"loss": 0.8333,
"step": 20
},
{
"epoch": 0.0029151119402985076,
"grad_norm": 1.137673984585037,
"learning_rate": 2.9137529137529138e-06,
"loss": 0.7745,
"step": 25
},
{
"epoch": 0.003498134328358209,
"grad_norm": 1.1933766134733839,
"learning_rate": 3.496503496503497e-06,
"loss": 0.7904,
"step": 30
},
{
"epoch": 0.00408115671641791,
"grad_norm": 1.0157304336095063,
"learning_rate": 4.079254079254079e-06,
"loss": 0.7624,
"step": 35
},
{
"epoch": 0.0046641791044776115,
"grad_norm": 0.9429820002012593,
"learning_rate": 4.662004662004663e-06,
"loss": 0.7285,
"step": 40
},
{
"epoch": 0.005247201492537314,
"grad_norm": 0.915430165066572,
"learning_rate": 5.244755244755245e-06,
"loss": 0.7469,
"step": 45
},
{
"epoch": 0.005830223880597015,
"grad_norm": 0.8161946795665328,
"learning_rate": 5.8275058275058275e-06,
"loss": 0.707,
"step": 50
},
{
"epoch": 0.006413246268656717,
"grad_norm": 0.8339668281290502,
"learning_rate": 6.41025641025641e-06,
"loss": 0.6905,
"step": 55
},
{
"epoch": 0.006996268656716418,
"grad_norm": 0.8574171717816127,
"learning_rate": 6.993006993006994e-06,
"loss": 0.7698,
"step": 60
},
{
"epoch": 0.00757929104477612,
"grad_norm": 0.8619956369841163,
"learning_rate": 7.5757575757575764e-06,
"loss": 0.7082,
"step": 65
},
{
"epoch": 0.00816231343283582,
"grad_norm": 0.8045557883502801,
"learning_rate": 8.158508158508159e-06,
"loss": 0.6587,
"step": 70
},
{
"epoch": 0.008745335820895522,
"grad_norm": 0.7946638318702278,
"learning_rate": 8.741258741258741e-06,
"loss": 0.6539,
"step": 75
},
{
"epoch": 0.009328358208955223,
"grad_norm": 0.8503979707980048,
"learning_rate": 9.324009324009325e-06,
"loss": 0.6879,
"step": 80
},
{
"epoch": 0.009911380597014926,
"grad_norm": 0.804711208532189,
"learning_rate": 9.906759906759908e-06,
"loss": 0.6743,
"step": 85
},
{
"epoch": 0.010494402985074628,
"grad_norm": 0.7951085316698083,
"learning_rate": 1.048951048951049e-05,
"loss": 0.6586,
"step": 90
},
{
"epoch": 0.011077425373134329,
"grad_norm": 0.793128900429868,
"learning_rate": 1.1072261072261073e-05,
"loss": 0.6842,
"step": 95
},
{
"epoch": 0.01166044776119403,
"grad_norm": 0.8075245421009535,
"learning_rate": 1.1655011655011655e-05,
"loss": 0.6684,
"step": 100
},
{
"epoch": 0.012243470149253732,
"grad_norm": 0.8581518728101114,
"learning_rate": 1.2237762237762239e-05,
"loss": 0.6207,
"step": 105
},
{
"epoch": 0.012826492537313433,
"grad_norm": 0.9082387952747121,
"learning_rate": 1.282051282051282e-05,
"loss": 0.6678,
"step": 110
},
{
"epoch": 0.013409514925373135,
"grad_norm": 0.9029958898223162,
"learning_rate": 1.3403263403263406e-05,
"loss": 0.6394,
"step": 115
},
{
"epoch": 0.013992537313432836,
"grad_norm": 0.8377367698700525,
"learning_rate": 1.3986013986013988e-05,
"loss": 0.6707,
"step": 120
},
{
"epoch": 0.014575559701492538,
"grad_norm": 0.7587966655835441,
"learning_rate": 1.456876456876457e-05,
"loss": 0.6258,
"step": 125
},
{
"epoch": 0.01515858208955224,
"grad_norm": 0.8173306054331542,
"learning_rate": 1.5151515151515153e-05,
"loss": 0.6619,
"step": 130
},
{
"epoch": 0.01574160447761194,
"grad_norm": 0.9376337343046882,
"learning_rate": 1.5734265734265734e-05,
"loss": 0.6673,
"step": 135
},
{
"epoch": 0.01632462686567164,
"grad_norm": 0.9363652268784762,
"learning_rate": 1.6317016317016318e-05,
"loss": 0.668,
"step": 140
},
{
"epoch": 0.016907649253731342,
"grad_norm": 0.8794426221093276,
"learning_rate": 1.68997668997669e-05,
"loss": 0.6569,
"step": 145
},
{
"epoch": 0.017490671641791043,
"grad_norm": 0.9407491307884387,
"learning_rate": 1.7482517482517483e-05,
"loss": 0.6778,
"step": 150
},
{
"epoch": 0.018073694029850745,
"grad_norm": 0.8720667910368793,
"learning_rate": 1.8065268065268067e-05,
"loss": 0.6291,
"step": 155
},
{
"epoch": 0.018656716417910446,
"grad_norm": 0.8603987252532928,
"learning_rate": 1.864801864801865e-05,
"loss": 0.6372,
"step": 160
},
{
"epoch": 0.019239738805970148,
"grad_norm": 0.8449689793110765,
"learning_rate": 1.923076923076923e-05,
"loss": 0.6586,
"step": 165
},
{
"epoch": 0.019822761194029852,
"grad_norm": 1.0325515822144362,
"learning_rate": 1.9813519813519816e-05,
"loss": 0.6386,
"step": 170
},
{
"epoch": 0.020405783582089554,
"grad_norm": 0.8513517345308381,
"learning_rate": 2.0396270396270396e-05,
"loss": 0.6311,
"step": 175
},
{
"epoch": 0.020988805970149255,
"grad_norm": 0.8847582203013,
"learning_rate": 2.097902097902098e-05,
"loss": 0.6834,
"step": 180
},
{
"epoch": 0.021571828358208957,
"grad_norm": 0.9006033563159529,
"learning_rate": 2.156177156177156e-05,
"loss": 0.6381,
"step": 185
},
{
"epoch": 0.022154850746268658,
"grad_norm": 0.8147114381410464,
"learning_rate": 2.2144522144522145e-05,
"loss": 0.6432,
"step": 190
},
{
"epoch": 0.02273787313432836,
"grad_norm": 0.9560611394473829,
"learning_rate": 2.272727272727273e-05,
"loss": 0.6261,
"step": 195
},
{
"epoch": 0.02332089552238806,
"grad_norm": 0.7951106600912393,
"learning_rate": 2.331002331002331e-05,
"loss": 0.6307,
"step": 200
},
{
"epoch": 0.023903917910447763,
"grad_norm": 0.855140755847949,
"learning_rate": 2.3892773892773894e-05,
"loss": 0.6213,
"step": 205
},
{
"epoch": 0.024486940298507464,
"grad_norm": 1.0420866065102492,
"learning_rate": 2.4475524475524478e-05,
"loss": 0.6275,
"step": 210
},
{
"epoch": 0.025069962686567165,
"grad_norm": 0.89019303742937,
"learning_rate": 2.505827505827506e-05,
"loss": 0.6275,
"step": 215
},
{
"epoch": 0.025652985074626867,
"grad_norm": 0.9758673014251983,
"learning_rate": 2.564102564102564e-05,
"loss": 0.6631,
"step": 220
},
{
"epoch": 0.02623600746268657,
"grad_norm": 0.8596860100346014,
"learning_rate": 2.6223776223776224e-05,
"loss": 0.6216,
"step": 225
},
{
"epoch": 0.02681902985074627,
"grad_norm": 0.8908792386022747,
"learning_rate": 2.680652680652681e-05,
"loss": 0.6406,
"step": 230
},
{
"epoch": 0.02740205223880597,
"grad_norm": 0.8721610635945679,
"learning_rate": 2.738927738927739e-05,
"loss": 0.6216,
"step": 235
},
{
"epoch": 0.027985074626865673,
"grad_norm": 0.8671325117829104,
"learning_rate": 2.7972027972027976e-05,
"loss": 0.6713,
"step": 240
},
{
"epoch": 0.028568097014925374,
"grad_norm": 0.8803714665797042,
"learning_rate": 2.8554778554778557e-05,
"loss": 0.629,
"step": 245
},
{
"epoch": 0.029151119402985076,
"grad_norm": 0.8308844754653052,
"learning_rate": 2.913752913752914e-05,
"loss": 0.6373,
"step": 250
},
{
"epoch": 0.029734141791044777,
"grad_norm": 0.829842731040146,
"learning_rate": 2.972027972027972e-05,
"loss": 0.6163,
"step": 255
},
{
"epoch": 0.03031716417910448,
"grad_norm": 0.9311493282737088,
"learning_rate": 3.0303030303030306e-05,
"loss": 0.6448,
"step": 260
},
{
"epoch": 0.03090018656716418,
"grad_norm": 0.8380478242539974,
"learning_rate": 3.088578088578088e-05,
"loss": 0.6123,
"step": 265
},
{
"epoch": 0.03148320895522388,
"grad_norm": 0.8612483599648676,
"learning_rate": 3.146853146853147e-05,
"loss": 0.6206,
"step": 270
},
{
"epoch": 0.03206623134328358,
"grad_norm": 0.7504191034824335,
"learning_rate": 3.205128205128206e-05,
"loss": 0.5956,
"step": 275
},
{
"epoch": 0.03264925373134328,
"grad_norm": 0.8435455726913331,
"learning_rate": 3.2634032634032635e-05,
"loss": 0.6147,
"step": 280
},
{
"epoch": 0.033232276119402986,
"grad_norm": 0.8228330420668449,
"learning_rate": 3.321678321678322e-05,
"loss": 0.6285,
"step": 285
},
{
"epoch": 0.033815298507462684,
"grad_norm": 0.7466739529712338,
"learning_rate": 3.37995337995338e-05,
"loss": 0.62,
"step": 290
},
{
"epoch": 0.03439832089552239,
"grad_norm": 0.8056760014927022,
"learning_rate": 3.438228438228439e-05,
"loss": 0.6216,
"step": 295
},
{
"epoch": 0.034981343283582086,
"grad_norm": 0.8113797852368729,
"learning_rate": 3.4965034965034965e-05,
"loss": 0.6105,
"step": 300
},
{
"epoch": 0.03556436567164179,
"grad_norm": 0.8890502750300378,
"learning_rate": 3.554778554778555e-05,
"loss": 0.6139,
"step": 305
},
{
"epoch": 0.03614738805970149,
"grad_norm": 0.7687545593968855,
"learning_rate": 3.613053613053613e-05,
"loss": 0.614,
"step": 310
},
{
"epoch": 0.036730410447761194,
"grad_norm": 0.8336903362213683,
"learning_rate": 3.671328671328672e-05,
"loss": 0.6278,
"step": 315
},
{
"epoch": 0.03731343283582089,
"grad_norm": 0.869875611794232,
"learning_rate": 3.72960372960373e-05,
"loss": 0.6645,
"step": 320
},
{
"epoch": 0.0378964552238806,
"grad_norm": 0.789926244606821,
"learning_rate": 3.787878787878788e-05,
"loss": 0.6027,
"step": 325
},
{
"epoch": 0.038479477611940295,
"grad_norm": 0.8392459694319648,
"learning_rate": 3.846153846153846e-05,
"loss": 0.6567,
"step": 330
},
{
"epoch": 0.0390625,
"grad_norm": 0.7918852769546142,
"learning_rate": 3.904428904428905e-05,
"loss": 0.6453,
"step": 335
},
{
"epoch": 0.039645522388059705,
"grad_norm": 0.7819447011686889,
"learning_rate": 3.962703962703963e-05,
"loss": 0.6,
"step": 340
},
{
"epoch": 0.0402285447761194,
"grad_norm": 0.7599671478703276,
"learning_rate": 4.020979020979021e-05,
"loss": 0.6097,
"step": 345
},
{
"epoch": 0.04081156716417911,
"grad_norm": 0.8770061789317196,
"learning_rate": 4.079254079254079e-05,
"loss": 0.6199,
"step": 350
},
{
"epoch": 0.041394589552238806,
"grad_norm": 0.8228590505028837,
"learning_rate": 4.1375291375291377e-05,
"loss": 0.63,
"step": 355
},
{
"epoch": 0.04197761194029851,
"grad_norm": 0.9266679768403295,
"learning_rate": 4.195804195804196e-05,
"loss": 0.6611,
"step": 360
},
{
"epoch": 0.04256063432835821,
"grad_norm": 0.776925533918814,
"learning_rate": 4.254079254079254e-05,
"loss": 0.6383,
"step": 365
},
{
"epoch": 0.043143656716417914,
"grad_norm": 0.7652459941000134,
"learning_rate": 4.312354312354312e-05,
"loss": 0.5967,
"step": 370
},
{
"epoch": 0.04372667910447761,
"grad_norm": 0.8438476797737248,
"learning_rate": 4.370629370629371e-05,
"loss": 0.6146,
"step": 375
},
{
"epoch": 0.044309701492537316,
"grad_norm": 0.8158888620815322,
"learning_rate": 4.428904428904429e-05,
"loss": 0.6237,
"step": 380
},
{
"epoch": 0.044892723880597014,
"grad_norm": 0.9144427364330033,
"learning_rate": 4.4871794871794874e-05,
"loss": 0.6098,
"step": 385
},
{
"epoch": 0.04547574626865672,
"grad_norm": 0.7260277122998825,
"learning_rate": 4.545454545454546e-05,
"loss": 0.6105,
"step": 390
},
{
"epoch": 0.04605876865671642,
"grad_norm": 0.774195742981333,
"learning_rate": 4.603729603729604e-05,
"loss": 0.6182,
"step": 395
},
{
"epoch": 0.04664179104477612,
"grad_norm": 0.8931158013280471,
"learning_rate": 4.662004662004662e-05,
"loss": 0.6269,
"step": 400
},
{
"epoch": 0.04722481343283582,
"grad_norm": 0.7219414153894123,
"learning_rate": 4.7202797202797204e-05,
"loss": 0.6159,
"step": 405
},
{
"epoch": 0.047807835820895525,
"grad_norm": 0.7392180125184434,
"learning_rate": 4.778554778554779e-05,
"loss": 0.6597,
"step": 410
},
{
"epoch": 0.04839085820895522,
"grad_norm": 0.7724562759848209,
"learning_rate": 4.836829836829837e-05,
"loss": 0.6588,
"step": 415
},
{
"epoch": 0.04897388059701493,
"grad_norm": 0.7698999004589875,
"learning_rate": 4.8951048951048956e-05,
"loss": 0.6431,
"step": 420
},
{
"epoch": 0.049556902985074626,
"grad_norm": 0.756825275252759,
"learning_rate": 4.9533799533799534e-05,
"loss": 0.6475,
"step": 425
},
{
"epoch": 0.05013992537313433,
"grad_norm": 0.7182404605677212,
"learning_rate": 4.9999998327150664e-05,
"loss": 0.6519,
"step": 430
},
{
"epoch": 0.05072294776119403,
"grad_norm": 0.7352734921536612,
"learning_rate": 4.999993977744981e-05,
"loss": 0.6367,
"step": 435
},
{
"epoch": 0.051305970149253734,
"grad_norm": 0.8663133271054537,
"learning_rate": 4.9999797585530614e-05,
"loss": 0.6367,
"step": 440
},
{
"epoch": 0.05188899253731343,
"grad_norm": 0.8007074508867144,
"learning_rate": 4.9999571751921666e-05,
"loss": 0.649,
"step": 445
},
{
"epoch": 0.05247201492537314,
"grad_norm": 0.8808725057777399,
"learning_rate": 4.999926227746247e-05,
"loss": 0.6407,
"step": 450
},
{
"epoch": 0.053055037313432835,
"grad_norm": 0.8207754194931053,
"learning_rate": 4.999886916330351e-05,
"loss": 0.6664,
"step": 455
},
{
"epoch": 0.05363805970149254,
"grad_norm": 0.7200644017362087,
"learning_rate": 4.9998392410906135e-05,
"loss": 0.6264,
"step": 460
},
{
"epoch": 0.05422108208955224,
"grad_norm": 0.7655394577960307,
"learning_rate": 4.9997832022042676e-05,
"loss": 0.6854,
"step": 465
},
{
"epoch": 0.05480410447761194,
"grad_norm": 1.5845328151194846,
"learning_rate": 4.9997187998796316e-05,
"loss": 0.6092,
"step": 470
},
{
"epoch": 0.05538712686567164,
"grad_norm": 0.763948541759687,
"learning_rate": 4.9996460343561184e-05,
"loss": 0.6601,
"step": 475
},
{
"epoch": 0.055970149253731345,
"grad_norm": 0.7370601652088519,
"learning_rate": 4.99956490590423e-05,
"loss": 0.6062,
"step": 480
},
{
"epoch": 0.05655317164179104,
"grad_norm": 0.6503724111090272,
"learning_rate": 4.9994754148255566e-05,
"loss": 0.597,
"step": 485
},
{
"epoch": 0.05713619402985075,
"grad_norm": 0.642633522142368,
"learning_rate": 4.999377561452776e-05,
"loss": 0.633,
"step": 490
},
{
"epoch": 0.057719216417910446,
"grad_norm": 0.6866312480811294,
"learning_rate": 4.999271346149652e-05,
"loss": 0.6421,
"step": 495
},
{
"epoch": 0.05830223880597015,
"grad_norm": 0.6472205796291965,
"learning_rate": 4.999156769311035e-05,
"loss": 0.615,
"step": 500
},
{
"epoch": 0.05888526119402985,
"grad_norm": 0.6977115669022897,
"learning_rate": 4.999033831362857e-05,
"loss": 0.6214,
"step": 505
},
{
"epoch": 0.059468283582089554,
"grad_norm": 0.6986043336376292,
"learning_rate": 4.998902532762132e-05,
"loss": 0.6193,
"step": 510
},
{
"epoch": 0.06005130597014925,
"grad_norm": 0.7475999210726344,
"learning_rate": 4.9987628739969554e-05,
"loss": 0.6224,
"step": 515
},
{
"epoch": 0.06063432835820896,
"grad_norm": 0.6813745517845872,
"learning_rate": 4.9986148555865016e-05,
"loss": 0.6177,
"step": 520
},
{
"epoch": 0.061217350746268655,
"grad_norm": 0.7015168996041015,
"learning_rate": 4.9984584780810196e-05,
"loss": 0.6768,
"step": 525
},
{
"epoch": 0.06180037313432836,
"grad_norm": 0.7037997780013406,
"learning_rate": 4.998293742061833e-05,
"loss": 0.6742,
"step": 530
},
{
"epoch": 0.06238339552238806,
"grad_norm": 0.7156012560309609,
"learning_rate": 4.998120648141338e-05,
"loss": 0.6304,
"step": 535
},
{
"epoch": 0.06296641791044776,
"grad_norm": 0.711127994617591,
"learning_rate": 4.997939196963004e-05,
"loss": 0.6765,
"step": 540
},
{
"epoch": 0.06354944029850747,
"grad_norm": 0.7367809530722819,
"learning_rate": 4.997749389201363e-05,
"loss": 0.6443,
"step": 545
},
{
"epoch": 0.06413246268656717,
"grad_norm": 0.7023254028972605,
"learning_rate": 4.997551225562014e-05,
"loss": 0.614,
"step": 550
},
{
"epoch": 0.06471548507462686,
"grad_norm": 0.7934049324407979,
"learning_rate": 4.99734470678162e-05,
"loss": 0.6804,
"step": 555
},
{
"epoch": 0.06529850746268656,
"grad_norm": 0.7211790595438915,
"learning_rate": 4.997129833627902e-05,
"loss": 0.6022,
"step": 560
},
{
"epoch": 0.06588152985074627,
"grad_norm": 0.7803510446535082,
"learning_rate": 4.996906606899639e-05,
"loss": 0.6324,
"step": 565
},
{
"epoch": 0.06646455223880597,
"grad_norm": 0.7268141035030203,
"learning_rate": 4.996675027426662e-05,
"loss": 0.6244,
"step": 570
},
{
"epoch": 0.06704757462686567,
"grad_norm": 1.937711458766374,
"learning_rate": 4.9964350960698564e-05,
"loss": 0.637,
"step": 575
},
{
"epoch": 0.06763059701492537,
"grad_norm": 0.6807880062940403,
"learning_rate": 4.996186813721152e-05,
"loss": 0.6009,
"step": 580
},
{
"epoch": 0.06821361940298508,
"grad_norm": 0.7031632269645188,
"learning_rate": 4.995930181303522e-05,
"loss": 0.6312,
"step": 585
},
{
"epoch": 0.06879664179104478,
"grad_norm": 0.6771751142969727,
"learning_rate": 4.995665199770986e-05,
"loss": 0.6604,
"step": 590
},
{
"epoch": 0.06937966417910447,
"grad_norm": 0.711794153251972,
"learning_rate": 4.995391870108595e-05,
"loss": 0.6527,
"step": 595
},
{
"epoch": 0.06996268656716417,
"grad_norm": 0.7336499566418994,
"learning_rate": 4.9951101933324374e-05,
"loss": 0.6056,
"step": 600
},
{
"epoch": 0.07054570895522388,
"grad_norm": 0.9123524114339076,
"learning_rate": 4.994820170489629e-05,
"loss": 0.6351,
"step": 605
},
{
"epoch": 0.07112873134328358,
"grad_norm": 0.6937351959939653,
"learning_rate": 4.9945218026583147e-05,
"loss": 0.6415,
"step": 610
},
{
"epoch": 0.07171175373134328,
"grad_norm": 0.6701402337273787,
"learning_rate": 4.9942150909476576e-05,
"loss": 0.616,
"step": 615
},
{
"epoch": 0.07229477611940298,
"grad_norm": 0.6737823946741938,
"learning_rate": 4.9939000364978424e-05,
"loss": 0.5916,
"step": 620
},
{
"epoch": 0.07287779850746269,
"grad_norm": 0.6443035255679863,
"learning_rate": 4.993576640480064e-05,
"loss": 0.593,
"step": 625
},
{
"epoch": 0.07346082089552239,
"grad_norm": 0.7194715850493604,
"learning_rate": 4.9932449040965296e-05,
"loss": 0.6537,
"step": 630
},
{
"epoch": 0.07404384328358209,
"grad_norm": 0.6638929998798874,
"learning_rate": 4.992904828580449e-05,
"loss": 0.651,
"step": 635
},
{
"epoch": 0.07462686567164178,
"grad_norm": 0.6403792391822621,
"learning_rate": 4.992556415196035e-05,
"loss": 0.5995,
"step": 640
},
{
"epoch": 0.0752098880597015,
"grad_norm": 0.7646937059991467,
"learning_rate": 4.9921996652384915e-05,
"loss": 0.6254,
"step": 645
},
{
"epoch": 0.0757929104477612,
"grad_norm": 0.708046339189841,
"learning_rate": 4.991834580034017e-05,
"loss": 0.6285,
"step": 650
},
{
"epoch": 0.07637593283582089,
"grad_norm": 0.6787038237176444,
"learning_rate": 4.991461160939795e-05,
"loss": 0.6089,
"step": 655
},
{
"epoch": 0.07695895522388059,
"grad_norm": 0.637822387437937,
"learning_rate": 4.991079409343989e-05,
"loss": 0.6339,
"step": 660
},
{
"epoch": 0.0775419776119403,
"grad_norm": 0.7266823799603278,
"learning_rate": 4.990689326665738e-05,
"loss": 0.601,
"step": 665
},
{
"epoch": 0.078125,
"grad_norm": 0.6501949889104538,
"learning_rate": 4.9902909143551516e-05,
"loss": 0.6016,
"step": 670
},
{
"epoch": 0.0787080223880597,
"grad_norm": 0.6820009767011616,
"learning_rate": 4.989884173893305e-05,
"loss": 0.6007,
"step": 675
},
{
"epoch": 0.07929104477611941,
"grad_norm": 0.6877686038902945,
"learning_rate": 4.989469106792231e-05,
"loss": 0.6136,
"step": 680
},
{
"epoch": 0.07987406716417911,
"grad_norm": 0.6085151630088098,
"learning_rate": 4.9890457145949186e-05,
"loss": 0.5881,
"step": 685
},
{
"epoch": 0.0804570895522388,
"grad_norm": 0.6214607197858061,
"learning_rate": 4.988613998875304e-05,
"loss": 0.6053,
"step": 690
},
{
"epoch": 0.0810401119402985,
"grad_norm": 0.6744013331308584,
"learning_rate": 4.988173961238264e-05,
"loss": 0.5802,
"step": 695
},
{
"epoch": 0.08162313432835822,
"grad_norm": 0.6808032691033227,
"learning_rate": 4.987725603319615e-05,
"loss": 0.6094,
"step": 700
},
{
"epoch": 0.08220615671641791,
"grad_norm": 0.6979985310436119,
"learning_rate": 4.987268926786098e-05,
"loss": 0.6323,
"step": 705
},
{
"epoch": 0.08278917910447761,
"grad_norm": 0.6226469149492844,
"learning_rate": 4.986803933335385e-05,
"loss": 0.6459,
"step": 710
},
{
"epoch": 0.08337220149253731,
"grad_norm": 0.676363656496179,
"learning_rate": 4.9863306246960605e-05,
"loss": 0.5761,
"step": 715
},
{
"epoch": 0.08395522388059702,
"grad_norm": 0.6685016854342901,
"learning_rate": 4.9858490026276226e-05,
"loss": 0.5988,
"step": 720
},
{
"epoch": 0.08453824626865672,
"grad_norm": 0.6396794014235626,
"learning_rate": 4.9853590689204715e-05,
"loss": 0.5854,
"step": 725
},
{
"epoch": 0.08512126865671642,
"grad_norm": 0.720334288248887,
"learning_rate": 4.9848608253959096e-05,
"loss": 0.6207,
"step": 730
},
{
"epoch": 0.08570429104477612,
"grad_norm": 0.6079861012766015,
"learning_rate": 4.984354273906127e-05,
"loss": 0.5953,
"step": 735
},
{
"epoch": 0.08628731343283583,
"grad_norm": 0.6782045394701461,
"learning_rate": 4.9838394163341993e-05,
"loss": 0.63,
"step": 740
},
{
"epoch": 0.08687033582089553,
"grad_norm": 0.6465534111571817,
"learning_rate": 4.983316254594081e-05,
"loss": 0.6006,
"step": 745
},
{
"epoch": 0.08745335820895522,
"grad_norm": 0.6048787945996146,
"learning_rate": 4.9827847906305934e-05,
"loss": 0.5937,
"step": 750
},
{
"epoch": 0.08803638059701492,
"grad_norm": 0.6704335077914664,
"learning_rate": 4.982245026419424e-05,
"loss": 0.6433,
"step": 755
},
{
"epoch": 0.08861940298507463,
"grad_norm": 0.6420162328599116,
"learning_rate": 4.981696963967116e-05,
"loss": 0.5806,
"step": 760
},
{
"epoch": 0.08920242537313433,
"grad_norm": 0.589278560052739,
"learning_rate": 4.981140605311057e-05,
"loss": 0.6049,
"step": 765
},
{
"epoch": 0.08978544776119403,
"grad_norm": 0.6681567808799939,
"learning_rate": 4.98057595251948e-05,
"loss": 0.6369,
"step": 770
},
{
"epoch": 0.09036847014925373,
"grad_norm": 0.5822554638380099,
"learning_rate": 4.980003007691449e-05,
"loss": 0.633,
"step": 775
},
{
"epoch": 0.09095149253731344,
"grad_norm": 0.6014893132349064,
"learning_rate": 4.979421772956852e-05,
"loss": 0.637,
"step": 780
},
{
"epoch": 0.09153451492537314,
"grad_norm": 0.6190152455325075,
"learning_rate": 4.9788322504763954e-05,
"loss": 0.626,
"step": 785
},
{
"epoch": 0.09211753731343283,
"grad_norm": 0.6043190595480243,
"learning_rate": 4.978234442441596e-05,
"loss": 0.5958,
"step": 790
},
{
"epoch": 0.09270055970149253,
"grad_norm": 0.7067979054927527,
"learning_rate": 4.977628351074769e-05,
"loss": 0.6391,
"step": 795
},
{
"epoch": 0.09328358208955224,
"grad_norm": 1.0900391421932065,
"learning_rate": 4.977013978629025e-05,
"loss": 0.5869,
"step": 800
},
{
"epoch": 0.09386660447761194,
"grad_norm": 0.7289575339056653,
"learning_rate": 4.976391327388257e-05,
"loss": 0.5817,
"step": 805
},
{
"epoch": 0.09444962686567164,
"grad_norm": 0.6772024103975534,
"learning_rate": 4.9757603996671354e-05,
"loss": 0.644,
"step": 810
},
{
"epoch": 0.09503264925373134,
"grad_norm": 0.6273836766329598,
"learning_rate": 4.975121197811096e-05,
"loss": 0.6343,
"step": 815
},
{
"epoch": 0.09561567164179105,
"grad_norm": 0.6925933519325139,
"learning_rate": 4.974473724196338e-05,
"loss": 0.6527,
"step": 820
},
{
"epoch": 0.09619869402985075,
"grad_norm": 0.7918262789810132,
"learning_rate": 4.973817981229802e-05,
"loss": 0.6043,
"step": 825
},
{
"epoch": 0.09678171641791045,
"grad_norm": 0.6915704244723462,
"learning_rate": 4.9731539713491776e-05,
"loss": 0.6101,
"step": 830
},
{
"epoch": 0.09736473880597014,
"grad_norm": 0.7026540645380082,
"learning_rate": 4.972481697022883e-05,
"loss": 0.6321,
"step": 835
},
{
"epoch": 0.09794776119402986,
"grad_norm": 0.6870682595723567,
"learning_rate": 4.971801160750057e-05,
"loss": 0.6431,
"step": 840
},
{
"epoch": 0.09853078358208955,
"grad_norm": 0.6762758634999102,
"learning_rate": 4.971112365060555e-05,
"loss": 0.6212,
"step": 845
},
{
"epoch": 0.09911380597014925,
"grad_norm": 1.0041969348534898,
"learning_rate": 4.970415312514936e-05,
"loss": 0.5934,
"step": 850
},
{
"epoch": 0.09969682835820895,
"grad_norm": 0.6273960276780831,
"learning_rate": 4.969710005704449e-05,
"loss": 0.6499,
"step": 855
},
{
"epoch": 0.10027985074626866,
"grad_norm": 0.6488940069009905,
"learning_rate": 4.9689964472510345e-05,
"loss": 0.5949,
"step": 860
},
{
"epoch": 0.10086287313432836,
"grad_norm": 0.763231378289182,
"learning_rate": 4.968274639807304e-05,
"loss": 0.6146,
"step": 865
},
{
"epoch": 0.10144589552238806,
"grad_norm": 0.6658920295533095,
"learning_rate": 4.967544586056532e-05,
"loss": 0.6049,
"step": 870
},
{
"epoch": 0.10202891791044776,
"grad_norm": 0.7074390644632157,
"learning_rate": 4.966806288712654e-05,
"loss": 0.6175,
"step": 875
},
{
"epoch": 0.10261194029850747,
"grad_norm": 0.695343426907801,
"learning_rate": 4.966059750520246e-05,
"loss": 0.591,
"step": 880
},
{
"epoch": 0.10319496268656717,
"grad_norm": 0.6418320757008029,
"learning_rate": 4.965304974254521e-05,
"loss": 0.6176,
"step": 885
},
{
"epoch": 0.10377798507462686,
"grad_norm": 0.7727923536862525,
"learning_rate": 4.9645419627213155e-05,
"loss": 0.6134,
"step": 890
},
{
"epoch": 0.10436100746268656,
"grad_norm": 0.6029656422873543,
"learning_rate": 4.96377071875708e-05,
"loss": 0.5654,
"step": 895
},
{
"epoch": 0.10494402985074627,
"grad_norm": 0.6358118150755105,
"learning_rate": 4.9629912452288696e-05,
"loss": 0.6252,
"step": 900
},
{
"epoch": 0.10552705223880597,
"grad_norm": 0.61344386800639,
"learning_rate": 4.962203545034332e-05,
"loss": 0.6059,
"step": 905
},
{
"epoch": 0.10611007462686567,
"grad_norm": 0.6135409466759565,
"learning_rate": 4.961407621101697e-05,
"loss": 0.5614,
"step": 910
},
{
"epoch": 0.10669309701492537,
"grad_norm": 1.0821214504665195,
"learning_rate": 4.960603476389765e-05,
"loss": 0.6162,
"step": 915
},
{
"epoch": 0.10727611940298508,
"grad_norm": 0.5759584165021776,
"learning_rate": 4.959791113887898e-05,
"loss": 0.6055,
"step": 920
},
{
"epoch": 0.10785914179104478,
"grad_norm": 0.5799139948464339,
"learning_rate": 4.958970536616006e-05,
"loss": 0.6067,
"step": 925
},
{
"epoch": 0.10844216417910447,
"grad_norm": 0.6049902178680088,
"learning_rate": 4.9581417476245365e-05,
"loss": 0.5808,
"step": 930
},
{
"epoch": 0.10902518656716417,
"grad_norm": 0.6216812883236675,
"learning_rate": 4.957304749994465e-05,
"loss": 0.5644,
"step": 935
},
{
"epoch": 0.10960820895522388,
"grad_norm": 0.6264774945250813,
"learning_rate": 4.956459546837283e-05,
"loss": 0.5889,
"step": 940
},
{
"epoch": 0.11019123134328358,
"grad_norm": 0.7584079121183279,
"learning_rate": 4.955606141294982e-05,
"loss": 0.6662,
"step": 945
},
{
"epoch": 0.11077425373134328,
"grad_norm": 0.6021040722392922,
"learning_rate": 4.954744536540048e-05,
"loss": 0.6075,
"step": 950
},
{
"epoch": 0.11135727611940298,
"grad_norm": 0.6516825177813871,
"learning_rate": 4.953874735775448e-05,
"loss": 0.6163,
"step": 955
},
{
"epoch": 0.11194029850746269,
"grad_norm": 0.6483316850868462,
"learning_rate": 4.9529967422346137e-05,
"loss": 0.6388,
"step": 960
},
{
"epoch": 0.11252332089552239,
"grad_norm": 1.0296359523919179,
"learning_rate": 4.952110559181437e-05,
"loss": 0.6187,
"step": 965
},
{
"epoch": 0.11310634328358209,
"grad_norm": 0.5963874485076113,
"learning_rate": 4.95121618991025e-05,
"loss": 0.653,
"step": 970
},
{
"epoch": 0.11368936567164178,
"grad_norm": 0.5827112927403142,
"learning_rate": 4.950313637745819e-05,
"loss": 0.6013,
"step": 975
},
{
"epoch": 0.1142723880597015,
"grad_norm": 0.5978368012555301,
"learning_rate": 4.9494029060433304e-05,
"loss": 0.5854,
"step": 980
},
{
"epoch": 0.1148554104477612,
"grad_norm": 0.6061921826097367,
"learning_rate": 4.9484839981883755e-05,
"loss": 0.6042,
"step": 985
},
{
"epoch": 0.11543843283582089,
"grad_norm": 0.577591628346454,
"learning_rate": 4.9475569175969414e-05,
"loss": 0.6033,
"step": 990
},
{
"epoch": 0.11602145522388059,
"grad_norm": 0.578876456470585,
"learning_rate": 4.9466216677153945e-05,
"loss": 0.5762,
"step": 995
},
{
"epoch": 0.1166044776119403,
"grad_norm": 0.6111711530967577,
"learning_rate": 4.945678252020475e-05,
"loss": 0.6078,
"step": 1000
},
{
"epoch": 0.1171875,
"grad_norm": 0.5389571001747652,
"learning_rate": 4.9447266740192735e-05,
"loss": 0.586,
"step": 1005
},
{
"epoch": 0.1177705223880597,
"grad_norm": 0.5665361736437953,
"learning_rate": 4.943766937249226e-05,
"loss": 0.5896,
"step": 1010
},
{
"epoch": 0.11835354477611941,
"grad_norm": 0.6004663875659557,
"learning_rate": 4.942799045278099e-05,
"loss": 0.5931,
"step": 1015
},
{
"epoch": 0.11893656716417911,
"grad_norm": 0.5877689595746617,
"learning_rate": 4.941823001703974e-05,
"loss": 0.5985,
"step": 1020
},
{
"epoch": 0.1195195895522388,
"grad_norm": 0.5935807095087586,
"learning_rate": 4.940838810155237e-05,
"loss": 0.5991,
"step": 1025
},
{
"epoch": 0.1201026119402985,
"grad_norm": 0.6656506474951656,
"learning_rate": 4.939846474290563e-05,
"loss": 0.6148,
"step": 1030
},
{
"epoch": 0.12068563432835822,
"grad_norm": 0.6400226054588848,
"learning_rate": 4.9388459977989016e-05,
"loss": 0.6035,
"step": 1035
},
{
"epoch": 0.12126865671641791,
"grad_norm": 0.5682054111741724,
"learning_rate": 4.937837384399467e-05,
"loss": 0.5942,
"step": 1040
},
{
"epoch": 0.12185167910447761,
"grad_norm": 0.5722092103161287,
"learning_rate": 4.936820637841721e-05,
"loss": 0.6297,
"step": 1045
},
{
"epoch": 0.12243470149253731,
"grad_norm": 0.5744583767360888,
"learning_rate": 4.935795761905359e-05,
"loss": 0.5799,
"step": 1050
},
{
"epoch": 0.12301772388059702,
"grad_norm": 0.5955258224599197,
"learning_rate": 4.934762760400299e-05,
"loss": 0.5901,
"step": 1055
},
{
"epoch": 0.12360074626865672,
"grad_norm": 0.5941450778138045,
"learning_rate": 4.933721637166662e-05,
"loss": 0.5932,
"step": 1060
},
{
"epoch": 0.12418376865671642,
"grad_norm": 0.6984953623677035,
"learning_rate": 4.9326723960747655e-05,
"loss": 0.5928,
"step": 1065
},
{
"epoch": 0.12476679104477612,
"grad_norm": 0.5556738969991728,
"learning_rate": 4.931615041025101e-05,
"loss": 0.5449,
"step": 1070
},
{
"epoch": 0.12534981343283583,
"grad_norm": 0.6202875202064847,
"learning_rate": 4.9305495759483246e-05,
"loss": 0.6245,
"step": 1075
},
{
"epoch": 0.1259328358208955,
"grad_norm": 0.6360648582792375,
"learning_rate": 4.929476004805241e-05,
"loss": 0.622,
"step": 1080
},
{
"epoch": 0.12651585820895522,
"grad_norm": 0.548961800240671,
"learning_rate": 4.928394331586788e-05,
"loss": 0.5878,
"step": 1085
},
{
"epoch": 0.12709888059701493,
"grad_norm": 0.6534887664001585,
"learning_rate": 4.927304560314023e-05,
"loss": 0.5985,
"step": 1090
},
{
"epoch": 0.12768190298507462,
"grad_norm": 0.6051258985285789,
"learning_rate": 4.9262066950381074e-05,
"loss": 0.5815,
"step": 1095
},
{
"epoch": 0.12826492537313433,
"grad_norm": 0.5940919877769313,
"learning_rate": 4.925100739840293e-05,
"loss": 0.6056,
"step": 1100
},
{
"epoch": 0.12884794776119404,
"grad_norm": 0.5795416953387396,
"learning_rate": 4.923986698831902e-05,
"loss": 0.593,
"step": 1105
},
{
"epoch": 0.12943097014925373,
"grad_norm": 0.6060003300391443,
"learning_rate": 4.922864576154318e-05,
"loss": 0.6293,
"step": 1110
},
{
"epoch": 0.13001399253731344,
"grad_norm": 0.5926697361708111,
"learning_rate": 4.921734375978966e-05,
"loss": 0.5863,
"step": 1115
},
{
"epoch": 0.13059701492537312,
"grad_norm": 0.5666399578950052,
"learning_rate": 4.9205961025073005e-05,
"loss": 0.6093,
"step": 1120
},
{
"epoch": 0.13118003731343283,
"grad_norm": 0.6290080779373148,
"learning_rate": 4.919449759970787e-05,
"loss": 0.6149,
"step": 1125
},
{
"epoch": 0.13176305970149255,
"grad_norm": 0.6030059289618742,
"learning_rate": 4.9182953526308866e-05,
"loss": 0.5983,
"step": 1130
},
{
"epoch": 0.13234608208955223,
"grad_norm": 0.6967548151886129,
"learning_rate": 4.9171328847790416e-05,
"loss": 0.5979,
"step": 1135
},
{
"epoch": 0.13292910447761194,
"grad_norm": 0.6568770405720038,
"learning_rate": 4.9159623607366587e-05,
"loss": 0.5857,
"step": 1140
},
{
"epoch": 0.13351212686567165,
"grad_norm": 0.5840444295366739,
"learning_rate": 4.914783784855093e-05,
"loss": 0.6438,
"step": 1145
},
{
"epoch": 0.13409514925373134,
"grad_norm": 0.7004511399791347,
"learning_rate": 4.913597161515633e-05,
"loss": 0.6184,
"step": 1150
},
{
"epoch": 0.13467817164179105,
"grad_norm": 0.550373143569304,
"learning_rate": 4.91240249512948e-05,
"loss": 0.567,
"step": 1155
},
{
"epoch": 0.13526119402985073,
"grad_norm": 0.5313433759195957,
"learning_rate": 4.9111997901377373e-05,
"loss": 0.5855,
"step": 1160
},
{
"epoch": 0.13584421641791045,
"grad_norm": 0.5655705667845206,
"learning_rate": 4.9099890510113924e-05,
"loss": 0.5661,
"step": 1165
},
{
"epoch": 0.13642723880597016,
"grad_norm": 0.6195584641586247,
"learning_rate": 4.908770282251296e-05,
"loss": 0.5905,
"step": 1170
},
{
"epoch": 0.13701026119402984,
"grad_norm": 0.5583698659457814,
"learning_rate": 4.9075434883881504e-05,
"loss": 0.6002,
"step": 1175
},
{
"epoch": 0.13759328358208955,
"grad_norm": 0.6389070121403235,
"learning_rate": 4.906308673982491e-05,
"loss": 0.5607,
"step": 1180
},
{
"epoch": 0.13817630597014927,
"grad_norm": 0.59774495236772,
"learning_rate": 4.905065843624668e-05,
"loss": 0.5678,
"step": 1185
},
{
"epoch": 0.13875932835820895,
"grad_norm": 0.6173804758323792,
"learning_rate": 4.903815001934832e-05,
"loss": 0.6002,
"step": 1190
},
{
"epoch": 0.13934235074626866,
"grad_norm": 0.5469869287414887,
"learning_rate": 4.9025561535629125e-05,
"loss": 0.5977,
"step": 1195
},
{
"epoch": 0.13992537313432835,
"grad_norm": 0.5624438700425504,
"learning_rate": 4.9012893031886075e-05,
"loss": 0.582,
"step": 1200
},
{
"epoch": 0.14050839552238806,
"grad_norm": 0.5985379982120503,
"learning_rate": 4.9000144555213575e-05,
"loss": 0.5825,
"step": 1205
},
{
"epoch": 0.14109141791044777,
"grad_norm": 0.5800377004502044,
"learning_rate": 4.898731615300336e-05,
"loss": 0.5706,
"step": 1210
},
{
"epoch": 0.14167444029850745,
"grad_norm": 0.5901666286652832,
"learning_rate": 4.8974407872944263e-05,
"loss": 0.5937,
"step": 1215
},
{
"epoch": 0.14225746268656717,
"grad_norm": 0.644009126842579,
"learning_rate": 4.8961419763022065e-05,
"loss": 0.5612,
"step": 1220
},
{
"epoch": 0.14284048507462688,
"grad_norm": 0.5922807627602008,
"learning_rate": 4.894835187151931e-05,
"loss": 0.6067,
"step": 1225
},
{
"epoch": 0.14342350746268656,
"grad_norm": 0.5722199265482402,
"learning_rate": 4.893520424701513e-05,
"loss": 0.6082,
"step": 1230
},
{
"epoch": 0.14400652985074627,
"grad_norm": 0.6126570623385963,
"learning_rate": 4.892197693838504e-05,
"loss": 0.5964,
"step": 1235
},
{
"epoch": 0.14458955223880596,
"grad_norm": 0.5973655677327272,
"learning_rate": 4.890866999480082e-05,
"loss": 0.5918,
"step": 1240
},
{
"epoch": 0.14517257462686567,
"grad_norm": 0.54393899856667,
"learning_rate": 4.889528346573023e-05,
"loss": 0.5827,
"step": 1245
},
{
"epoch": 0.14575559701492538,
"grad_norm": 0.5508599028743483,
"learning_rate": 4.888181740093693e-05,
"loss": 0.5966,
"step": 1250
},
{
"epoch": 0.14633861940298507,
"grad_norm": 0.5748810947596694,
"learning_rate": 4.886827185048023e-05,
"loss": 0.6069,
"step": 1255
},
{
"epoch": 0.14692164179104478,
"grad_norm": 0.5843720987519714,
"learning_rate": 4.8854646864714906e-05,
"loss": 0.626,
"step": 1260
},
{
"epoch": 0.1475046641791045,
"grad_norm": 0.5574624014831214,
"learning_rate": 4.884094249429109e-05,
"loss": 0.5513,
"step": 1265
},
{
"epoch": 0.14808768656716417,
"grad_norm": 0.6341906909347658,
"learning_rate": 4.882715879015396e-05,
"loss": 0.597,
"step": 1270
},
{
"epoch": 0.14867070895522388,
"grad_norm": 0.575813430423846,
"learning_rate": 4.881329580354363e-05,
"loss": 0.6081,
"step": 1275
},
{
"epoch": 0.14925373134328357,
"grad_norm": 0.6445668774873804,
"learning_rate": 4.8799353585994954e-05,
"loss": 0.6087,
"step": 1280
},
{
"epoch": 0.14983675373134328,
"grad_norm": 0.5859378252774211,
"learning_rate": 4.8785332189337325e-05,
"loss": 0.5602,
"step": 1285
},
{
"epoch": 0.150419776119403,
"grad_norm": 0.6268377383651598,
"learning_rate": 4.877123166569445e-05,
"loss": 0.621,
"step": 1290
},
{
"epoch": 0.15100279850746268,
"grad_norm": 0.539111443847871,
"learning_rate": 4.8757052067484216e-05,
"loss": 0.5778,
"step": 1295
},
{
"epoch": 0.1515858208955224,
"grad_norm": 0.6230886477061637,
"learning_rate": 4.874279344741846e-05,
"loss": 0.6052,
"step": 1300
},
{
"epoch": 0.1521688432835821,
"grad_norm": 0.5804714929817124,
"learning_rate": 4.872845585850277e-05,
"loss": 0.5704,
"step": 1305
},
{
"epoch": 0.15275186567164178,
"grad_norm": 0.5644557466436395,
"learning_rate": 4.87140393540363e-05,
"loss": 0.6135,
"step": 1310
},
{
"epoch": 0.1533348880597015,
"grad_norm": 0.5979238448583827,
"learning_rate": 4.869954398761156e-05,
"loss": 0.5891,
"step": 1315
},
{
"epoch": 0.15391791044776118,
"grad_norm": 0.5684870674237956,
"learning_rate": 4.868496981311424e-05,
"loss": 0.6057,
"step": 1320
},
{
"epoch": 0.1545009328358209,
"grad_norm": 0.6084433820062218,
"learning_rate": 4.8670316884722984e-05,
"loss": 0.6183,
"step": 1325
},
{
"epoch": 0.1550839552238806,
"grad_norm": 0.5168917985718466,
"learning_rate": 4.86555852569092e-05,
"loss": 0.558,
"step": 1330
},
{
"epoch": 0.1556669776119403,
"grad_norm": 0.6169882009143579,
"learning_rate": 4.864077498443687e-05,
"loss": 0.5894,
"step": 1335
},
{
"epoch": 0.15625,
"grad_norm": 0.5573154647203309,
"learning_rate": 4.8625886122362305e-05,
"loss": 0.6047,
"step": 1340
},
{
"epoch": 0.1568330223880597,
"grad_norm": 0.5430765223123376,
"learning_rate": 4.861091872603399e-05,
"loss": 0.5936,
"step": 1345
},
{
"epoch": 0.1574160447761194,
"grad_norm": 0.5952512138917507,
"learning_rate": 4.859587285109235e-05,
"loss": 0.6323,
"step": 1350
},
{
"epoch": 0.1579990671641791,
"grad_norm": 0.6257733920419736,
"learning_rate": 4.8580748553469554e-05,
"loss": 0.621,
"step": 1355
},
{
"epoch": 0.15858208955223882,
"grad_norm": 0.6484544777362733,
"learning_rate": 4.8565545889389295e-05,
"loss": 0.6225,
"step": 1360
},
{
"epoch": 0.1591651119402985,
"grad_norm": 0.6001757890053516,
"learning_rate": 4.85502649153666e-05,
"loss": 0.7485,
"step": 1365
},
{
"epoch": 0.15974813432835822,
"grad_norm": 0.661835508508043,
"learning_rate": 4.853490568820759e-05,
"loss": 0.8953,
"step": 1370
},
{
"epoch": 0.1603311567164179,
"grad_norm": 0.5424000700147391,
"learning_rate": 4.851946826500932e-05,
"loss": 0.6088,
"step": 1375
},
{
"epoch": 0.1609141791044776,
"grad_norm": 0.5642506132746299,
"learning_rate": 4.8503952703159485e-05,
"loss": 0.5788,
"step": 1380
},
{
"epoch": 0.16149720149253732,
"grad_norm": 1.5770888453223382,
"learning_rate": 4.8488359060336314e-05,
"loss": 0.6464,
"step": 1385
},
{
"epoch": 0.162080223880597,
"grad_norm": 0.5798567582108715,
"learning_rate": 4.847268739450825e-05,
"loss": 0.5962,
"step": 1390
},
{
"epoch": 0.16266324626865672,
"grad_norm": 0.6247134338348835,
"learning_rate": 4.84569377639338e-05,
"loss": 0.5762,
"step": 1395
},
{
"epoch": 0.16324626865671643,
"grad_norm": 0.6403693981471119,
"learning_rate": 4.84411102271613e-05,
"loss": 0.6033,
"step": 1400
},
{
"epoch": 0.16382929104477612,
"grad_norm": 0.6328461284086258,
"learning_rate": 4.842520484302871e-05,
"loss": 0.6124,
"step": 1405
},
{
"epoch": 0.16441231343283583,
"grad_norm": 0.5690467644511553,
"learning_rate": 4.840922167066335e-05,
"loss": 0.6058,
"step": 1410
},
{
"epoch": 0.1649953358208955,
"grad_norm": 0.5991271256192877,
"learning_rate": 4.8393160769481755e-05,
"loss": 0.5759,
"step": 1415
},
{
"epoch": 0.16557835820895522,
"grad_norm": 0.5518688454925015,
"learning_rate": 4.8377022199189374e-05,
"loss": 0.5758,
"step": 1420
},
{
"epoch": 0.16616138059701493,
"grad_norm": 0.8358279547212623,
"learning_rate": 4.836080601978043e-05,
"loss": 0.614,
"step": 1425
},
{
"epoch": 0.16674440298507462,
"grad_norm": 0.5736501078455153,
"learning_rate": 4.83445122915376e-05,
"loss": 0.6268,
"step": 1430
},
{
"epoch": 0.16732742537313433,
"grad_norm": 0.501139191324822,
"learning_rate": 4.832814107503188e-05,
"loss": 0.5522,
"step": 1435
},
{
"epoch": 0.16791044776119404,
"grad_norm": 0.5414270989894118,
"learning_rate": 4.8311692431122326e-05,
"loss": 0.5985,
"step": 1440
},
{
"epoch": 0.16849347014925373,
"grad_norm": 0.5503417116648404,
"learning_rate": 4.82951664209558e-05,
"loss": 0.5883,
"step": 1445
},
{
"epoch": 0.16907649253731344,
"grad_norm": 0.5828254158920338,
"learning_rate": 4.82785631059668e-05,
"loss": 0.5973,
"step": 1450
},
{
"epoch": 0.16965951492537312,
"grad_norm": 0.5750003990997431,
"learning_rate": 4.826188254787717e-05,
"loss": 0.5929,
"step": 1455
},
{
"epoch": 0.17024253731343283,
"grad_norm": 0.5931754407423996,
"learning_rate": 4.824512480869593e-05,
"loss": 0.584,
"step": 1460
},
{
"epoch": 0.17082555970149255,
"grad_norm": 0.5457054499115934,
"learning_rate": 4.822828995071899e-05,
"loss": 0.5711,
"step": 1465
},
{
"epoch": 0.17140858208955223,
"grad_norm": 0.5229569450197525,
"learning_rate": 4.821137803652896e-05,
"loss": 0.5459,
"step": 1470
},
{
"epoch": 0.17199160447761194,
"grad_norm": 0.6027439958371525,
"learning_rate": 4.819438912899489e-05,
"loss": 0.5814,
"step": 1475
},
{
"epoch": 0.17257462686567165,
"grad_norm": 0.5702069062273913,
"learning_rate": 4.8177323291272066e-05,
"loss": 0.6299,
"step": 1480
},
{
"epoch": 0.17315764925373134,
"grad_norm": 0.5780484151492952,
"learning_rate": 4.8160180586801744e-05,
"loss": 0.5777,
"step": 1485
},
{
"epoch": 0.17374067164179105,
"grad_norm": 0.575917162342149,
"learning_rate": 4.814296107931093e-05,
"loss": 0.5547,
"step": 1490
},
{
"epoch": 0.17432369402985073,
"grad_norm": 0.5192602462313004,
"learning_rate": 4.812566483281216e-05,
"loss": 0.5669,
"step": 1495
},
{
"epoch": 0.17490671641791045,
"grad_norm": 0.5345650171129616,
"learning_rate": 4.81082919116032e-05,
"loss": 0.5917,
"step": 1500
},
{
"epoch": 0.17548973880597016,
"grad_norm": 0.5475736070312467,
"learning_rate": 4.809084238026689e-05,
"loss": 0.602,
"step": 1505
},
{
"epoch": 0.17607276119402984,
"grad_norm": 0.5786621619994619,
"learning_rate": 4.8073316303670835e-05,
"loss": 0.5861,
"step": 1510
},
{
"epoch": 0.17665578358208955,
"grad_norm": 0.5448850237456587,
"learning_rate": 4.8055713746967216e-05,
"loss": 0.5857,
"step": 1515
},
{
"epoch": 0.17723880597014927,
"grad_norm": 0.5779873112523268,
"learning_rate": 4.803803477559252e-05,
"loss": 0.5665,
"step": 1520
},
{
"epoch": 0.17782182835820895,
"grad_norm": 0.5652601574196091,
"learning_rate": 4.8020279455267274e-05,
"loss": 0.5794,
"step": 1525
},
{
"epoch": 0.17840485074626866,
"grad_norm": 0.5437626429311837,
"learning_rate": 4.800244785199588e-05,
"loss": 0.547,
"step": 1530
},
{
"epoch": 0.17898787313432835,
"grad_norm": 0.5657663832577134,
"learning_rate": 4.7984540032066266e-05,
"loss": 0.5645,
"step": 1535
},
{
"epoch": 0.17957089552238806,
"grad_norm": 0.5396084565585528,
"learning_rate": 4.796655606204971e-05,
"loss": 0.5917,
"step": 1540
},
{
"epoch": 0.18015391791044777,
"grad_norm": 0.5659614973871961,
"learning_rate": 4.794849600880059e-05,
"loss": 0.5888,
"step": 1545
},
{
"epoch": 0.18073694029850745,
"grad_norm": 0.610344463972923,
"learning_rate": 4.793035993945609e-05,
"loss": 0.5973,
"step": 1550
},
{
"epoch": 0.18131996268656717,
"grad_norm": 0.5315364790581317,
"learning_rate": 4.7912147921436e-05,
"loss": 0.5793,
"step": 1555
},
{
"epoch": 0.18190298507462688,
"grad_norm": 0.5215816759572497,
"learning_rate": 4.789386002244244e-05,
"loss": 0.5773,
"step": 1560
},
{
"epoch": 0.18248600746268656,
"grad_norm": 0.48268753752530635,
"learning_rate": 4.7875496310459607e-05,
"loss": 0.5439,
"step": 1565
},
{
"epoch": 0.18306902985074627,
"grad_norm": 0.5928345158273391,
"learning_rate": 4.7857056853753536e-05,
"loss": 0.5946,
"step": 1570
},
{
"epoch": 0.18365205223880596,
"grad_norm": 0.5217322463974674,
"learning_rate": 4.783854172087183e-05,
"loss": 0.5633,
"step": 1575
},
{
"epoch": 0.18423507462686567,
"grad_norm": 0.5361438235075061,
"learning_rate": 4.781995098064343e-05,
"loss": 0.5616,
"step": 1580
},
{
"epoch": 0.18481809701492538,
"grad_norm": 0.5319409918375809,
"learning_rate": 4.780128470217833e-05,
"loss": 0.5959,
"step": 1585
},
{
"epoch": 0.18540111940298507,
"grad_norm": 0.5656220383824416,
"learning_rate": 4.778254295486732e-05,
"loss": 0.5941,
"step": 1590
},
{
"epoch": 0.18598414179104478,
"grad_norm": 0.664107972745298,
"learning_rate": 4.7763725808381777e-05,
"loss": 0.5932,
"step": 1595
},
{
"epoch": 0.1865671641791045,
"grad_norm": 0.5594876020852717,
"learning_rate": 4.7744833332673336e-05,
"loss": 0.5754,
"step": 1600
},
{
"epoch": 0.18715018656716417,
"grad_norm": 0.5748188546122887,
"learning_rate": 4.7725865597973684e-05,
"loss": 0.5929,
"step": 1605
},
{
"epoch": 0.18773320895522388,
"grad_norm": 0.5608090588962066,
"learning_rate": 4.770682267479427e-05,
"loss": 0.6,
"step": 1610
},
{
"epoch": 0.18831623134328357,
"grad_norm": 0.5614561936493092,
"learning_rate": 4.7687704633926056e-05,
"loss": 0.5885,
"step": 1615
},
{
"epoch": 0.18889925373134328,
"grad_norm": 0.5020575112535327,
"learning_rate": 4.766851154643924e-05,
"loss": 0.5766,
"step": 1620
},
{
"epoch": 0.189482276119403,
"grad_norm": 0.5350491170633971,
"learning_rate": 4.7649243483683015e-05,
"loss": 0.6016,
"step": 1625
},
{
"epoch": 0.19006529850746268,
"grad_norm": 0.5723463466212365,
"learning_rate": 4.762990051728529e-05,
"loss": 0.5938,
"step": 1630
},
{
"epoch": 0.1906483208955224,
"grad_norm": 0.528038155312228,
"learning_rate": 4.7610482719152404e-05,
"loss": 0.5919,
"step": 1635
},
{
"epoch": 0.1912313432835821,
"grad_norm": 0.5195292947087375,
"learning_rate": 4.7590990161468906e-05,
"loss": 0.587,
"step": 1640
},
{
"epoch": 0.19181436567164178,
"grad_norm": 0.5309411178378524,
"learning_rate": 4.757142291669724e-05,
"loss": 0.5584,
"step": 1645
},
{
"epoch": 0.1923973880597015,
"grad_norm": 0.5235366705843203,
"learning_rate": 4.755178105757751e-05,
"loss": 0.5332,
"step": 1650
},
{
"epoch": 0.19298041044776118,
"grad_norm": 0.5662388478996405,
"learning_rate": 4.753206465712717e-05,
"loss": 0.6003,
"step": 1655
},
{
"epoch": 0.1935634328358209,
"grad_norm": 0.541515776380167,
"learning_rate": 4.751227378864081e-05,
"loss": 0.6167,
"step": 1660
},
{
"epoch": 0.1941464552238806,
"grad_norm": 0.5451263614433529,
"learning_rate": 4.749240852568981e-05,
"loss": 0.5795,
"step": 1665
},
{
"epoch": 0.1947294776119403,
"grad_norm": 0.5539737195153389,
"learning_rate": 4.747246894212216e-05,
"loss": 0.6156,
"step": 1670
},
{
"epoch": 0.1953125,
"grad_norm": 0.4838135257147904,
"learning_rate": 4.7452455112062076e-05,
"loss": 0.586,
"step": 1675
},
{
"epoch": 0.1958955223880597,
"grad_norm": 0.6206912586835625,
"learning_rate": 4.743236710990982e-05,
"loss": 0.5835,
"step": 1680
},
{
"epoch": 0.1964785447761194,
"grad_norm": 0.5575344781989854,
"learning_rate": 4.7412205010341385e-05,
"loss": 0.5615,
"step": 1685
},
{
"epoch": 0.1970615671641791,
"grad_norm": 0.5465760234872273,
"learning_rate": 4.739196888830818e-05,
"loss": 0.5614,
"step": 1690
},
{
"epoch": 0.19764458955223882,
"grad_norm": 0.6116107229280592,
"learning_rate": 4.737165881903683e-05,
"loss": 0.5777,
"step": 1695
},
{
"epoch": 0.1982276119402985,
"grad_norm": 0.526352835776055,
"learning_rate": 4.735127487802882e-05,
"loss": 0.5499,
"step": 1700
},
{
"epoch": 0.19881063432835822,
"grad_norm": 0.5790768202270913,
"learning_rate": 4.7330817141060284e-05,
"loss": 0.6062,
"step": 1705
},
{
"epoch": 0.1993936567164179,
"grad_norm": 0.5786125838532169,
"learning_rate": 4.731028568418167e-05,
"loss": 0.5853,
"step": 1710
},
{
"epoch": 0.1999766791044776,
"grad_norm": 0.5220981929969758,
"learning_rate": 4.728968058371746e-05,
"loss": 0.5917,
"step": 1715
},
{
"epoch": 0.20055970149253732,
"grad_norm": 0.5962771524705726,
"learning_rate": 4.726900191626592e-05,
"loss": 0.615,
"step": 1720
},
{
"epoch": 0.201142723880597,
"grad_norm": 0.614179304629615,
"learning_rate": 4.724824975869881e-05,
"loss": 0.6088,
"step": 1725
},
{
"epoch": 0.20172574626865672,
"grad_norm": 0.5339809717905718,
"learning_rate": 4.722742418816106e-05,
"loss": 0.5856,
"step": 1730
},
{
"epoch": 0.20230876865671643,
"grad_norm": 0.5775559838523747,
"learning_rate": 4.7206525282070514e-05,
"loss": 0.5525,
"step": 1735
},
{
"epoch": 0.20289179104477612,
"grad_norm": 0.5787127160424809,
"learning_rate": 4.718555311811764e-05,
"loss": 0.5889,
"step": 1740
},
{
"epoch": 0.20347481343283583,
"grad_norm": 0.5507748006986873,
"learning_rate": 4.716450777426525e-05,
"loss": 0.5811,
"step": 1745
},
{
"epoch": 0.2040578358208955,
"grad_norm": 0.5386322774377927,
"learning_rate": 4.7143389328748174e-05,
"loss": 0.5681,
"step": 1750
},
{
"epoch": 0.20464085820895522,
"grad_norm": 0.5253847435139612,
"learning_rate": 4.712219786007302e-05,
"loss": 0.5352,
"step": 1755
},
{
"epoch": 0.20522388059701493,
"grad_norm": 0.5229542272956548,
"learning_rate": 4.710093344701782e-05,
"loss": 0.5878,
"step": 1760
},
{
"epoch": 0.20580690298507462,
"grad_norm": 0.5347624824538455,
"learning_rate": 4.707959616863181e-05,
"loss": 0.5622,
"step": 1765
},
{
"epoch": 0.20638992537313433,
"grad_norm": 0.5561607185138324,
"learning_rate": 4.7058186104235086e-05,
"loss": 0.5797,
"step": 1770
},
{
"epoch": 0.20697294776119404,
"grad_norm": 0.4892165988190017,
"learning_rate": 4.70367033334183e-05,
"loss": 0.5403,
"step": 1775
},
{
"epoch": 0.20755597014925373,
"grad_norm": 0.5510594553982379,
"learning_rate": 4.701514793604242e-05,
"loss": 0.5559,
"step": 1780
},
{
"epoch": 0.20813899253731344,
"grad_norm": 0.5473696320978395,
"learning_rate": 4.699351999223838e-05,
"loss": 0.5753,
"step": 1785
},
{
"epoch": 0.20872201492537312,
"grad_norm": 0.4873353162223706,
"learning_rate": 4.697181958240679e-05,
"loss": 0.5492,
"step": 1790
},
{
"epoch": 0.20930503731343283,
"grad_norm": 0.5064007191967134,
"learning_rate": 4.695004678721768e-05,
"loss": 0.5858,
"step": 1795
},
{
"epoch": 0.20988805970149255,
"grad_norm": 0.5667430033926578,
"learning_rate": 4.692820168761014e-05,
"loss": 0.6062,
"step": 1800
},
{
"epoch": 0.21047108208955223,
"grad_norm": 0.5212510033253821,
"learning_rate": 4.690628436479206e-05,
"loss": 0.5598,
"step": 1805
},
{
"epoch": 0.21105410447761194,
"grad_norm": 0.5588622002334643,
"learning_rate": 4.688429490023982e-05,
"loss": 0.5763,
"step": 1810
},
{
"epoch": 0.21163712686567165,
"grad_norm": 0.6512372849264118,
"learning_rate": 4.6862233375697964e-05,
"loss": 0.5808,
"step": 1815
},
{
"epoch": 0.21222014925373134,
"grad_norm": 0.600056817158455,
"learning_rate": 4.684009987317894e-05,
"loss": 0.5929,
"step": 1820
},
{
"epoch": 0.21280317164179105,
"grad_norm": 0.5525047925178632,
"learning_rate": 4.6817894474962756e-05,
"loss": 0.5598,
"step": 1825
},
{
"epoch": 0.21338619402985073,
"grad_norm": 0.49741821795163615,
"learning_rate": 4.679561726359668e-05,
"loss": 0.5689,
"step": 1830
},
{
"epoch": 0.21396921641791045,
"grad_norm": 0.5369544968462594,
"learning_rate": 4.677326832189496e-05,
"loss": 0.5846,
"step": 1835
},
{
"epoch": 0.21455223880597016,
"grad_norm": 0.578124984252202,
"learning_rate": 4.675084773293848e-05,
"loss": 0.6068,
"step": 1840
},
{
"epoch": 0.21513526119402984,
"grad_norm": 0.5489389574421355,
"learning_rate": 4.6728355580074476e-05,
"loss": 0.5799,
"step": 1845
},
{
"epoch": 0.21571828358208955,
"grad_norm": 0.5492579167575462,
"learning_rate": 4.6705791946916236e-05,
"loss": 0.5969,
"step": 1850
},
{
"epoch": 0.21630130597014927,
"grad_norm": 0.4976854096096666,
"learning_rate": 4.6683156917342726e-05,
"loss": 0.6034,
"step": 1855
},
{
"epoch": 0.21688432835820895,
"grad_norm": 0.600940601658767,
"learning_rate": 4.666045057549838e-05,
"loss": 0.5946,
"step": 1860
},
{
"epoch": 0.21746735074626866,
"grad_norm": 0.5506775932621145,
"learning_rate": 4.663767300579268e-05,
"loss": 0.5847,
"step": 1865
},
{
"epoch": 0.21805037313432835,
"grad_norm": 0.5271315191050433,
"learning_rate": 4.661482429289994e-05,
"loss": 0.5662,
"step": 1870
},
{
"epoch": 0.21863339552238806,
"grad_norm": 0.5806047125500812,
"learning_rate": 4.659190452175891e-05,
"loss": 0.5717,
"step": 1875
},
{
"epoch": 0.21921641791044777,
"grad_norm": 0.5548551191672546,
"learning_rate": 4.65689137775725e-05,
"loss": 0.5918,
"step": 1880
},
{
"epoch": 0.21979944029850745,
"grad_norm": 0.5224077396834346,
"learning_rate": 4.654585214580749e-05,
"loss": 0.5764,
"step": 1885
},
{
"epoch": 0.22038246268656717,
"grad_norm": 0.595212892553765,
"learning_rate": 4.652271971219412e-05,
"loss": 0.5718,
"step": 1890
},
{
"epoch": 0.22096548507462688,
"grad_norm": 0.5205351842902096,
"learning_rate": 4.6499516562725906e-05,
"loss": 0.5843,
"step": 1895
},
{
"epoch": 0.22154850746268656,
"grad_norm": 0.5193771435539847,
"learning_rate": 4.647624278365917e-05,
"loss": 0.6024,
"step": 1900
},
{
"epoch": 0.22213152985074627,
"grad_norm": 0.5762529929483823,
"learning_rate": 4.6452898461512866e-05,
"loss": 0.5841,
"step": 1905
},
{
"epoch": 0.22271455223880596,
"grad_norm": 0.5613099807801801,
"learning_rate": 4.642948368306814e-05,
"loss": 0.5909,
"step": 1910
},
{
"epoch": 0.22329757462686567,
"grad_norm": 0.5857098538355904,
"learning_rate": 4.640599853536806e-05,
"loss": 0.5986,
"step": 1915
},
{
"epoch": 0.22388059701492538,
"grad_norm": 0.5346551057187519,
"learning_rate": 4.6382443105717324e-05,
"loss": 0.5651,
"step": 1920
},
{
"epoch": 0.22446361940298507,
"grad_norm": 0.6133229780421945,
"learning_rate": 4.635881748168184e-05,
"loss": 0.6051,
"step": 1925
},
{
"epoch": 0.22504664179104478,
"grad_norm": 0.564856310131765,
"learning_rate": 4.633512175108851e-05,
"loss": 0.5645,
"step": 1930
},
{
"epoch": 0.2256296641791045,
"grad_norm": 0.6086776045173963,
"learning_rate": 4.631135600202485e-05,
"loss": 0.611,
"step": 1935
},
{
"epoch": 0.22621268656716417,
"grad_norm": 0.5053374707380298,
"learning_rate": 4.628752032283862e-05,
"loss": 0.6359,
"step": 1940
},
{
"epoch": 0.22679570895522388,
"grad_norm": 0.5177309796242289,
"learning_rate": 4.626361480213759e-05,
"loss": 0.5161,
"step": 1945
},
{
"epoch": 0.22737873134328357,
"grad_norm": 0.5655297929571091,
"learning_rate": 4.623963952878914e-05,
"loss": 0.5215,
"step": 1950
},
{
"epoch": 0.22796175373134328,
"grad_norm": 0.5007101086568317,
"learning_rate": 4.621559459191996e-05,
"loss": 0.5672,
"step": 1955
},
{
"epoch": 0.228544776119403,
"grad_norm": 0.5604337094818254,
"learning_rate": 4.619148008091569e-05,
"loss": 0.5495,
"step": 1960
},
{
"epoch": 0.22912779850746268,
"grad_norm": 0.5600231646159735,
"learning_rate": 4.616729608542064e-05,
"loss": 0.5914,
"step": 1965
},
{
"epoch": 0.2297108208955224,
"grad_norm": 0.5261560439698072,
"learning_rate": 4.61430426953374e-05,
"loss": 0.5761,
"step": 1970
},
{
"epoch": 0.2302938432835821,
"grad_norm": 0.5177160101117486,
"learning_rate": 4.611872000082654e-05,
"loss": 0.6105,
"step": 1975
},
{
"epoch": 0.23087686567164178,
"grad_norm": 0.4991176586608243,
"learning_rate": 4.609432809230627e-05,
"loss": 0.5576,
"step": 1980
},
{
"epoch": 0.2314598880597015,
"grad_norm": 0.49011496350766276,
"learning_rate": 4.606986706045207e-05,
"loss": 0.5217,
"step": 1985
},
{
"epoch": 0.23204291044776118,
"grad_norm": 0.49863205095192953,
"learning_rate": 4.604533699619643e-05,
"loss": 0.5606,
"step": 1990
},
{
"epoch": 0.2326259328358209,
"grad_norm": 0.5202522274733646,
"learning_rate": 4.602073799072841e-05,
"loss": 0.582,
"step": 1995
},
{
"epoch": 0.2332089552238806,
"grad_norm": 0.5778738097193321,
"learning_rate": 4.5996070135493426e-05,
"loss": 0.5608,
"step": 2000
},
{
"epoch": 0.2337919776119403,
"grad_norm": 0.5383730742466348,
"learning_rate": 4.597133352219275e-05,
"loss": 0.5833,
"step": 2005
},
{
"epoch": 0.234375,
"grad_norm": 0.504768235114523,
"learning_rate": 4.594652824278333e-05,
"loss": 0.5428,
"step": 2010
},
{
"epoch": 0.2349580223880597,
"grad_norm": 0.5424141377382832,
"learning_rate": 4.592165438947734e-05,
"loss": 0.5234,
"step": 2015
},
{
"epoch": 0.2355410447761194,
"grad_norm": 0.560600079358499,
"learning_rate": 4.589671205474189e-05,
"loss": 0.5949,
"step": 2020
},
{
"epoch": 0.2361240671641791,
"grad_norm": 0.5710842213852398,
"learning_rate": 4.587170133129867e-05,
"loss": 0.5675,
"step": 2025
},
{
"epoch": 0.23670708955223882,
"grad_norm": 0.6954740937313674,
"learning_rate": 4.5846622312123566e-05,
"loss": 0.5593,
"step": 2030
},
{
"epoch": 0.2372901119402985,
"grad_norm": 0.4936861843329207,
"learning_rate": 4.582147509044639e-05,
"loss": 0.544,
"step": 2035
},
{
"epoch": 0.23787313432835822,
"grad_norm": 0.5990080756647657,
"learning_rate": 4.579625975975047e-05,
"loss": 0.6093,
"step": 2040
},
{
"epoch": 0.2384561567164179,
"grad_norm": 0.46710498554704555,
"learning_rate": 4.577097641377234e-05,
"loss": 0.5421,
"step": 2045
},
{
"epoch": 0.2390391791044776,
"grad_norm": 0.6220614730173019,
"learning_rate": 4.574562514650137e-05,
"loss": 0.5959,
"step": 2050
},
{
"epoch": 0.23962220149253732,
"grad_norm": 0.5736687871170686,
"learning_rate": 4.572020605217941e-05,
"loss": 0.5705,
"step": 2055
},
{
"epoch": 0.240205223880597,
"grad_norm": 0.5700891932767567,
"learning_rate": 4.569471922530048e-05,
"loss": 0.5812,
"step": 2060
},
{
"epoch": 0.24078824626865672,
"grad_norm": 0.5076933159103915,
"learning_rate": 4.566916476061036e-05,
"loss": 0.565,
"step": 2065
},
{
"epoch": 0.24137126865671643,
"grad_norm": 0.5485662358157128,
"learning_rate": 4.56435427531063e-05,
"loss": 0.576,
"step": 2070
},
{
"epoch": 0.24195429104477612,
"grad_norm": 0.5699225728909032,
"learning_rate": 4.5617853298036634e-05,
"loss": 0.5984,
"step": 2075
},
{
"epoch": 0.24253731343283583,
"grad_norm": 0.5970990880266768,
"learning_rate": 4.559209649090039e-05,
"loss": 0.5648,
"step": 2080
},
{
"epoch": 0.2431203358208955,
"grad_norm": 0.5310811562750187,
"learning_rate": 4.556627242744703e-05,
"loss": 0.5616,
"step": 2085
},
{
"epoch": 0.24370335820895522,
"grad_norm": 0.4892958829299151,
"learning_rate": 4.5540381203675994e-05,
"loss": 0.5867,
"step": 2090
},
{
"epoch": 0.24428638059701493,
"grad_norm": 0.5227922779707119,
"learning_rate": 4.55144229158364e-05,
"loss": 0.5534,
"step": 2095
},
{
"epoch": 0.24486940298507462,
"grad_norm": 0.5041942400636352,
"learning_rate": 4.548839766042668e-05,
"loss": 0.5371,
"step": 2100
},
{
"epoch": 0.24545242537313433,
"grad_norm": 0.49660405961707166,
"learning_rate": 4.5462305534194204e-05,
"loss": 0.572,
"step": 2105
},
{
"epoch": 0.24603544776119404,
"grad_norm": 0.5499042134726534,
"learning_rate": 4.543614663413493e-05,
"loss": 0.5611,
"step": 2110
},
{
"epoch": 0.24661847014925373,
"grad_norm": 0.5496526828713953,
"learning_rate": 4.5409921057493064e-05,
"loss": 0.57,
"step": 2115
},
{
"epoch": 0.24720149253731344,
"grad_norm": 0.5599675685862884,
"learning_rate": 4.538362890176066e-05,
"loss": 0.5618,
"step": 2120
},
{
"epoch": 0.24778451492537312,
"grad_norm": 0.5579889614380124,
"learning_rate": 4.535727026467727e-05,
"loss": 0.5682,
"step": 2125
},
{
"epoch": 0.24836753731343283,
"grad_norm": 0.5300253135611827,
"learning_rate": 4.533084524422959e-05,
"loss": 0.5828,
"step": 2130
},
{
"epoch": 0.24895055970149255,
"grad_norm": 0.498730689430679,
"learning_rate": 4.530435393865111e-05,
"loss": 0.5535,
"step": 2135
},
{
"epoch": 0.24953358208955223,
"grad_norm": 0.5177126292248756,
"learning_rate": 4.527779644642172e-05,
"loss": 0.5661,
"step": 2140
},
{
"epoch": 0.2501166044776119,
"grad_norm": 0.5708740823647661,
"learning_rate": 4.525117286626734e-05,
"loss": 0.5764,
"step": 2145
},
{
"epoch": 0.25069962686567165,
"grad_norm": 0.5326668265886716,
"learning_rate": 4.522448329715959e-05,
"loss": 0.562,
"step": 2150
},
{
"epoch": 0.25128264925373134,
"grad_norm": 0.4878189090302149,
"learning_rate": 4.51977278383154e-05,
"loss": 0.5389,
"step": 2155
},
{
"epoch": 0.251865671641791,
"grad_norm": 0.5459043435498905,
"learning_rate": 4.517090658919662e-05,
"loss": 0.5456,
"step": 2160
},
{
"epoch": 0.25244869402985076,
"grad_norm": 0.5451507549234296,
"learning_rate": 4.5144019649509694e-05,
"loss": 0.5619,
"step": 2165
},
{
"epoch": 0.25303171641791045,
"grad_norm": 0.48817546876328066,
"learning_rate": 4.5117067119205256e-05,
"loss": 0.5328,
"step": 2170
},
{
"epoch": 0.25361473880597013,
"grad_norm": 0.5573481611505946,
"learning_rate": 4.5090049098477756e-05,
"loss": 0.6166,
"step": 2175
},
{
"epoch": 0.25419776119402987,
"grad_norm": 0.46166230865153285,
"learning_rate": 4.506296568776513e-05,
"loss": 0.5603,
"step": 2180
},
{
"epoch": 0.25478078358208955,
"grad_norm": 0.5255525351801149,
"learning_rate": 4.503581698774838e-05,
"loss": 0.5597,
"step": 2185
},
{
"epoch": 0.25536380597014924,
"grad_norm": 0.5036713071359231,
"learning_rate": 4.5008603099351235e-05,
"loss": 0.5572,
"step": 2190
},
{
"epoch": 0.255946828358209,
"grad_norm": 0.5629197395168302,
"learning_rate": 4.498132412373972e-05,
"loss": 0.5549,
"step": 2195
},
{
"epoch": 0.25652985074626866,
"grad_norm": 0.5281653518122478,
"learning_rate": 4.4953980162321845e-05,
"loss": 0.5783,
"step": 2200
},
{
"epoch": 0.25711287313432835,
"grad_norm": 0.52357818056316,
"learning_rate": 4.492657131674722e-05,
"loss": 0.5365,
"step": 2205
},
{
"epoch": 0.2576958955223881,
"grad_norm": 0.5581473844002001,
"learning_rate": 4.48990976889066e-05,
"loss": 0.5314,
"step": 2210
},
{
"epoch": 0.25827891791044777,
"grad_norm": 0.5463641738254288,
"learning_rate": 4.487155938093163e-05,
"loss": 0.5678,
"step": 2215
},
{
"epoch": 0.25886194029850745,
"grad_norm": 0.580427123805623,
"learning_rate": 4.484395649519435e-05,
"loss": 0.5831,
"step": 2220
},
{
"epoch": 0.25944496268656714,
"grad_norm": 0.5515157134010146,
"learning_rate": 4.48162891343069e-05,
"loss": 0.5916,
"step": 2225
},
{
"epoch": 0.2600279850746269,
"grad_norm": 0.4614774802102389,
"learning_rate": 4.478855740112107e-05,
"loss": 0.5747,
"step": 2230
},
{
"epoch": 0.26061100746268656,
"grad_norm": 0.5263544516058696,
"learning_rate": 4.476076139872797e-05,
"loss": 0.5873,
"step": 2235
},
{
"epoch": 0.26119402985074625,
"grad_norm": 0.5279147571827948,
"learning_rate": 4.473290123045764e-05,
"loss": 0.5694,
"step": 2240
},
{
"epoch": 0.261777052238806,
"grad_norm": 0.9742556257776407,
"learning_rate": 4.470497699987861e-05,
"loss": 0.5686,
"step": 2245
},
{
"epoch": 0.26236007462686567,
"grad_norm": 0.5328912557304052,
"learning_rate": 4.4676988810797596e-05,
"loss": 0.5591,
"step": 2250
},
{
"epoch": 0.26294309701492535,
"grad_norm": 0.5633803039126256,
"learning_rate": 4.464893676725906e-05,
"loss": 0.5728,
"step": 2255
},
{
"epoch": 0.2635261194029851,
"grad_norm": 0.5134466833234131,
"learning_rate": 4.4620820973544866e-05,
"loss": 0.588,
"step": 2260
},
{
"epoch": 0.2641091417910448,
"grad_norm": 0.5049188829803447,
"learning_rate": 4.459264153417381e-05,
"loss": 0.5973,
"step": 2265
},
{
"epoch": 0.26469216417910446,
"grad_norm": 0.5404106159275316,
"learning_rate": 4.4564398553901344e-05,
"loss": 0.5788,
"step": 2270
},
{
"epoch": 0.2652751865671642,
"grad_norm": 0.6037489755180414,
"learning_rate": 4.4536092137719094e-05,
"loss": 0.5935,
"step": 2275
},
{
"epoch": 0.2658582089552239,
"grad_norm": 0.5161487013459114,
"learning_rate": 4.450772239085452e-05,
"loss": 0.5371,
"step": 2280
},
{
"epoch": 0.26644123134328357,
"grad_norm": 0.5278496691940492,
"learning_rate": 4.44792894187705e-05,
"loss": 0.5234,
"step": 2285
},
{
"epoch": 0.2670242537313433,
"grad_norm": 0.5092042449789793,
"learning_rate": 4.445079332716497e-05,
"loss": 0.5925,
"step": 2290
},
{
"epoch": 0.267607276119403,
"grad_norm": 0.5223498221449892,
"learning_rate": 4.4422234221970475e-05,
"loss": 0.5629,
"step": 2295
},
{
"epoch": 0.2681902985074627,
"grad_norm": 0.5456470351851613,
"learning_rate": 4.439361220935385e-05,
"loss": 0.5416,
"step": 2300
},
{
"epoch": 0.26877332089552236,
"grad_norm": 0.5476526134142059,
"learning_rate": 4.436492739571575e-05,
"loss": 0.5499,
"step": 2305
},
{
"epoch": 0.2693563432835821,
"grad_norm": 0.49505688936501147,
"learning_rate": 4.433617988769031e-05,
"loss": 0.5431,
"step": 2310
},
{
"epoch": 0.2699393656716418,
"grad_norm": 0.5342065164923998,
"learning_rate": 4.43073697921447e-05,
"loss": 0.5693,
"step": 2315
},
{
"epoch": 0.27052238805970147,
"grad_norm": 0.5145761250584094,
"learning_rate": 4.4278497216178805e-05,
"loss": 0.5795,
"step": 2320
},
{
"epoch": 0.2711054104477612,
"grad_norm": 0.5273626768859015,
"learning_rate": 4.4249562267124735e-05,
"loss": 0.5549,
"step": 2325
},
{
"epoch": 0.2716884328358209,
"grad_norm": 0.5079624559473638,
"learning_rate": 4.422056505254648e-05,
"loss": 0.5746,
"step": 2330
},
{
"epoch": 0.2722714552238806,
"grad_norm": 0.505058134321932,
"learning_rate": 4.4191505680239494e-05,
"loss": 0.5565,
"step": 2335
},
{
"epoch": 0.2728544776119403,
"grad_norm": 0.5668879894960839,
"learning_rate": 4.416238425823031e-05,
"loss": 0.5578,
"step": 2340
},
{
"epoch": 0.2734375,
"grad_norm": 0.5141499926926323,
"learning_rate": 4.413320089477612e-05,
"loss": 0.5489,
"step": 2345
},
{
"epoch": 0.2740205223880597,
"grad_norm": 0.5109670063026779,
"learning_rate": 4.4103955698364394e-05,
"loss": 0.5592,
"step": 2350
},
{
"epoch": 0.2746035447761194,
"grad_norm": 0.5082096864806888,
"learning_rate": 4.407464877771243e-05,
"loss": 0.5213,
"step": 2355
},
{
"epoch": 0.2751865671641791,
"grad_norm": 0.5083698199510284,
"learning_rate": 4.4045280241767024e-05,
"loss": 0.5493,
"step": 2360
},
{
"epoch": 0.2757695895522388,
"grad_norm": 0.5053920381989545,
"learning_rate": 4.401585019970397e-05,
"loss": 0.5686,
"step": 2365
},
{
"epoch": 0.27635261194029853,
"grad_norm": 0.4963501376761793,
"learning_rate": 4.3986358760927774e-05,
"loss": 0.5545,
"step": 2370
},
{
"epoch": 0.2769356343283582,
"grad_norm": 0.5288831258581409,
"learning_rate": 4.3956806035071123e-05,
"loss": 0.5249,
"step": 2375
},
{
"epoch": 0.2775186567164179,
"grad_norm": 0.62308836025263,
"learning_rate": 4.392719213199457e-05,
"loss": 0.6042,
"step": 2380
},
{
"epoch": 0.27810167910447764,
"grad_norm": 0.4832296950151894,
"learning_rate": 4.389751716178606e-05,
"loss": 0.5633,
"step": 2385
},
{
"epoch": 0.2786847014925373,
"grad_norm": 0.5237890773447569,
"learning_rate": 4.386778123476059e-05,
"loss": 0.5271,
"step": 2390
},
{
"epoch": 0.279267723880597,
"grad_norm": 0.5595739424570599,
"learning_rate": 4.383798446145973e-05,
"loss": 0.6001,
"step": 2395
},
{
"epoch": 0.2798507462686567,
"grad_norm": 0.5332931182325285,
"learning_rate": 4.380812695265126e-05,
"loss": 0.6024,
"step": 2400
},
{
"epoch": 0.28043376865671643,
"grad_norm": 0.4918731480131771,
"learning_rate": 4.3778208819328724e-05,
"loss": 0.56,
"step": 2405
},
{
"epoch": 0.2810167910447761,
"grad_norm": 0.4625310842353903,
"learning_rate": 4.374823017271105e-05,
"loss": 0.5161,
"step": 2410
},
{
"epoch": 0.2815998134328358,
"grad_norm": 0.5070557648263391,
"learning_rate": 4.371819112424212e-05,
"loss": 0.518,
"step": 2415
},
{
"epoch": 0.28218283582089554,
"grad_norm": 0.56213077270057,
"learning_rate": 4.368809178559034e-05,
"loss": 0.5645,
"step": 2420
},
{
"epoch": 0.2827658582089552,
"grad_norm": 0.5027850751505164,
"learning_rate": 4.365793226864825e-05,
"loss": 0.5516,
"step": 2425
},
{
"epoch": 0.2833488805970149,
"grad_norm": 0.496969778411097,
"learning_rate": 4.3627712685532104e-05,
"loss": 0.5661,
"step": 2430
},
{
"epoch": 0.28393190298507465,
"grad_norm": 0.5203901533643535,
"learning_rate": 4.3597433148581465e-05,
"loss": 0.5564,
"step": 2435
},
{
"epoch": 0.28451492537313433,
"grad_norm": 0.5761459191507305,
"learning_rate": 4.3567093770358724e-05,
"loss": 0.584,
"step": 2440
},
{
"epoch": 0.285097947761194,
"grad_norm": 0.5440010534822834,
"learning_rate": 4.353669466364877e-05,
"loss": 0.5833,
"step": 2445
},
{
"epoch": 0.28568097014925375,
"grad_norm": 0.5575628208489772,
"learning_rate": 4.3506235941458516e-05,
"loss": 0.5926,
"step": 2450
},
{
"epoch": 0.28626399253731344,
"grad_norm": 0.500211988994469,
"learning_rate": 4.347571771701648e-05,
"loss": 0.5651,
"step": 2455
},
{
"epoch": 0.2868470149253731,
"grad_norm": 0.46825863619458763,
"learning_rate": 4.34451401037724e-05,
"loss": 0.5461,
"step": 2460
},
{
"epoch": 0.28743003731343286,
"grad_norm": 0.5521309621293441,
"learning_rate": 4.3414503215396776e-05,
"loss": 0.5659,
"step": 2465
},
{
"epoch": 0.28801305970149255,
"grad_norm": 0.4609786323278,
"learning_rate": 4.338380716578046e-05,
"loss": 0.5487,
"step": 2470
},
{
"epoch": 0.28859608208955223,
"grad_norm": 0.499738534899754,
"learning_rate": 4.3353052069034214e-05,
"loss": 0.5381,
"step": 2475
},
{
"epoch": 0.2891791044776119,
"grad_norm": 0.5917418512742165,
"learning_rate": 4.332223803948834e-05,
"loss": 0.5434,
"step": 2480
},
{
"epoch": 0.28976212686567165,
"grad_norm": 0.48180474136483686,
"learning_rate": 4.3291365191692204e-05,
"loss": 0.5734,
"step": 2485
},
{
"epoch": 0.29034514925373134,
"grad_norm": 0.5007564237127459,
"learning_rate": 4.326043364041381e-05,
"loss": 0.5352,
"step": 2490
},
{
"epoch": 0.290928171641791,
"grad_norm": 0.47349703509695523,
"learning_rate": 4.3229443500639414e-05,
"loss": 0.5553,
"step": 2495
},
{
"epoch": 0.29151119402985076,
"grad_norm": 0.5192204895800268,
"learning_rate": 4.319839488757305e-05,
"loss": 0.5398,
"step": 2500
},
{
"epoch": 0.29209421641791045,
"grad_norm": 0.5757316112230765,
"learning_rate": 4.3167287916636145e-05,
"loss": 0.5795,
"step": 2505
},
{
"epoch": 0.29267723880597013,
"grad_norm": 0.5046004476804008,
"learning_rate": 4.3136122703467045e-05,
"loss": 0.5657,
"step": 2510
},
{
"epoch": 0.29326026119402987,
"grad_norm": 0.6116298626095651,
"learning_rate": 4.3104899363920616e-05,
"loss": 0.5892,
"step": 2515
},
{
"epoch": 0.29384328358208955,
"grad_norm": 0.5028293067243745,
"learning_rate": 4.3073618014067824e-05,
"loss": 0.5614,
"step": 2520
},
{
"epoch": 0.29442630597014924,
"grad_norm": 0.5284740774719737,
"learning_rate": 4.304227877019525e-05,
"loss": 0.5509,
"step": 2525
},
{
"epoch": 0.295009328358209,
"grad_norm": 0.5043452273361897,
"learning_rate": 4.301088174880472e-05,
"loss": 0.5734,
"step": 2530
},
{
"epoch": 0.29559235074626866,
"grad_norm": 0.5491908726146885,
"learning_rate": 4.297942706661283e-05,
"loss": 0.5994,
"step": 2535
},
{
"epoch": 0.29617537313432835,
"grad_norm": 0.5967635993887994,
"learning_rate": 4.2947914840550544e-05,
"loss": 0.5895,
"step": 2540
},
{
"epoch": 0.2967583955223881,
"grad_norm": 0.49957976391073805,
"learning_rate": 4.291634518776273e-05,
"loss": 0.5559,
"step": 2545
},
{
"epoch": 0.29734141791044777,
"grad_norm": 0.586452369328343,
"learning_rate": 4.2884718225607736e-05,
"loss": 0.5987,
"step": 2550
},
{
"epoch": 0.29792444029850745,
"grad_norm": 0.5079777653661454,
"learning_rate": 4.285303407165694e-05,
"loss": 0.5892,
"step": 2555
},
{
"epoch": 0.29850746268656714,
"grad_norm": 0.45945289033221137,
"learning_rate": 4.282129284369436e-05,
"loss": 0.5353,
"step": 2560
},
{
"epoch": 0.2990904850746269,
"grad_norm": 0.47293762371540526,
"learning_rate": 4.278949465971616e-05,
"loss": 0.5451,
"step": 2565
},
{
"epoch": 0.29967350746268656,
"grad_norm": 0.5125425687772994,
"learning_rate": 4.2757639637930246e-05,
"loss": 0.588,
"step": 2570
},
{
"epoch": 0.30025652985074625,
"grad_norm": 0.560469657657478,
"learning_rate": 4.2725727896755804e-05,
"loss": 0.5353,
"step": 2575
},
{
"epoch": 0.300839552238806,
"grad_norm": 0.5254788469907365,
"learning_rate": 4.269375955482287e-05,
"loss": 0.547,
"step": 2580
},
{
"epoch": 0.30142257462686567,
"grad_norm": 0.48987507837663824,
"learning_rate": 4.266173473097192e-05,
"loss": 0.5175,
"step": 2585
},
{
"epoch": 0.30200559701492535,
"grad_norm": 0.5484732726515896,
"learning_rate": 4.262965354425335e-05,
"loss": 0.6067,
"step": 2590
},
{
"epoch": 0.3025886194029851,
"grad_norm": 0.5064490667857149,
"learning_rate": 4.259751611392712e-05,
"loss": 0.533,
"step": 2595
},
{
"epoch": 0.3031716417910448,
"grad_norm": 0.5014575792637805,
"learning_rate": 4.256532255946226e-05,
"loss": 0.5913,
"step": 2600
},
{
"epoch": 0.30375466417910446,
"grad_norm": 0.48901348419554475,
"learning_rate": 4.253307300053643e-05,
"loss": 0.5543,
"step": 2605
},
{
"epoch": 0.3043376865671642,
"grad_norm": 0.5232458296317694,
"learning_rate": 4.25007675570355e-05,
"loss": 0.5759,
"step": 2610
},
{
"epoch": 0.3049207089552239,
"grad_norm": 0.5070730983893377,
"learning_rate": 4.246840634905307e-05,
"loss": 0.5515,
"step": 2615
},
{
"epoch": 0.30550373134328357,
"grad_norm": 0.4828637362156461,
"learning_rate": 4.2435989496890054e-05,
"loss": 0.5817,
"step": 2620
},
{
"epoch": 0.3060867537313433,
"grad_norm": 0.45681007096402865,
"learning_rate": 4.240351712105422e-05,
"loss": 0.5538,
"step": 2625
},
{
"epoch": 0.306669776119403,
"grad_norm": 0.5633919156991598,
"learning_rate": 4.237098934225973e-05,
"loss": 0.5619,
"step": 2630
},
{
"epoch": 0.3072527985074627,
"grad_norm": 0.4744742163372956,
"learning_rate": 4.233840628142672e-05,
"loss": 0.5749,
"step": 2635
},
{
"epoch": 0.30783582089552236,
"grad_norm": 0.47221088669237776,
"learning_rate": 4.2305768059680806e-05,
"loss": 0.5689,
"step": 2640
},
{
"epoch": 0.3084188432835821,
"grad_norm": 0.4762515982934417,
"learning_rate": 4.2273074798352706e-05,
"loss": 0.5507,
"step": 2645
},
{
"epoch": 0.3090018656716418,
"grad_norm": 0.4718217399644616,
"learning_rate": 4.22403266189777e-05,
"loss": 0.5483,
"step": 2650
},
{
"epoch": 0.30958488805970147,
"grad_norm": 0.4671780211152603,
"learning_rate": 4.2207523643295253e-05,
"loss": 0.5463,
"step": 2655
},
{
"epoch": 0.3101679104477612,
"grad_norm": 0.49004196187956206,
"learning_rate": 4.2174665993248505e-05,
"loss": 0.5474,
"step": 2660
},
{
"epoch": 0.3107509328358209,
"grad_norm": 0.5092130863744895,
"learning_rate": 4.214175379098388e-05,
"loss": 0.5512,
"step": 2665
},
{
"epoch": 0.3113339552238806,
"grad_norm": 0.4925159889445034,
"learning_rate": 4.210878715885056e-05,
"loss": 0.5609,
"step": 2670
},
{
"epoch": 0.3119169776119403,
"grad_norm": 0.5438536521932689,
"learning_rate": 4.2075766219400095e-05,
"loss": 0.5949,
"step": 2675
},
{
"epoch": 0.3125,
"grad_norm": 0.4985765329826545,
"learning_rate": 4.20426910953859e-05,
"loss": 0.5726,
"step": 2680
},
{
"epoch": 0.3130830223880597,
"grad_norm": 0.5036063584187528,
"learning_rate": 4.200956190976284e-05,
"loss": 0.5255,
"step": 2685
},
{
"epoch": 0.3136660447761194,
"grad_norm": 0.5634159268083127,
"learning_rate": 4.1976378785686715e-05,
"loss": 0.5254,
"step": 2690
},
{
"epoch": 0.3142490671641791,
"grad_norm": 0.483875275656377,
"learning_rate": 4.1943141846513886e-05,
"loss": 0.5518,
"step": 2695
},
{
"epoch": 0.3148320895522388,
"grad_norm": 0.5510683619678165,
"learning_rate": 4.190985121580071e-05,
"loss": 0.557,
"step": 2700
},
{
"epoch": 0.31541511194029853,
"grad_norm": 0.45786711381238016,
"learning_rate": 4.18765070173032e-05,
"loss": 0.521,
"step": 2705
},
{
"epoch": 0.3159981343283582,
"grad_norm": 0.47954794442661985,
"learning_rate": 4.184310937497647e-05,
"loss": 0.55,
"step": 2710
},
{
"epoch": 0.3165811567164179,
"grad_norm": 0.5348827941441646,
"learning_rate": 4.1809658412974314e-05,
"loss": 0.5668,
"step": 2715
},
{
"epoch": 0.31716417910447764,
"grad_norm": 0.5035567889348412,
"learning_rate": 4.177615425564872e-05,
"loss": 0.5584,
"step": 2720
},
{
"epoch": 0.3177472014925373,
"grad_norm": 0.5014049010285576,
"learning_rate": 4.174259702754947e-05,
"loss": 0.5538,
"step": 2725
},
{
"epoch": 0.318330223880597,
"grad_norm": 0.7108080618041146,
"learning_rate": 4.17089868534236e-05,
"loss": 0.593,
"step": 2730
},
{
"epoch": 0.3189132462686567,
"grad_norm": 0.4901344322857948,
"learning_rate": 4.1675323858214975e-05,
"loss": 0.5504,
"step": 2735
},
{
"epoch": 0.31949626865671643,
"grad_norm": 0.5455638212871261,
"learning_rate": 4.164160816706383e-05,
"loss": 0.5781,
"step": 2740
},
{
"epoch": 0.3200792910447761,
"grad_norm": 0.4691865337288382,
"learning_rate": 4.160783990530629e-05,
"loss": 0.5349,
"step": 2745
},
{
"epoch": 0.3206623134328358,
"grad_norm": 0.5397164117756076,
"learning_rate": 4.157401919847389e-05,
"loss": 0.5635,
"step": 2750
},
{
"epoch": 0.32124533582089554,
"grad_norm": 0.49203133885978545,
"learning_rate": 4.1540146172293154e-05,
"loss": 0.553,
"step": 2755
},
{
"epoch": 0.3218283582089552,
"grad_norm": 0.6956383718262041,
"learning_rate": 4.150622095268508e-05,
"loss": 0.5454,
"step": 2760
},
{
"epoch": 0.3224113805970149,
"grad_norm": 0.5236479255481173,
"learning_rate": 4.1472243665764715e-05,
"loss": 0.546,
"step": 2765
},
{
"epoch": 0.32299440298507465,
"grad_norm": 0.5226139094665005,
"learning_rate": 4.1438214437840625e-05,
"loss": 0.5685,
"step": 2770
},
{
"epoch": 0.32357742537313433,
"grad_norm": 0.519532395995944,
"learning_rate": 4.140413339541451e-05,
"loss": 0.5621,
"step": 2775
},
{
"epoch": 0.324160447761194,
"grad_norm": 0.5221117984856967,
"learning_rate": 4.137000066518065e-05,
"loss": 0.5945,
"step": 2780
},
{
"epoch": 0.32474347014925375,
"grad_norm": 0.4662764439552845,
"learning_rate": 4.13358163740255e-05,
"loss": 0.5439,
"step": 2785
},
{
"epoch": 0.32532649253731344,
"grad_norm": 0.4825709779338081,
"learning_rate": 4.1301580649027154e-05,
"loss": 0.5487,
"step": 2790
},
{
"epoch": 0.3259095149253731,
"grad_norm": 0.49595525241615773,
"learning_rate": 4.126729361745495e-05,
"loss": 0.5456,
"step": 2795
},
{
"epoch": 0.32649253731343286,
"grad_norm": 0.46259611525800753,
"learning_rate": 4.1232955406768925e-05,
"loss": 0.5385,
"step": 2800
},
{
"epoch": 0.32707555970149255,
"grad_norm": 0.48140441921381905,
"learning_rate": 4.119856614461938e-05,
"loss": 0.5289,
"step": 2805
},
{
"epoch": 0.32765858208955223,
"grad_norm": 0.4878844018028054,
"learning_rate": 4.11641259588464e-05,
"loss": 0.5698,
"step": 2810
},
{
"epoch": 0.3282416044776119,
"grad_norm": 0.521050501093044,
"learning_rate": 4.1129634977479375e-05,
"loss": 0.5609,
"step": 2815
},
{
"epoch": 0.32882462686567165,
"grad_norm": 0.49828928878222534,
"learning_rate": 4.109509332873653e-05,
"loss": 0.5421,
"step": 2820
},
{
"epoch": 0.32940764925373134,
"grad_norm": 0.5526407883676059,
"learning_rate": 4.106050114102443e-05,
"loss": 0.5789,
"step": 2825
},
{
"epoch": 0.329990671641791,
"grad_norm": 0.5516248967750637,
"learning_rate": 4.102585854293751e-05,
"loss": 0.526,
"step": 2830
},
{
"epoch": 0.33057369402985076,
"grad_norm": 0.5461716849773546,
"learning_rate": 4.0991165663257636e-05,
"loss": 0.542,
"step": 2835
},
{
"epoch": 0.33115671641791045,
"grad_norm": 0.4524230892035543,
"learning_rate": 4.095642263095356e-05,
"loss": 0.5429,
"step": 2840
},
{
"epoch": 0.33173973880597013,
"grad_norm": 0.4907608846635581,
"learning_rate": 4.0921629575180485e-05,
"loss": 0.5536,
"step": 2845
},
{
"epoch": 0.33232276119402987,
"grad_norm": 0.5284875026264199,
"learning_rate": 4.088678662527959e-05,
"loss": 0.6164,
"step": 2850
},
{
"epoch": 0.33290578358208955,
"grad_norm": 0.48638662847344827,
"learning_rate": 4.085189391077749e-05,
"loss": 0.576,
"step": 2855
},
{
"epoch": 0.33348880597014924,
"grad_norm": 0.5695349098133777,
"learning_rate": 4.0816951561385836e-05,
"loss": 0.5521,
"step": 2860
},
{
"epoch": 0.334071828358209,
"grad_norm": 0.44821729017419637,
"learning_rate": 4.078195970700079e-05,
"loss": 0.526,
"step": 2865
},
{
"epoch": 0.33465485074626866,
"grad_norm": 0.5289515772805433,
"learning_rate": 4.074691847770251e-05,
"loss": 0.5505,
"step": 2870
},
{
"epoch": 0.33523787313432835,
"grad_norm": 0.5491964687999824,
"learning_rate": 4.0711828003754764e-05,
"loss": 0.5795,
"step": 2875
},
{
"epoch": 0.3358208955223881,
"grad_norm": 0.5184535389920294,
"learning_rate": 4.067668841560432e-05,
"loss": 0.5864,
"step": 2880
},
{
"epoch": 0.33640391791044777,
"grad_norm": 0.4640726695845034,
"learning_rate": 4.064149984388057e-05,
"loss": 0.5377,
"step": 2885
},
{
"epoch": 0.33698694029850745,
"grad_norm": 0.5223555480526277,
"learning_rate": 4.060626241939499e-05,
"loss": 0.5528,
"step": 2890
},
{
"epoch": 0.33756996268656714,
"grad_norm": 0.47444935336532806,
"learning_rate": 4.057097627314063e-05,
"loss": 0.5366,
"step": 2895
},
{
"epoch": 0.3381529850746269,
"grad_norm": 0.49936360299242055,
"learning_rate": 4.0535641536291725e-05,
"loss": 0.5981,
"step": 2900
},
{
"epoch": 0.33873600746268656,
"grad_norm": 0.49567180693590535,
"learning_rate": 4.050025834020307e-05,
"loss": 0.564,
"step": 2905
},
{
"epoch": 0.33931902985074625,
"grad_norm": 0.48231305889253284,
"learning_rate": 4.046482681640967e-05,
"loss": 0.5559,
"step": 2910
},
{
"epoch": 0.339902052238806,
"grad_norm": 0.5713231807843953,
"learning_rate": 4.042934709662613e-05,
"loss": 0.6046,
"step": 2915
},
{
"epoch": 0.34048507462686567,
"grad_norm": 0.5599418550405636,
"learning_rate": 4.039381931274626e-05,
"loss": 0.5459,
"step": 2920
},
{
"epoch": 0.34106809701492535,
"grad_norm": 0.5244470856538797,
"learning_rate": 4.035824359684253e-05,
"loss": 0.5559,
"step": 2925
},
{
"epoch": 0.3416511194029851,
"grad_norm": 0.46218246493450177,
"learning_rate": 4.032262008116559e-05,
"loss": 0.5496,
"step": 2930
},
{
"epoch": 0.3422341417910448,
"grad_norm": 0.5460308556475504,
"learning_rate": 4.02869488981438e-05,
"loss": 0.5726,
"step": 2935
},
{
"epoch": 0.34281716417910446,
"grad_norm": 0.4623737344709334,
"learning_rate": 4.025123018038271e-05,
"loss": 0.5422,
"step": 2940
},
{
"epoch": 0.3434001865671642,
"grad_norm": 0.5813565724181794,
"learning_rate": 4.0215464060664564e-05,
"loss": 0.5822,
"step": 2945
},
{
"epoch": 0.3439832089552239,
"grad_norm": 0.49376149527368923,
"learning_rate": 4.017965067194783e-05,
"loss": 0.5821,
"step": 2950
},
{
"epoch": 0.34456623134328357,
"grad_norm": 0.5169325234827248,
"learning_rate": 4.0143790147366724e-05,
"loss": 0.5482,
"step": 2955
},
{
"epoch": 0.3451492537313433,
"grad_norm": 0.515737079500457,
"learning_rate": 4.010788262023064e-05,
"loss": 0.5709,
"step": 2960
},
{
"epoch": 0.345732276119403,
"grad_norm": 0.5087208722494978,
"learning_rate": 4.007192822402372e-05,
"loss": 0.5565,
"step": 2965
},
{
"epoch": 0.3463152985074627,
"grad_norm": 0.4796963642209258,
"learning_rate": 4.003592709240438e-05,
"loss": 0.5494,
"step": 2970
},
{
"epoch": 0.34689832089552236,
"grad_norm": 0.49755643832004437,
"learning_rate": 3.9999879359204676e-05,
"loss": 0.5382,
"step": 2975
},
{
"epoch": 0.3474813432835821,
"grad_norm": 0.4690304314754959,
"learning_rate": 3.996378515843001e-05,
"loss": 0.5334,
"step": 2980
},
{
"epoch": 0.3480643656716418,
"grad_norm": 0.4772189221465707,
"learning_rate": 3.9927644624258445e-05,
"loss": 0.5902,
"step": 2985
},
{
"epoch": 0.34864738805970147,
"grad_norm": 0.46404845790530064,
"learning_rate": 3.989145789104033e-05,
"loss": 0.5431,
"step": 2990
},
{
"epoch": 0.3492304104477612,
"grad_norm": 0.5340362988246943,
"learning_rate": 3.985522509329775e-05,
"loss": 0.5486,
"step": 2995
},
{
"epoch": 0.3498134328358209,
"grad_norm": 0.510771920504794,
"learning_rate": 3.9818946365724004e-05,
"loss": 0.5401,
"step": 3000
},
{
"epoch": 0.3503964552238806,
"grad_norm": 0.49718158294490394,
"learning_rate": 3.978262184318317e-05,
"loss": 0.5626,
"step": 3005
},
{
"epoch": 0.3509794776119403,
"grad_norm": 0.5158633729934479,
"learning_rate": 3.974625166070953e-05,
"loss": 0.5364,
"step": 3010
},
{
"epoch": 0.3515625,
"grad_norm": 0.5204824836614376,
"learning_rate": 3.970983595350714e-05,
"loss": 0.5545,
"step": 3015
},
{
"epoch": 0.3521455223880597,
"grad_norm": 0.506085761814899,
"learning_rate": 3.967337485694929e-05,
"loss": 0.5492,
"step": 3020
},
{
"epoch": 0.3527285447761194,
"grad_norm": 0.5167191975821872,
"learning_rate": 3.963686850657795e-05,
"loss": 0.5326,
"step": 3025
},
{
"epoch": 0.3533115671641791,
"grad_norm": 0.49528479007831655,
"learning_rate": 3.9600317038103385e-05,
"loss": 0.548,
"step": 3030
},
{
"epoch": 0.3538945895522388,
"grad_norm": 0.514405428467145,
"learning_rate": 3.956372058740354e-05,
"loss": 0.5708,
"step": 3035
},
{
"epoch": 0.35447761194029853,
"grad_norm": 0.4760653718744451,
"learning_rate": 3.952707929052359e-05,
"loss": 0.5385,
"step": 3040
},
{
"epoch": 0.3550606343283582,
"grad_norm": 0.5086645138695323,
"learning_rate": 3.9490393283675445e-05,
"loss": 0.5425,
"step": 3045
},
{
"epoch": 0.3556436567164179,
"grad_norm": 0.48138382467030305,
"learning_rate": 3.9453662703237186e-05,
"loss": 0.5599,
"step": 3050
},
{
"epoch": 0.35622667910447764,
"grad_norm": 0.594129469541421,
"learning_rate": 3.941688768575261e-05,
"loss": 0.558,
"step": 3055
},
{
"epoch": 0.3568097014925373,
"grad_norm": 0.4896479074572315,
"learning_rate": 3.938006836793073e-05,
"loss": 0.5399,
"step": 3060
},
{
"epoch": 0.357392723880597,
"grad_norm": 0.5430860896504877,
"learning_rate": 3.934320488664519e-05,
"loss": 0.5614,
"step": 3065
},
{
"epoch": 0.3579757462686567,
"grad_norm": 0.5171549846963465,
"learning_rate": 3.9306297378933855e-05,
"loss": 0.5338,
"step": 3070
},
{
"epoch": 0.35855876865671643,
"grad_norm": 0.5030297563055794,
"learning_rate": 3.926934598199824e-05,
"loss": 0.5671,
"step": 3075
},
{
"epoch": 0.3591417910447761,
"grad_norm": 0.4738604088361965,
"learning_rate": 3.923235083320301e-05,
"loss": 0.5204,
"step": 3080
},
{
"epoch": 0.3597248134328358,
"grad_norm": 0.5081725848955316,
"learning_rate": 3.919531207007548e-05,
"loss": 0.5382,
"step": 3085
},
{
"epoch": 0.36030783582089554,
"grad_norm": 0.5336505974453601,
"learning_rate": 3.915822983030512e-05,
"loss": 0.5498,
"step": 3090
},
{
"epoch": 0.3608908582089552,
"grad_norm": 0.45862890034907244,
"learning_rate": 3.912110425174296e-05,
"loss": 0.5478,
"step": 3095
},
{
"epoch": 0.3614738805970149,
"grad_norm": 0.5692817255847623,
"learning_rate": 3.9083935472401214e-05,
"loss": 0.5511,
"step": 3100
},
{
"epoch": 0.36205690298507465,
"grad_norm": 0.5588630515105258,
"learning_rate": 3.904672363045265e-05,
"loss": 0.5713,
"step": 3105
},
{
"epoch": 0.36263992537313433,
"grad_norm": 0.7882164929530883,
"learning_rate": 3.900946886423012e-05,
"loss": 0.5967,
"step": 3110
},
{
"epoch": 0.363222947761194,
"grad_norm": 0.44828952211943274,
"learning_rate": 3.897217131222606e-05,
"loss": 0.5117,
"step": 3115
},
{
"epoch": 0.36380597014925375,
"grad_norm": 0.492899081711114,
"learning_rate": 3.893483111309196e-05,
"loss": 0.5381,
"step": 3120
},
{
"epoch": 0.36438899253731344,
"grad_norm": 0.46780156460670047,
"learning_rate": 3.889744840563781e-05,
"loss": 0.5561,
"step": 3125
},
{
"epoch": 0.3649720149253731,
"grad_norm": 0.5062858013949219,
"learning_rate": 3.886002332883169e-05,
"loss": 0.5501,
"step": 3130
},
{
"epoch": 0.36555503731343286,
"grad_norm": 0.4693101482730097,
"learning_rate": 3.8822556021799114e-05,
"loss": 0.5193,
"step": 3135
},
{
"epoch": 0.36613805970149255,
"grad_norm": 0.5060844807583814,
"learning_rate": 3.878504662382264e-05,
"loss": 0.5532,
"step": 3140
},
{
"epoch": 0.36672108208955223,
"grad_norm": 0.5201363699624885,
"learning_rate": 3.8747495274341274e-05,
"loss": 0.5845,
"step": 3145
},
{
"epoch": 0.3673041044776119,
"grad_norm": 0.5073064897043211,
"learning_rate": 3.870990211294997e-05,
"loss": 0.5444,
"step": 3150
},
{
"epoch": 0.36788712686567165,
"grad_norm": 0.5344016844623599,
"learning_rate": 3.867226727939912e-05,
"loss": 0.5606,
"step": 3155
},
{
"epoch": 0.36847014925373134,
"grad_norm": 0.5139307938716214,
"learning_rate": 3.863459091359401e-05,
"loss": 0.5882,
"step": 3160
},
{
"epoch": 0.369053171641791,
"grad_norm": 0.4425759816838131,
"learning_rate": 3.8596873155594385e-05,
"loss": 0.5202,
"step": 3165
},
{
"epoch": 0.36963619402985076,
"grad_norm": 0.5742393732533151,
"learning_rate": 3.855911414561378e-05,
"loss": 0.5662,
"step": 3170
},
{
"epoch": 0.37021921641791045,
"grad_norm": 0.44175421568947043,
"learning_rate": 3.852131402401914e-05,
"loss": 0.4854,
"step": 3175
},
{
"epoch": 0.37080223880597013,
"grad_norm": 0.4984055964239209,
"learning_rate": 3.848347293133021e-05,
"loss": 0.573,
"step": 3180
},
{
"epoch": 0.37138526119402987,
"grad_norm": 0.49757126890693293,
"learning_rate": 3.844559100821906e-05,
"loss": 0.5549,
"step": 3185
},
{
"epoch": 0.37196828358208955,
"grad_norm": 0.57357388001981,
"learning_rate": 3.8407668395509526e-05,
"loss": 0.5574,
"step": 3190
},
{
"epoch": 0.37255130597014924,
"grad_norm": 0.5019623696079913,
"learning_rate": 3.8369705234176726e-05,
"loss": 0.5894,
"step": 3195
},
{
"epoch": 0.373134328358209,
"grad_norm": 0.4858155386150297,
"learning_rate": 3.8331701665346495e-05,
"loss": 0.5383,
"step": 3200
},
{
"epoch": 0.37371735074626866,
"grad_norm": 0.5624113035875654,
"learning_rate": 3.829365783029492e-05,
"loss": 0.5585,
"step": 3205
},
{
"epoch": 0.37430037313432835,
"grad_norm": 0.5052807550374429,
"learning_rate": 3.8255573870447716e-05,
"loss": 0.5439,
"step": 3210
},
{
"epoch": 0.3748833955223881,
"grad_norm": 0.5085388458884429,
"learning_rate": 3.82174499273798e-05,
"loss": 0.565,
"step": 3215
},
{
"epoch": 0.37546641791044777,
"grad_norm": 0.4989534046655282,
"learning_rate": 3.817928614281471e-05,
"loss": 0.5324,
"step": 3220
},
{
"epoch": 0.37604944029850745,
"grad_norm": 0.8403204163889687,
"learning_rate": 3.8141082658624106e-05,
"loss": 0.5626,
"step": 3225
},
{
"epoch": 0.37663246268656714,
"grad_norm": 0.5559098643501764,
"learning_rate": 3.810283961682722e-05,
"loss": 0.5583,
"step": 3230
},
{
"epoch": 0.3772154850746269,
"grad_norm": 0.47644518540184416,
"learning_rate": 3.806455715959032e-05,
"loss": 0.5233,
"step": 3235
},
{
"epoch": 0.37779850746268656,
"grad_norm": 0.4930068420607561,
"learning_rate": 3.8026235429226236e-05,
"loss": 0.5642,
"step": 3240
},
{
"epoch": 0.37838152985074625,
"grad_norm": 0.5290426787630793,
"learning_rate": 3.798787456819377e-05,
"loss": 0.5234,
"step": 3245
},
{
"epoch": 0.378964552238806,
"grad_norm": 0.4868258441116788,
"learning_rate": 3.794947471909719e-05,
"loss": 0.5395,
"step": 3250
},
{
"epoch": 0.37954757462686567,
"grad_norm": 0.47174968576411624,
"learning_rate": 3.791103602468569e-05,
"loss": 0.5395,
"step": 3255
},
{
"epoch": 0.38013059701492535,
"grad_norm": 0.5363107908185231,
"learning_rate": 3.7872558627852905e-05,
"loss": 0.5557,
"step": 3260
},
{
"epoch": 0.3807136194029851,
"grad_norm": 0.49188998699814623,
"learning_rate": 3.78340426716363e-05,
"loss": 0.5268,
"step": 3265
},
{
"epoch": 0.3812966417910448,
"grad_norm": 0.48927529818476645,
"learning_rate": 3.779548829921673e-05,
"loss": 0.5387,
"step": 3270
},
{
"epoch": 0.38187966417910446,
"grad_norm": 0.47364382567752494,
"learning_rate": 3.775689565391781e-05,
"loss": 0.5129,
"step": 3275
},
{
"epoch": 0.3824626865671642,
"grad_norm": 0.5126016691334612,
"learning_rate": 3.771826487920546e-05,
"loss": 0.5467,
"step": 3280
},
{
"epoch": 0.3830457089552239,
"grad_norm": 0.5430085481644645,
"learning_rate": 3.767959611868734e-05,
"loss": 0.5536,
"step": 3285
},
{
"epoch": 0.38362873134328357,
"grad_norm": 0.4763687695806715,
"learning_rate": 3.764088951611233e-05,
"loss": 0.5071,
"step": 3290
},
{
"epoch": 0.3842117537313433,
"grad_norm": 0.47538986078328227,
"learning_rate": 3.7602145215369965e-05,
"loss": 0.5431,
"step": 3295
},
{
"epoch": 0.384794776119403,
"grad_norm": 0.44993211618209894,
"learning_rate": 3.756336336048994e-05,
"loss": 0.5608,
"step": 3300
},
{
"epoch": 0.3853777985074627,
"grad_norm": 0.4857510707261703,
"learning_rate": 3.752454409564152e-05,
"loss": 0.5307,
"step": 3305
},
{
"epoch": 0.38596082089552236,
"grad_norm": 0.49584291918089757,
"learning_rate": 3.74856875651331e-05,
"loss": 0.6047,
"step": 3310
},
{
"epoch": 0.3865438432835821,
"grad_norm": 0.5136523336548113,
"learning_rate": 3.744679391341157e-05,
"loss": 0.5516,
"step": 3315
},
{
"epoch": 0.3871268656716418,
"grad_norm": 0.5181136974711912,
"learning_rate": 3.740786328506179e-05,
"loss": 0.5439,
"step": 3320
},
{
"epoch": 0.38770988805970147,
"grad_norm": 0.5089785637081085,
"learning_rate": 3.7368895824806146e-05,
"loss": 0.5585,
"step": 3325
},
{
"epoch": 0.3882929104477612,
"grad_norm": 0.49682305539963434,
"learning_rate": 3.732989167750388e-05,
"loss": 0.562,
"step": 3330
},
{
"epoch": 0.3888759328358209,
"grad_norm": 0.5243718121493276,
"learning_rate": 3.7290850988150644e-05,
"loss": 0.5851,
"step": 3335
},
{
"epoch": 0.3894589552238806,
"grad_norm": 0.5098618228761496,
"learning_rate": 3.7251773901877945e-05,
"loss": 0.5191,
"step": 3340
},
{
"epoch": 0.3900419776119403,
"grad_norm": 0.47672882853039394,
"learning_rate": 3.721266056395257e-05,
"loss": 0.5284,
"step": 3345
},
{
"epoch": 0.390625,
"grad_norm": 0.4822727732005445,
"learning_rate": 3.7173511119776086e-05,
"loss": 0.5794,
"step": 3350
},
{
"epoch": 0.3912080223880597,
"grad_norm": 0.5222854783235708,
"learning_rate": 3.713432571488427e-05,
"loss": 0.5391,
"step": 3355
},
{
"epoch": 0.3917910447761194,
"grad_norm": 7.9962975681807595,
"learning_rate": 3.70951044949466e-05,
"loss": 0.5534,
"step": 3360
},
{
"epoch": 0.3923740671641791,
"grad_norm": 0.4764348433947033,
"learning_rate": 3.705584760576566e-05,
"loss": 0.5452,
"step": 3365
},
{
"epoch": 0.3929570895522388,
"grad_norm": 0.49080831286781906,
"learning_rate": 3.7016555193276667e-05,
"loss": 0.5746,
"step": 3370
},
{
"epoch": 0.39354011194029853,
"grad_norm": 0.5101400424865905,
"learning_rate": 3.697722740354688e-05,
"loss": 0.5729,
"step": 3375
},
{
"epoch": 0.3941231343283582,
"grad_norm": 0.4479587001588852,
"learning_rate": 3.6937864382775076e-05,
"loss": 0.5447,
"step": 3380
},
{
"epoch": 0.3947061567164179,
"grad_norm": 0.48466656620240023,
"learning_rate": 3.689846627729098e-05,
"loss": 0.5327,
"step": 3385
},
{
"epoch": 0.39528917910447764,
"grad_norm": 0.5157423888893855,
"learning_rate": 3.685903323355477e-05,
"loss": 0.5381,
"step": 3390
},
{
"epoch": 0.3958722014925373,
"grad_norm": 0.4820557574738787,
"learning_rate": 3.68195653981565e-05,
"loss": 0.5183,
"step": 3395
},
{
"epoch": 0.396455223880597,
"grad_norm": 0.48067690178778916,
"learning_rate": 3.678006291781555e-05,
"loss": 0.6121,
"step": 3400
},
{
"epoch": 0.3970382462686567,
"grad_norm": 0.48140274545401546,
"learning_rate": 3.6740525939380084e-05,
"loss": 0.5763,
"step": 3405
},
{
"epoch": 0.39762126865671643,
"grad_norm": 0.4821460133421974,
"learning_rate": 3.6700954609826535e-05,
"loss": 0.5498,
"step": 3410
},
{
"epoch": 0.3982042910447761,
"grad_norm": 0.5062442615898354,
"learning_rate": 3.6661349076259015e-05,
"loss": 0.5301,
"step": 3415
},
{
"epoch": 0.3987873134328358,
"grad_norm": 0.4894842886999361,
"learning_rate": 3.662170948590879e-05,
"loss": 0.5647,
"step": 3420
},
{
"epoch": 0.39937033582089554,
"grad_norm": 0.4635809485827499,
"learning_rate": 3.658203598613375e-05,
"loss": 0.5658,
"step": 3425
},
{
"epoch": 0.3999533582089552,
"grad_norm": 0.4756142318658808,
"learning_rate": 3.65423287244178e-05,
"loss": 0.5434,
"step": 3430
},
{
"epoch": 0.4005363805970149,
"grad_norm": 0.4768793799786962,
"learning_rate": 3.6502587848370395e-05,
"loss": 0.5129,
"step": 3435
},
{
"epoch": 0.40111940298507465,
"grad_norm": 0.49735047427469553,
"learning_rate": 3.6462813505725915e-05,
"loss": 0.5693,
"step": 3440
},
{
"epoch": 0.40170242537313433,
"grad_norm": 0.46768242382881264,
"learning_rate": 3.642300584434319e-05,
"loss": 0.5424,
"step": 3445
},
{
"epoch": 0.402285447761194,
"grad_norm": 0.48929227885921284,
"learning_rate": 3.638316501220487e-05,
"loss": 0.5613,
"step": 3450
},
{
"epoch": 0.40286847014925375,
"grad_norm": 0.49035092965063865,
"learning_rate": 3.6343291157416937e-05,
"loss": 0.5747,
"step": 3455
},
{
"epoch": 0.40345149253731344,
"grad_norm": 0.4920873718095396,
"learning_rate": 3.630338442820814e-05,
"loss": 0.5621,
"step": 3460
},
{
"epoch": 0.4040345149253731,
"grad_norm": 0.5721821448291733,
"learning_rate": 3.6263444972929395e-05,
"loss": 0.5642,
"step": 3465
},
{
"epoch": 0.40461753731343286,
"grad_norm": 0.4900664650078509,
"learning_rate": 3.622347294005334e-05,
"loss": 0.5637,
"step": 3470
},
{
"epoch": 0.40520055970149255,
"grad_norm": 0.4784540540109224,
"learning_rate": 3.618346847817366e-05,
"loss": 0.5166,
"step": 3475
},
{
"epoch": 0.40578358208955223,
"grad_norm": 0.5033729998350225,
"learning_rate": 3.6143431736004636e-05,
"loss": 0.5604,
"step": 3480
},
{
"epoch": 0.4063666044776119,
"grad_norm": 0.47667240088779256,
"learning_rate": 3.610336286238051e-05,
"loss": 0.5492,
"step": 3485
},
{
"epoch": 0.40694962686567165,
"grad_norm": 0.4874513651352348,
"learning_rate": 3.6063262006255006e-05,
"loss": 0.5563,
"step": 3490
},
{
"epoch": 0.40753264925373134,
"grad_norm": 0.523091113003475,
"learning_rate": 3.602312931670073e-05,
"loss": 0.5488,
"step": 3495
},
{
"epoch": 0.408115671641791,
"grad_norm": 0.5531785395812051,
"learning_rate": 3.59829649429086e-05,
"loss": 0.6013,
"step": 3500
},
{
"epoch": 0.40869869402985076,
"grad_norm": 0.47185275980717517,
"learning_rate": 3.5942769034187354e-05,
"loss": 0.536,
"step": 3505
},
{
"epoch": 0.40928171641791045,
"grad_norm": 0.46288526087001597,
"learning_rate": 3.590254173996295e-05,
"loss": 0.5169,
"step": 3510
},
{
"epoch": 0.40986473880597013,
"grad_norm": 0.48524872738357405,
"learning_rate": 3.586228320977801e-05,
"loss": 0.5271,
"step": 3515
},
{
"epoch": 0.41044776119402987,
"grad_norm": 0.5450240726382458,
"learning_rate": 3.582199359329129e-05,
"loss": 0.5427,
"step": 3520
},
{
"epoch": 0.41103078358208955,
"grad_norm": 0.48693873450848696,
"learning_rate": 3.5781673040277084e-05,
"loss": 0.5088,
"step": 3525
},
{
"epoch": 0.41161380597014924,
"grad_norm": 0.46864955154362703,
"learning_rate": 3.5741321700624726e-05,
"loss": 0.5174,
"step": 3530
},
{
"epoch": 0.412196828358209,
"grad_norm": 0.46728302060083887,
"learning_rate": 3.570093972433794e-05,
"loss": 0.5687,
"step": 3535
},
{
"epoch": 0.41277985074626866,
"grad_norm": 0.4634797945645771,
"learning_rate": 3.56605272615344e-05,
"loss": 0.528,
"step": 3540
},
{
"epoch": 0.41336287313432835,
"grad_norm": 0.5238785746405048,
"learning_rate": 3.562008446244509e-05,
"loss": 0.5375,
"step": 3545
},
{
"epoch": 0.4139458955223881,
"grad_norm": 0.5187794803590449,
"learning_rate": 3.557961147741376e-05,
"loss": 0.5985,
"step": 3550
},
{
"epoch": 0.41452891791044777,
"grad_norm": 0.4379680762591855,
"learning_rate": 3.553910845689638e-05,
"loss": 0.5554,
"step": 3555
},
{
"epoch": 0.41511194029850745,
"grad_norm": 0.46577740202461215,
"learning_rate": 3.549857555146056e-05,
"loss": 0.5472,
"step": 3560
},
{
"epoch": 0.41569496268656714,
"grad_norm": 0.4789932179858049,
"learning_rate": 3.5458012911785036e-05,
"loss": 0.5451,
"step": 3565
},
{
"epoch": 0.4162779850746269,
"grad_norm": 0.49654652817675987,
"learning_rate": 3.541742068865907e-05,
"loss": 0.5513,
"step": 3570
},
{
"epoch": 0.41686100746268656,
"grad_norm": 0.4692894243690009,
"learning_rate": 3.537679903298187e-05,
"loss": 0.5185,
"step": 3575
},
{
"epoch": 0.41744402985074625,
"grad_norm": 0.47941339153241536,
"learning_rate": 3.53361480957621e-05,
"loss": 0.5692,
"step": 3580
},
{
"epoch": 0.418027052238806,
"grad_norm": 0.46544865766435817,
"learning_rate": 3.529546802811725e-05,
"loss": 0.5405,
"step": 3585
},
{
"epoch": 0.41861007462686567,
"grad_norm": 0.46403588392065775,
"learning_rate": 3.5254758981273106e-05,
"loss": 0.5437,
"step": 3590
},
{
"epoch": 0.41919309701492535,
"grad_norm": 0.5179906749611947,
"learning_rate": 3.521402110656318e-05,
"loss": 0.5593,
"step": 3595
},
{
"epoch": 0.4197761194029851,
"grad_norm": 0.4781025455029735,
"learning_rate": 3.517325455542815e-05,
"loss": 0.5498,
"step": 3600
},
{
"epoch": 0.4203591417910448,
"grad_norm": 0.4934656687592932,
"learning_rate": 3.513245947941531e-05,
"loss": 0.5215,
"step": 3605
},
{
"epoch": 0.42094216417910446,
"grad_norm": 0.5326144345367541,
"learning_rate": 3.5091636030177995e-05,
"loss": 0.535,
"step": 3610
},
{
"epoch": 0.4215251865671642,
"grad_norm": 0.47801066829681754,
"learning_rate": 3.505078435947498e-05,
"loss": 0.5137,
"step": 3615
},
{
"epoch": 0.4221082089552239,
"grad_norm": 0.5286215614880156,
"learning_rate": 3.500990461916998e-05,
"loss": 0.5733,
"step": 3620
},
{
"epoch": 0.42269123134328357,
"grad_norm": 0.5073258841460547,
"learning_rate": 3.496899696123107e-05,
"loss": 0.5746,
"step": 3625
},
{
"epoch": 0.4232742537313433,
"grad_norm": 0.48126213143188057,
"learning_rate": 3.492806153773007e-05,
"loss": 0.559,
"step": 3630
},
{
"epoch": 0.423857276119403,
"grad_norm": 0.43727021104667335,
"learning_rate": 3.488709850084206e-05,
"loss": 0.4998,
"step": 3635
},
{
"epoch": 0.4244402985074627,
"grad_norm": 0.4897298712066961,
"learning_rate": 3.484610800284473e-05,
"loss": 0.5463,
"step": 3640
},
{
"epoch": 0.42502332089552236,
"grad_norm": 0.523356089677995,
"learning_rate": 3.480509019611788e-05,
"loss": 0.5659,
"step": 3645
},
{
"epoch": 0.4256063432835821,
"grad_norm": 0.5277720573146358,
"learning_rate": 3.476404523314282e-05,
"loss": 0.5241,
"step": 3650
},
{
"epoch": 0.4261893656716418,
"grad_norm": 0.526635621277189,
"learning_rate": 3.472297326650183e-05,
"loss": 0.5543,
"step": 3655
},
{
"epoch": 0.42677238805970147,
"grad_norm": 0.45989723897065593,
"learning_rate": 3.468187444887754e-05,
"loss": 0.4939,
"step": 3660
},
{
"epoch": 0.4273554104477612,
"grad_norm": 0.4622944623408051,
"learning_rate": 3.464074893305242e-05,
"loss": 0.5297,
"step": 3665
},
{
"epoch": 0.4279384328358209,
"grad_norm": 0.4783006627278798,
"learning_rate": 3.45995968719082e-05,
"loss": 0.5255,
"step": 3670
},
{
"epoch": 0.4285214552238806,
"grad_norm": 0.5059594374936409,
"learning_rate": 3.455841841842524e-05,
"loss": 0.5773,
"step": 3675
},
{
"epoch": 0.4291044776119403,
"grad_norm": 0.44346639279558187,
"learning_rate": 3.4517213725682085e-05,
"loss": 0.5196,
"step": 3680
},
{
"epoch": 0.4296875,
"grad_norm": 0.4742923141132099,
"learning_rate": 3.447598294685476e-05,
"loss": 0.5211,
"step": 3685
},
{
"epoch": 0.4302705223880597,
"grad_norm": 0.5096994312042691,
"learning_rate": 3.443472623521631e-05,
"loss": 0.5576,
"step": 3690
},
{
"epoch": 0.4308535447761194,
"grad_norm": 0.4757579513441239,
"learning_rate": 3.4393443744136136e-05,
"loss": 0.5342,
"step": 3695
},
{
"epoch": 0.4314365671641791,
"grad_norm": 0.4525723387319913,
"learning_rate": 3.435213562707953e-05,
"loss": 0.521,
"step": 3700
},
{
"epoch": 0.4320195895522388,
"grad_norm": 0.4655324458390628,
"learning_rate": 3.431080203760699e-05,
"loss": 0.5143,
"step": 3705
},
{
"epoch": 0.43260261194029853,
"grad_norm": 0.46493164741995624,
"learning_rate": 3.426944312937376e-05,
"loss": 0.5448,
"step": 3710
},
{
"epoch": 0.4331856343283582,
"grad_norm": 0.4705834455942846,
"learning_rate": 3.422805905612914e-05,
"loss": 0.5132,
"step": 3715
},
{
"epoch": 0.4337686567164179,
"grad_norm": 0.44735510242287085,
"learning_rate": 3.4186649971716044e-05,
"loss": 0.5078,
"step": 3720
},
{
"epoch": 0.43435167910447764,
"grad_norm": 0.4248977052828472,
"learning_rate": 3.4145216030070344e-05,
"loss": 0.5224,
"step": 3725
},
{
"epoch": 0.4349347014925373,
"grad_norm": 0.48361810889787654,
"learning_rate": 3.410375738522028e-05,
"loss": 0.5696,
"step": 3730
},
{
"epoch": 0.435517723880597,
"grad_norm": 0.5169029233248057,
"learning_rate": 3.406227419128596e-05,
"loss": 0.548,
"step": 3735
},
{
"epoch": 0.4361007462686567,
"grad_norm": 0.47352743387761187,
"learning_rate": 3.402076660247878e-05,
"loss": 0.5735,
"step": 3740
},
{
"epoch": 0.43668376865671643,
"grad_norm": 0.4598793206254637,
"learning_rate": 3.397923477310074e-05,
"loss": 0.5286,
"step": 3745
},
{
"epoch": 0.4372667910447761,
"grad_norm": 0.7329855818387587,
"learning_rate": 3.393767885754405e-05,
"loss": 0.5274,
"step": 3750
},
{
"epoch": 0.4378498134328358,
"grad_norm": 0.5374277209323949,
"learning_rate": 3.389609901029038e-05,
"loss": 0.569,
"step": 3755
},
{
"epoch": 0.43843283582089554,
"grad_norm": 0.4851132196571775,
"learning_rate": 3.38544953859104e-05,
"loss": 0.5658,
"step": 3760
},
{
"epoch": 0.4390158582089552,
"grad_norm": 0.4588569006808473,
"learning_rate": 3.381286813906317e-05,
"loss": 0.5463,
"step": 3765
},
{
"epoch": 0.4395988805970149,
"grad_norm": 0.49195337073823053,
"learning_rate": 3.3771217424495555e-05,
"loss": 0.5855,
"step": 3770
},
{
"epoch": 0.44018190298507465,
"grad_norm": 0.47991606249122304,
"learning_rate": 3.372954339704167e-05,
"loss": 0.5496,
"step": 3775
},
{
"epoch": 0.44076492537313433,
"grad_norm": 0.4690325622228386,
"learning_rate": 3.368784621162229e-05,
"loss": 0.5647,
"step": 3780
},
{
"epoch": 0.441347947761194,
"grad_norm": 0.5399827946397292,
"learning_rate": 3.364612602324429e-05,
"loss": 0.554,
"step": 3785
},
{
"epoch": 0.44193097014925375,
"grad_norm": 0.44956816879532197,
"learning_rate": 3.3604382987000016e-05,
"loss": 0.5033,
"step": 3790
},
{
"epoch": 0.44251399253731344,
"grad_norm": 0.49304141117786027,
"learning_rate": 3.356261725806681e-05,
"loss": 0.5651,
"step": 3795
},
{
"epoch": 0.4430970149253731,
"grad_norm": 0.4984710533841555,
"learning_rate": 3.352082899170631e-05,
"loss": 0.5238,
"step": 3800
},
{
"epoch": 0.44368003731343286,
"grad_norm": 0.5007565543332921,
"learning_rate": 3.3479018343264e-05,
"loss": 0.5519,
"step": 3805
},
{
"epoch": 0.44426305970149255,
"grad_norm": 0.5011462730265147,
"learning_rate": 3.343718546816852e-05,
"loss": 0.5523,
"step": 3810
},
{
"epoch": 0.44484608208955223,
"grad_norm": 0.49568146210324576,
"learning_rate": 3.339533052193114e-05,
"loss": 0.5132,
"step": 3815
},
{
"epoch": 0.4454291044776119,
"grad_norm": 0.45124156599655735,
"learning_rate": 3.335345366014522e-05,
"loss": 0.55,
"step": 3820
},
{
"epoch": 0.44601212686567165,
"grad_norm": 0.47223468640192207,
"learning_rate": 3.331155503848553e-05,
"loss": 0.5438,
"step": 3825
},
{
"epoch": 0.44659514925373134,
"grad_norm": 0.5287373716087295,
"learning_rate": 3.326963481270778e-05,
"loss": 0.5411,
"step": 3830
},
{
"epoch": 0.447178171641791,
"grad_norm": 0.44674599573765417,
"learning_rate": 3.322769313864796e-05,
"loss": 0.5363,
"step": 3835
},
{
"epoch": 0.44776119402985076,
"grad_norm": 0.4434254783508246,
"learning_rate": 3.3185730172221814e-05,
"loss": 0.537,
"step": 3840
},
{
"epoch": 0.44834421641791045,
"grad_norm": 0.5046794647102569,
"learning_rate": 3.3143746069424215e-05,
"loss": 0.5582,
"step": 3845
},
{
"epoch": 0.44892723880597013,
"grad_norm": 0.489005919425733,
"learning_rate": 3.310174098632865e-05,
"loss": 0.5549,
"step": 3850
},
{
"epoch": 0.44951026119402987,
"grad_norm": 0.49266732159431287,
"learning_rate": 3.305971507908655e-05,
"loss": 0.5313,
"step": 3855
},
{
"epoch": 0.45009328358208955,
"grad_norm": 0.5315237555379242,
"learning_rate": 3.301766850392681e-05,
"loss": 0.5581,
"step": 3860
},
{
"epoch": 0.45067630597014924,
"grad_norm": 0.46700902921771087,
"learning_rate": 3.29756014171551e-05,
"loss": 0.5371,
"step": 3865
},
{
"epoch": 0.451259328358209,
"grad_norm": 0.4395133061533244,
"learning_rate": 3.2933513975153384e-05,
"loss": 0.5399,
"step": 3870
},
{
"epoch": 0.45184235074626866,
"grad_norm": 0.485910728377908,
"learning_rate": 3.2891406334379285e-05,
"loss": 0.525,
"step": 3875
},
{
"epoch": 0.45242537313432835,
"grad_norm": 0.46551066433925664,
"learning_rate": 3.284927865136551e-05,
"loss": 0.4913,
"step": 3880
},
{
"epoch": 0.4530083955223881,
"grad_norm": 0.4792517888689515,
"learning_rate": 3.280713108271926e-05,
"loss": 0.5237,
"step": 3885
},
{
"epoch": 0.45359141791044777,
"grad_norm": 0.44580552095045006,
"learning_rate": 3.276496378512168e-05,
"loss": 0.5716,
"step": 3890
},
{
"epoch": 0.45417444029850745,
"grad_norm": 0.46092489328912967,
"learning_rate": 3.272277691532725e-05,
"loss": 0.5402,
"step": 3895
},
{
"epoch": 0.45475746268656714,
"grad_norm": 0.4413897932753195,
"learning_rate": 3.268057063016319e-05,
"loss": 0.5305,
"step": 3900
},
{
"epoch": 0.4553404850746269,
"grad_norm": 0.4782676785091174,
"learning_rate": 3.263834508652894e-05,
"loss": 0.5946,
"step": 3905
},
{
"epoch": 0.45592350746268656,
"grad_norm": 0.5238873675452039,
"learning_rate": 3.259610044139548e-05,
"loss": 0.5478,
"step": 3910
},
{
"epoch": 0.45650652985074625,
"grad_norm": 0.5182499166196407,
"learning_rate": 3.255383685180484e-05,
"loss": 0.5454,
"step": 3915
},
{
"epoch": 0.457089552238806,
"grad_norm": 0.47638810180727925,
"learning_rate": 3.251155447486945e-05,
"loss": 0.5446,
"step": 3920
},
{
"epoch": 0.45767257462686567,
"grad_norm": 0.4523229716036492,
"learning_rate": 3.246925346777158e-05,
"loss": 0.522,
"step": 3925
},
{
"epoch": 0.45825559701492535,
"grad_norm": 0.46909930899902835,
"learning_rate": 3.2426933987762785e-05,
"loss": 0.5393,
"step": 3930
},
{
"epoch": 0.4588386194029851,
"grad_norm": 0.43989917162010367,
"learning_rate": 3.238459619216326e-05,
"loss": 0.5211,
"step": 3935
},
{
"epoch": 0.4594216417910448,
"grad_norm": 0.48026394452289217,
"learning_rate": 3.23422402383613e-05,
"loss": 0.5547,
"step": 3940
},
{
"epoch": 0.46000466417910446,
"grad_norm": 0.48185947326085543,
"learning_rate": 3.22998662838127e-05,
"loss": 0.5041,
"step": 3945
},
{
"epoch": 0.4605876865671642,
"grad_norm": 0.4734018707747314,
"learning_rate": 3.2257474486040166e-05,
"loss": 0.5038,
"step": 3950
},
{
"epoch": 0.4611707089552239,
"grad_norm": 0.5075617095429634,
"learning_rate": 3.221506500263276e-05,
"loss": 0.5447,
"step": 3955
},
{
"epoch": 0.46175373134328357,
"grad_norm": 0.5046083432543891,
"learning_rate": 3.217263799124527e-05,
"loss": 0.5772,
"step": 3960
},
{
"epoch": 0.4623367537313433,
"grad_norm": 0.47069172866304554,
"learning_rate": 3.213019360959762e-05,
"loss": 0.5341,
"step": 3965
},
{
"epoch": 0.462919776119403,
"grad_norm": 0.47866635232292687,
"learning_rate": 3.2087732015474366e-05,
"loss": 0.5208,
"step": 3970
},
{
"epoch": 0.4635027985074627,
"grad_norm": 0.4516545061013375,
"learning_rate": 3.204525336672399e-05,
"loss": 0.5382,
"step": 3975
},
{
"epoch": 0.46408582089552236,
"grad_norm": 0.4987443516763455,
"learning_rate": 3.200275782125842e-05,
"loss": 0.5319,
"step": 3980
},
{
"epoch": 0.4646688432835821,
"grad_norm": 0.5260827157599703,
"learning_rate": 3.196024553705235e-05,
"loss": 0.5355,
"step": 3985
},
{
"epoch": 0.4652518656716418,
"grad_norm": 0.4668030990105745,
"learning_rate": 3.1917716672142746e-05,
"loss": 0.5057,
"step": 3990
},
{
"epoch": 0.46583488805970147,
"grad_norm": 0.4426983151393079,
"learning_rate": 3.187517138462819e-05,
"loss": 0.5254,
"step": 3995
},
{
"epoch": 0.4664179104477612,
"grad_norm": 0.4925988353031197,
"learning_rate": 3.1832609832668314e-05,
"loss": 0.5422,
"step": 4000
},
{
"epoch": 0.4670009328358209,
"grad_norm": 0.4560283471593519,
"learning_rate": 3.179003217448321e-05,
"loss": 0.5013,
"step": 4005
},
{
"epoch": 0.4675839552238806,
"grad_norm": 0.5244233872415475,
"learning_rate": 3.1747438568352844e-05,
"loss": 0.5736,
"step": 4010
},
{
"epoch": 0.4681669776119403,
"grad_norm": 0.5512150192012447,
"learning_rate": 3.170482917261648e-05,
"loss": 0.5688,
"step": 4015
},
{
"epoch": 0.46875,
"grad_norm": 0.5009381195241619,
"learning_rate": 3.166220414567206e-05,
"loss": 0.5468,
"step": 4020
},
{
"epoch": 0.4693330223880597,
"grad_norm": 0.47235216856933904,
"learning_rate": 3.161956364597566e-05,
"loss": 0.5318,
"step": 4025
},
{
"epoch": 0.4699160447761194,
"grad_norm": 0.5634765857129355,
"learning_rate": 3.1576907832040855e-05,
"loss": 0.5188,
"step": 4030
},
{
"epoch": 0.4704990671641791,
"grad_norm": 0.48072091370811076,
"learning_rate": 3.153423686243813e-05,
"loss": 0.5313,
"step": 4035
},
{
"epoch": 0.4710820895522388,
"grad_norm": 0.5120155067954245,
"learning_rate": 3.149155089579437e-05,
"loss": 0.5572,
"step": 4040
},
{
"epoch": 0.47166511194029853,
"grad_norm": 0.4812745114629162,
"learning_rate": 3.144885009079215e-05,
"loss": 0.5578,
"step": 4045
},
{
"epoch": 0.4722481343283582,
"grad_norm": 0.5058486137109877,
"learning_rate": 3.140613460616924e-05,
"loss": 0.5199,
"step": 4050
},
{
"epoch": 0.4728311567164179,
"grad_norm": 0.48002106325579913,
"learning_rate": 3.1363404600717965e-05,
"loss": 0.5659,
"step": 4055
},
{
"epoch": 0.47341417910447764,
"grad_norm": 0.48245091505853244,
"learning_rate": 3.132066023328465e-05,
"loss": 0.533,
"step": 4060
},
{
"epoch": 0.4739972014925373,
"grad_norm": 0.47536248109948226,
"learning_rate": 3.1277901662768983e-05,
"loss": 0.5433,
"step": 4065
},
{
"epoch": 0.474580223880597,
"grad_norm": 0.47200232932636466,
"learning_rate": 3.123512904812347e-05,
"loss": 0.5322,
"step": 4070
},
{
"epoch": 0.4751632462686567,
"grad_norm": 0.44740442637527467,
"learning_rate": 3.119234254835282e-05,
"loss": 0.5107,
"step": 4075
},
{
"epoch": 0.47574626865671643,
"grad_norm": 0.46154208469718466,
"learning_rate": 3.114954232251336e-05,
"loss": 0.527,
"step": 4080
},
{
"epoch": 0.4763292910447761,
"grad_norm": 0.5341929895695878,
"learning_rate": 3.110672852971243e-05,
"loss": 0.5689,
"step": 4085
},
{
"epoch": 0.4769123134328358,
"grad_norm": 0.4652305557977976,
"learning_rate": 3.1063901329107843e-05,
"loss": 0.5167,
"step": 4090
},
{
"epoch": 0.47749533582089554,
"grad_norm": 0.44417381469876005,
"learning_rate": 3.10210608799072e-05,
"loss": 0.5413,
"step": 4095
},
{
"epoch": 0.4780783582089552,
"grad_norm": 0.4507749342993046,
"learning_rate": 3.097820734136739e-05,
"loss": 0.5282,
"step": 4100
},
{
"epoch": 0.4786613805970149,
"grad_norm": 0.48297525105833106,
"learning_rate": 3.093534087279397e-05,
"loss": 0.5347,
"step": 4105
},
{
"epoch": 0.47924440298507465,
"grad_norm": 0.5100316904667305,
"learning_rate": 3.089246163354051e-05,
"loss": 0.5395,
"step": 4110
},
{
"epoch": 0.47982742537313433,
"grad_norm": 0.4995786265506999,
"learning_rate": 3.084956978300812e-05,
"loss": 0.5311,
"step": 4115
},
{
"epoch": 0.480410447761194,
"grad_norm": 0.4471832602242875,
"learning_rate": 3.080666548064475e-05,
"loss": 0.5193,
"step": 4120
},
{
"epoch": 0.48099347014925375,
"grad_norm": 0.46140397088213214,
"learning_rate": 3.076374888594464e-05,
"loss": 0.5345,
"step": 4125
},
{
"epoch": 0.48157649253731344,
"grad_norm": 0.44371711691404125,
"learning_rate": 3.0720820158447766e-05,
"loss": 0.5072,
"step": 4130
},
{
"epoch": 0.4821595149253731,
"grad_norm": 0.4561270995882789,
"learning_rate": 3.067787945773915e-05,
"loss": 0.5181,
"step": 4135
},
{
"epoch": 0.48274253731343286,
"grad_norm": 0.4468599316613869,
"learning_rate": 3.063492694344835e-05,
"loss": 0.5286,
"step": 4140
},
{
"epoch": 0.48332555970149255,
"grad_norm": 0.4921853335929309,
"learning_rate": 3.059196277524886e-05,
"loss": 0.5075,
"step": 4145
},
{
"epoch": 0.48390858208955223,
"grad_norm": 0.47049020704612843,
"learning_rate": 3.054898711285747e-05,
"loss": 0.5475,
"step": 4150
},
{
"epoch": 0.4844916044776119,
"grad_norm": 1.0060177708740357,
"learning_rate": 3.05060001160337e-05,
"loss": 0.5293,
"step": 4155
},
{
"epoch": 0.48507462686567165,
"grad_norm": 0.48784465637028546,
"learning_rate": 3.046300194457923e-05,
"loss": 0.534,
"step": 4160
},
{
"epoch": 0.48565764925373134,
"grad_norm": 0.4518879881598059,
"learning_rate": 3.0419992758337235e-05,
"loss": 0.5209,
"step": 4165
},
{
"epoch": 0.486240671641791,
"grad_norm": 0.4452974474644216,
"learning_rate": 3.0376972717191894e-05,
"loss": 0.5245,
"step": 4170
},
{
"epoch": 0.48682369402985076,
"grad_norm": 0.44833371464722693,
"learning_rate": 3.0333941981067688e-05,
"loss": 0.5156,
"step": 4175
},
{
"epoch": 0.48740671641791045,
"grad_norm": 0.44934997583495073,
"learning_rate": 3.029090070992889e-05,
"loss": 0.5144,
"step": 4180
},
{
"epoch": 0.48798973880597013,
"grad_norm": 0.47274610781733095,
"learning_rate": 3.0247849063778917e-05,
"loss": 0.5317,
"step": 4185
},
{
"epoch": 0.48857276119402987,
"grad_norm": 0.47590787280797764,
"learning_rate": 3.020478720265977e-05,
"loss": 0.51,
"step": 4190
},
{
"epoch": 0.48915578358208955,
"grad_norm": 0.44563102351746864,
"learning_rate": 3.01617152866514e-05,
"loss": 0.5301,
"step": 4195
},
{
"epoch": 0.48973880597014924,
"grad_norm": 0.4916861483726837,
"learning_rate": 3.0118633475871167e-05,
"loss": 0.5772,
"step": 4200
},
{
"epoch": 0.490321828358209,
"grad_norm": 0.4684037067244426,
"learning_rate": 3.0075541930473183e-05,
"loss": 0.4969,
"step": 4205
},
{
"epoch": 0.49090485074626866,
"grad_norm": 0.4797812232043466,
"learning_rate": 3.0032440810647783e-05,
"loss": 0.5038,
"step": 4210
},
{
"epoch": 0.49148787313432835,
"grad_norm": 0.4525162593921102,
"learning_rate": 2.998933027662086e-05,
"loss": 0.5266,
"step": 4215
},
{
"epoch": 0.4920708955223881,
"grad_norm": 0.4822829175013609,
"learning_rate": 2.9946210488653316e-05,
"loss": 0.5243,
"step": 4220
},
{
"epoch": 0.49265391791044777,
"grad_norm": 0.45841709338227005,
"learning_rate": 2.9903081607040473e-05,
"loss": 0.4948,
"step": 4225
},
{
"epoch": 0.49323694029850745,
"grad_norm": 0.4465507453121582,
"learning_rate": 2.9859943792111422e-05,
"loss": 0.5228,
"step": 4230
},
{
"epoch": 0.49381996268656714,
"grad_norm": 0.613185904602815,
"learning_rate": 2.9816797204228497e-05,
"loss": 0.5345,
"step": 4235
},
{
"epoch": 0.4944029850746269,
"grad_norm": 0.49396745715444307,
"learning_rate": 2.9773642003786627e-05,
"loss": 0.5412,
"step": 4240
},
{
"epoch": 0.49498600746268656,
"grad_norm": 0.5239548031082005,
"learning_rate": 2.9730478351212754e-05,
"loss": 0.5821,
"step": 4245
},
{
"epoch": 0.49556902985074625,
"grad_norm": 0.4221114790983556,
"learning_rate": 2.968730640696526e-05,
"loss": 0.5199,
"step": 4250
},
{
"epoch": 0.496152052238806,
"grad_norm": 0.480540611050451,
"learning_rate": 2.9644126331533328e-05,
"loss": 0.5149,
"step": 4255
},
{
"epoch": 0.49673507462686567,
"grad_norm": 0.4972848047633413,
"learning_rate": 2.9600938285436385e-05,
"loss": 0.5155,
"step": 4260
},
{
"epoch": 0.49731809701492535,
"grad_norm": 0.5100758433643319,
"learning_rate": 2.9557742429223478e-05,
"loss": 0.5495,
"step": 4265
},
{
"epoch": 0.4979011194029851,
"grad_norm": 0.46169645444430685,
"learning_rate": 2.95145389234727e-05,
"loss": 0.5432,
"step": 4270
},
{
"epoch": 0.4984841417910448,
"grad_norm": 0.478683421695544,
"learning_rate": 2.947132792879056e-05,
"loss": 0.5622,
"step": 4275
},
{
"epoch": 0.49906716417910446,
"grad_norm": 0.4937574145856353,
"learning_rate": 2.9428109605811427e-05,
"loss": 0.5292,
"step": 4280
},
{
"epoch": 0.4996501865671642,
"grad_norm": 0.4482984531417574,
"learning_rate": 2.9384884115196898e-05,
"loss": 0.5115,
"step": 4285
},
{
"epoch": 0.5002332089552238,
"grad_norm": 0.4581506836353169,
"learning_rate": 2.9341651617635236e-05,
"loss": 0.5369,
"step": 4290
},
{
"epoch": 0.5008162313432836,
"grad_norm": 0.4636121196389006,
"learning_rate": 2.929841227384072e-05,
"loss": 0.563,
"step": 4295
},
{
"epoch": 0.5013992537313433,
"grad_norm": 0.49303476266346524,
"learning_rate": 2.925516624455311e-05,
"loss": 0.5323,
"step": 4300
},
{
"epoch": 0.5019822761194029,
"grad_norm": 0.4792765198500162,
"learning_rate": 2.9211913690537003e-05,
"loss": 0.5238,
"step": 4305
},
{
"epoch": 0.5025652985074627,
"grad_norm": 0.47595384636734767,
"learning_rate": 2.9168654772581257e-05,
"loss": 0.5487,
"step": 4310
},
{
"epoch": 0.5031483208955224,
"grad_norm": 0.45475337905445856,
"learning_rate": 2.9125389651498374e-05,
"loss": 0.5551,
"step": 4315
},
{
"epoch": 0.503731343283582,
"grad_norm": 0.45368396432436275,
"learning_rate": 2.908211848812394e-05,
"loss": 0.5096,
"step": 4320
},
{
"epoch": 0.5043143656716418,
"grad_norm": 0.4720732964199861,
"learning_rate": 2.903884144331598e-05,
"loss": 0.5375,
"step": 4325
},
{
"epoch": 0.5048973880597015,
"grad_norm": 0.4609507788137596,
"learning_rate": 2.89955586779544e-05,
"loss": 0.5193,
"step": 4330
},
{
"epoch": 0.5054804104477612,
"grad_norm": 0.4437193884044816,
"learning_rate": 2.8952270352940362e-05,
"loss": 0.538,
"step": 4335
},
{
"epoch": 0.5060634328358209,
"grad_norm": 0.4686721143893824,
"learning_rate": 2.89089766291957e-05,
"loss": 0.5359,
"step": 4340
},
{
"epoch": 0.5066464552238806,
"grad_norm": 0.48522442554794026,
"learning_rate": 2.886567766766231e-05,
"loss": 0.5367,
"step": 4345
},
{
"epoch": 0.5072294776119403,
"grad_norm": 0.4682334708596764,
"learning_rate": 2.8822373629301573e-05,
"loss": 0.4979,
"step": 4350
},
{
"epoch": 0.5078125,
"grad_norm": 0.4461441009977717,
"learning_rate": 2.8779064675093724e-05,
"loss": 0.5473,
"step": 4355
},
{
"epoch": 0.5083955223880597,
"grad_norm": 0.523425795341376,
"learning_rate": 2.8735750966037295e-05,
"loss": 0.592,
"step": 4360
},
{
"epoch": 0.5089785447761194,
"grad_norm": 0.41208414865400855,
"learning_rate": 2.869243266314847e-05,
"loss": 0.5146,
"step": 4365
},
{
"epoch": 0.5095615671641791,
"grad_norm": 0.45512638703063274,
"learning_rate": 2.8649109927460533e-05,
"loss": 0.517,
"step": 4370
},
{
"epoch": 0.5101445895522388,
"grad_norm": 0.449015485615699,
"learning_rate": 2.8605782920023227e-05,
"loss": 0.5235,
"step": 4375
},
{
"epoch": 0.5107276119402985,
"grad_norm": 0.49156178005481366,
"learning_rate": 2.8562451801902197e-05,
"loss": 0.5308,
"step": 4380
},
{
"epoch": 0.5113106343283582,
"grad_norm": 0.4348710846671963,
"learning_rate": 2.8519116734178336e-05,
"loss": 0.5387,
"step": 4385
},
{
"epoch": 0.511893656716418,
"grad_norm": 0.4902492973827986,
"learning_rate": 2.8475777877947264e-05,
"loss": 0.5417,
"step": 4390
},
{
"epoch": 0.5124766791044776,
"grad_norm": 0.6547931036800456,
"learning_rate": 2.843243539431863e-05,
"loss": 0.5444,
"step": 4395
},
{
"epoch": 0.5130597014925373,
"grad_norm": 0.4876654982418252,
"learning_rate": 2.838908944441562e-05,
"loss": 0.5558,
"step": 4400
},
{
"epoch": 0.5136427238805971,
"grad_norm": 0.46099868761271035,
"learning_rate": 2.834574018937428e-05,
"loss": 0.5605,
"step": 4405
},
{
"epoch": 0.5142257462686567,
"grad_norm": 0.4545024437386275,
"learning_rate": 2.8302387790342943e-05,
"loss": 0.5175,
"step": 4410
},
{
"epoch": 0.5148087686567164,
"grad_norm": 0.4336617180340361,
"learning_rate": 2.8259032408481635e-05,
"loss": 0.5038,
"step": 4415
},
{
"epoch": 0.5153917910447762,
"grad_norm": 0.47838755970386043,
"learning_rate": 2.8215674204961462e-05,
"loss": 0.5594,
"step": 4420
},
{
"epoch": 0.5159748134328358,
"grad_norm": 0.5142813582002242,
"learning_rate": 2.817231334096403e-05,
"loss": 0.5177,
"step": 4425
},
{
"epoch": 0.5165578358208955,
"grad_norm": 0.44541441218084665,
"learning_rate": 2.812894997768083e-05,
"loss": 0.5465,
"step": 4430
},
{
"epoch": 0.5171408582089553,
"grad_norm": 0.4560004171104499,
"learning_rate": 2.8085584276312644e-05,
"loss": 0.5201,
"step": 4435
},
{
"epoch": 0.5177238805970149,
"grad_norm": 0.47964290891517664,
"learning_rate": 2.8042216398068942e-05,
"loss": 0.5247,
"step": 4440
},
{
"epoch": 0.5183069029850746,
"grad_norm": 0.8168752970003079,
"learning_rate": 2.7998846504167308e-05,
"loss": 0.5277,
"step": 4445
},
{
"epoch": 0.5188899253731343,
"grad_norm": 2.544656980904975,
"learning_rate": 2.7955474755832784e-05,
"loss": 0.505,
"step": 4450
},
{
"epoch": 0.519472947761194,
"grad_norm": 0.4659395065613913,
"learning_rate": 2.7912101314297327e-05,
"loss": 0.5269,
"step": 4455
},
{
"epoch": 0.5200559701492538,
"grad_norm": 2.6969407750363943,
"learning_rate": 2.7868726340799184e-05,
"loss": 0.5306,
"step": 4460
},
{
"epoch": 0.5206389925373134,
"grad_norm": 0.5270843633040919,
"learning_rate": 2.7825349996582313e-05,
"loss": 0.5711,
"step": 4465
},
{
"epoch": 0.5212220149253731,
"grad_norm": 0.5091777799722457,
"learning_rate": 2.7781972442895726e-05,
"loss": 0.5365,
"step": 4470
},
{
"epoch": 0.5218050373134329,
"grad_norm": 1.3533639882343287,
"learning_rate": 2.7738593840992975e-05,
"loss": 0.5539,
"step": 4475
},
{
"epoch": 0.5223880597014925,
"grad_norm": 0.4547594298673975,
"learning_rate": 2.769521435213149e-05,
"loss": 0.544,
"step": 4480
},
{
"epoch": 0.5229710820895522,
"grad_norm": 0.5510873748275549,
"learning_rate": 2.7651834137572003e-05,
"loss": 0.5353,
"step": 4485
},
{
"epoch": 0.523554104477612,
"grad_norm": 0.4328373857127542,
"learning_rate": 2.760845335857793e-05,
"loss": 0.5132,
"step": 4490
},
{
"epoch": 0.5241371268656716,
"grad_norm": 0.45053079578268007,
"learning_rate": 2.7565072176414803e-05,
"loss": 0.5231,
"step": 4495
},
{
"epoch": 0.5247201492537313,
"grad_norm": 0.49741815715434656,
"learning_rate": 2.7521690752349643e-05,
"loss": 0.5281,
"step": 4500
},
{
"epoch": 0.5253031716417911,
"grad_norm": 0.4607291755452952,
"learning_rate": 2.7478309247650362e-05,
"loss": 0.5558,
"step": 4505
},
{
"epoch": 0.5258861940298507,
"grad_norm": 0.5194917261722904,
"learning_rate": 2.7434927823585206e-05,
"loss": 0.539,
"step": 4510
},
{
"epoch": 0.5264692164179104,
"grad_norm": 0.45463126850370955,
"learning_rate": 2.739154664142208e-05,
"loss": 0.5215,
"step": 4515
},
{
"epoch": 0.5270522388059702,
"grad_norm": 0.4892107161079775,
"learning_rate": 2.7348165862428e-05,
"loss": 0.4891,
"step": 4520
},
{
"epoch": 0.5276352611940298,
"grad_norm": 0.5045457990411999,
"learning_rate": 2.7304785647868507e-05,
"loss": 0.5416,
"step": 4525
},
{
"epoch": 0.5282182835820896,
"grad_norm": 0.44346437150982526,
"learning_rate": 2.726140615900703e-05,
"loss": 0.5048,
"step": 4530
},
{
"epoch": 0.5288013059701493,
"grad_norm": 0.48666851854991494,
"learning_rate": 2.7218027557104286e-05,
"loss": 0.5034,
"step": 4535
},
{
"epoch": 0.5293843283582089,
"grad_norm": 0.4611162809005146,
"learning_rate": 2.7174650003417696e-05,
"loss": 0.5142,
"step": 4540
},
{
"epoch": 0.5299673507462687,
"grad_norm": 0.45514376590541833,
"learning_rate": 2.7131273659200818e-05,
"loss": 0.5162,
"step": 4545
},
{
"epoch": 0.5305503731343284,
"grad_norm": 0.48960501435122616,
"learning_rate": 2.7087898685702685e-05,
"loss": 0.5069,
"step": 4550
},
{
"epoch": 0.531133395522388,
"grad_norm": 0.4797862989191752,
"learning_rate": 2.704452524416722e-05,
"loss": 0.531,
"step": 4555
},
{
"epoch": 0.5317164179104478,
"grad_norm": 0.5893011284468495,
"learning_rate": 2.7001153495832697e-05,
"loss": 0.5061,
"step": 4560
},
{
"epoch": 0.5322994402985075,
"grad_norm": 0.4461810976300674,
"learning_rate": 2.6957783601931063e-05,
"loss": 0.5203,
"step": 4565
},
{
"epoch": 0.5328824626865671,
"grad_norm": 0.49982880523004874,
"learning_rate": 2.691441572368737e-05,
"loss": 0.5284,
"step": 4570
},
{
"epoch": 0.5334654850746269,
"grad_norm": 0.8194747676964631,
"learning_rate": 2.6871050022319177e-05,
"loss": 0.5642,
"step": 4575
},
{
"epoch": 0.5340485074626866,
"grad_norm": 0.45941020004877126,
"learning_rate": 2.6827686659035983e-05,
"loss": 0.5444,
"step": 4580
},
{
"epoch": 0.5346315298507462,
"grad_norm": 0.7401627959971693,
"learning_rate": 2.678432579503855e-05,
"loss": 0.5138,
"step": 4585
},
{
"epoch": 0.535214552238806,
"grad_norm": 0.49796442033400656,
"learning_rate": 2.6740967591518374e-05,
"loss": 0.5264,
"step": 4590
},
{
"epoch": 0.5357975746268657,
"grad_norm": 1.3853714172153995,
"learning_rate": 2.6697612209657063e-05,
"loss": 0.5375,
"step": 4595
},
{
"epoch": 0.5363805970149254,
"grad_norm": 0.4543560797621967,
"learning_rate": 2.665425981062573e-05,
"loss": 0.5155,
"step": 4600
},
{
"epoch": 0.5369636194029851,
"grad_norm": 0.484187959090983,
"learning_rate": 2.6610910555584384e-05,
"loss": 0.5478,
"step": 4605
},
{
"epoch": 0.5375466417910447,
"grad_norm": 0.49626488411866204,
"learning_rate": 2.6567564605681376e-05,
"loss": 0.5352,
"step": 4610
},
{
"epoch": 0.5381296641791045,
"grad_norm": 0.514181721385795,
"learning_rate": 2.652422212205275e-05,
"loss": 0.5387,
"step": 4615
},
{
"epoch": 0.5387126865671642,
"grad_norm": 0.5832049097179766,
"learning_rate": 2.6480883265821673e-05,
"loss": 0.5355,
"step": 4620
},
{
"epoch": 0.5392957089552238,
"grad_norm": 1.5120035406279293,
"learning_rate": 2.643754819809781e-05,
"loss": 0.5144,
"step": 4625
},
{
"epoch": 0.5398787313432836,
"grad_norm": 0.49157920203901545,
"learning_rate": 2.639421707997678e-05,
"loss": 0.5176,
"step": 4630
},
{
"epoch": 0.5404617537313433,
"grad_norm": 0.52902066701415,
"learning_rate": 2.6350890072539476e-05,
"loss": 0.5635,
"step": 4635
},
{
"epoch": 0.5410447761194029,
"grad_norm": 0.4637572082859708,
"learning_rate": 2.630756733685153e-05,
"loss": 0.5346,
"step": 4640
},
{
"epoch": 0.5416277985074627,
"grad_norm": 0.47595145214824397,
"learning_rate": 2.6264249033962713e-05,
"loss": 0.5249,
"step": 4645
},
{
"epoch": 0.5422108208955224,
"grad_norm": 1.6764787888943493,
"learning_rate": 2.622093532490628e-05,
"loss": 0.5064,
"step": 4650
},
{
"epoch": 0.542793843283582,
"grad_norm": 0.5788528459186878,
"learning_rate": 2.6177626370698443e-05,
"loss": 0.5051,
"step": 4655
},
{
"epoch": 0.5433768656716418,
"grad_norm": 0.48162500018331245,
"learning_rate": 2.6134322332337695e-05,
"loss": 0.5043,
"step": 4660
},
{
"epoch": 0.5439598880597015,
"grad_norm": 0.4737951941511267,
"learning_rate": 2.6091023370804307e-05,
"loss": 0.5142,
"step": 4665
},
{
"epoch": 0.5445429104477612,
"grad_norm": 0.5575644816288676,
"learning_rate": 2.604772964705965e-05,
"loss": 0.5505,
"step": 4670
},
{
"epoch": 0.5451259328358209,
"grad_norm": 0.4758372366326835,
"learning_rate": 2.6004441322045603e-05,
"loss": 0.5373,
"step": 4675
},
{
"epoch": 0.5457089552238806,
"grad_norm": 0.508958909313539,
"learning_rate": 2.596115855668403e-05,
"loss": 0.5353,
"step": 4680
},
{
"epoch": 0.5462919776119403,
"grad_norm": 0.4683786155826753,
"learning_rate": 2.5917881511876073e-05,
"loss": 0.4946,
"step": 4685
},
{
"epoch": 0.546875,
"grad_norm": 0.5456678942382134,
"learning_rate": 2.5874610348501632e-05,
"loss": 0.5371,
"step": 4690
},
{
"epoch": 0.5474580223880597,
"grad_norm": 0.4552682312653168,
"learning_rate": 2.5831345227418752e-05,
"loss": 0.5605,
"step": 4695
},
{
"epoch": 0.5480410447761194,
"grad_norm": 0.4776001785979188,
"learning_rate": 2.5788086309463006e-05,
"loss": 0.5005,
"step": 4700
},
{
"epoch": 0.5486240671641791,
"grad_norm": 0.5453229696717531,
"learning_rate": 2.57448337554469e-05,
"loss": 0.5194,
"step": 4705
},
{
"epoch": 0.5492070895522388,
"grad_norm": 0.5387817919914101,
"learning_rate": 2.570158772615928e-05,
"loss": 0.5333,
"step": 4710
},
{
"epoch": 0.5497901119402985,
"grad_norm": 0.48603015940285904,
"learning_rate": 2.5658348382364773e-05,
"loss": 0.5117,
"step": 4715
},
{
"epoch": 0.5503731343283582,
"grad_norm": 0.4789950555392135,
"learning_rate": 2.5615115884803108e-05,
"loss": 0.5168,
"step": 4720
},
{
"epoch": 0.550956156716418,
"grad_norm": 0.4593000342095894,
"learning_rate": 2.557189039418858e-05,
"loss": 0.5238,
"step": 4725
},
{
"epoch": 0.5515391791044776,
"grad_norm": 0.4671404581672369,
"learning_rate": 2.552867207120945e-05,
"loss": 0.5336,
"step": 4730
},
{
"epoch": 0.5521222014925373,
"grad_norm": 0.4867006683126969,
"learning_rate": 2.5485461076527308e-05,
"loss": 0.5432,
"step": 4735
},
{
"epoch": 0.5527052238805971,
"grad_norm": 0.4603754841175435,
"learning_rate": 2.5442257570776527e-05,
"loss": 0.5464,
"step": 4740
},
{
"epoch": 0.5532882462686567,
"grad_norm": 0.5322095390809611,
"learning_rate": 2.539906171456362e-05,
"loss": 0.5102,
"step": 4745
},
{
"epoch": 0.5538712686567164,
"grad_norm": 0.4501063054653902,
"learning_rate": 2.5355873668466677e-05,
"loss": 0.526,
"step": 4750
},
{
"epoch": 0.5544542910447762,
"grad_norm": 0.4645529985985766,
"learning_rate": 2.5312693593034746e-05,
"loss": 0.5035,
"step": 4755
},
{
"epoch": 0.5550373134328358,
"grad_norm": 0.4764710401107973,
"learning_rate": 2.5269521648787247e-05,
"loss": 0.5458,
"step": 4760
},
{
"epoch": 0.5556203358208955,
"grad_norm": 0.4534954950893274,
"learning_rate": 2.5226357996213378e-05,
"loss": 0.5199,
"step": 4765
},
{
"epoch": 0.5562033582089553,
"grad_norm": 0.4810706685920549,
"learning_rate": 2.518320279577151e-05,
"loss": 0.5263,
"step": 4770
},
{
"epoch": 0.5567863805970149,
"grad_norm": 0.4690472358261824,
"learning_rate": 2.514005620788858e-05,
"loss": 0.5245,
"step": 4775
},
{
"epoch": 0.5573694029850746,
"grad_norm": 0.46880508544260313,
"learning_rate": 2.5096918392959532e-05,
"loss": 0.5232,
"step": 4780
},
{
"epoch": 0.5579524253731343,
"grad_norm": 0.513925549118589,
"learning_rate": 2.5053789511346693e-05,
"loss": 0.5157,
"step": 4785
},
{
"epoch": 0.558535447761194,
"grad_norm": 0.5043455996761089,
"learning_rate": 2.5010669723379154e-05,
"loss": 0.5128,
"step": 4790
},
{
"epoch": 0.5591184701492538,
"grad_norm": 0.5924109641064016,
"learning_rate": 2.4967559189352226e-05,
"loss": 0.4921,
"step": 4795
},
{
"epoch": 0.5597014925373134,
"grad_norm": 0.47567528373039897,
"learning_rate": 2.492445806952682e-05,
"loss": 0.5406,
"step": 4800
},
{
"epoch": 0.5602845149253731,
"grad_norm": 0.4707447199114358,
"learning_rate": 2.4881366524128845e-05,
"loss": 0.5325,
"step": 4805
},
{
"epoch": 0.5608675373134329,
"grad_norm": 0.444463840534227,
"learning_rate": 2.4838284713348602e-05,
"loss": 0.4945,
"step": 4810
},
{
"epoch": 0.5614505597014925,
"grad_norm": 0.51359075031383,
"learning_rate": 2.479521279734024e-05,
"loss": 0.4986,
"step": 4815
},
{
"epoch": 0.5620335820895522,
"grad_norm": 0.508318043421021,
"learning_rate": 2.475215093622109e-05,
"loss": 0.5362,
"step": 4820
},
{
"epoch": 0.562616604477612,
"grad_norm": 0.4798125508084004,
"learning_rate": 2.4709099290071126e-05,
"loss": 0.498,
"step": 4825
},
{
"epoch": 0.5631996268656716,
"grad_norm": 0.46383793437769893,
"learning_rate": 2.4666058018932314e-05,
"loss": 0.5254,
"step": 4830
},
{
"epoch": 0.5637826492537313,
"grad_norm": 0.4825804953446005,
"learning_rate": 2.4623027282808114e-05,
"loss": 0.515,
"step": 4835
},
{
"epoch": 0.5643656716417911,
"grad_norm": 0.49356516542621953,
"learning_rate": 2.4580007241662773e-05,
"loss": 0.4913,
"step": 4840
},
{
"epoch": 0.5649486940298507,
"grad_norm": 0.46504399968658483,
"learning_rate": 2.4536998055420783e-05,
"loss": 0.5055,
"step": 4845
},
{
"epoch": 0.5655317164179104,
"grad_norm": 0.5178266867135491,
"learning_rate": 2.4493999883966308e-05,
"loss": 0.5338,
"step": 4850
},
{
"epoch": 0.5661147388059702,
"grad_norm": 0.443773396289914,
"learning_rate": 2.445101288714254e-05,
"loss": 0.527,
"step": 4855
},
{
"epoch": 0.5666977611940298,
"grad_norm": 0.4361179409054666,
"learning_rate": 2.440803722475114e-05,
"loss": 0.5115,
"step": 4860
},
{
"epoch": 0.5672807835820896,
"grad_norm": 0.4957031944579591,
"learning_rate": 2.436507305655165e-05,
"loss": 0.5224,
"step": 4865
},
{
"epoch": 0.5678638059701493,
"grad_norm": 0.5179909459677171,
"learning_rate": 2.4322120542260864e-05,
"loss": 0.5026,
"step": 4870
},
{
"epoch": 0.5684468283582089,
"grad_norm": 0.45931411628609564,
"learning_rate": 2.4279179841552246e-05,
"loss": 0.5492,
"step": 4875
},
{
"epoch": 0.5690298507462687,
"grad_norm": 0.4718292947061309,
"learning_rate": 2.4236251114055358e-05,
"loss": 0.5485,
"step": 4880
},
{
"epoch": 0.5696128731343284,
"grad_norm": 0.4509426009796579,
"learning_rate": 2.419333451935526e-05,
"loss": 0.5262,
"step": 4885
},
{
"epoch": 0.570195895522388,
"grad_norm": 0.520970250021266,
"learning_rate": 2.4150430216991888e-05,
"loss": 0.54,
"step": 4890
},
{
"epoch": 0.5707789179104478,
"grad_norm": 0.48125165415242044,
"learning_rate": 2.4107538366459494e-05,
"loss": 0.5535,
"step": 4895
},
{
"epoch": 0.5713619402985075,
"grad_norm": 0.49924612989919914,
"learning_rate": 2.406465912720604e-05,
"loss": 0.5378,
"step": 4900
},
{
"epoch": 0.5719449626865671,
"grad_norm": 0.4618022420362027,
"learning_rate": 2.4021792658632612e-05,
"loss": 0.5385,
"step": 4905
},
{
"epoch": 0.5725279850746269,
"grad_norm": 0.43524447366727087,
"learning_rate": 2.3978939120092814e-05,
"loss": 0.5275,
"step": 4910
},
{
"epoch": 0.5731110074626866,
"grad_norm": 0.5235831961960491,
"learning_rate": 2.3936098670892165e-05,
"loss": 0.5171,
"step": 4915
},
{
"epoch": 0.5736940298507462,
"grad_norm": 0.47253982117527504,
"learning_rate": 2.389327147028757e-05,
"loss": 0.5149,
"step": 4920
},
{
"epoch": 0.574277052238806,
"grad_norm": 0.536733235694404,
"learning_rate": 2.3850457677486655e-05,
"loss": 0.5385,
"step": 4925
},
{
"epoch": 0.5748600746268657,
"grad_norm": 0.49487264913586265,
"learning_rate": 2.380765745164718e-05,
"loss": 0.5453,
"step": 4930
},
{
"epoch": 0.5754430970149254,
"grad_norm": 0.4782990502193716,
"learning_rate": 2.376487095187654e-05,
"loss": 0.496,
"step": 4935
},
{
"epoch": 0.5760261194029851,
"grad_norm": 1.5309625007086856,
"learning_rate": 2.3722098337231025e-05,
"loss": 0.507,
"step": 4940
},
{
"epoch": 0.5766091417910447,
"grad_norm": 0.4155961423348815,
"learning_rate": 2.3679339766715358e-05,
"loss": 0.5219,
"step": 4945
},
{
"epoch": 0.5771921641791045,
"grad_norm": 0.4914382394629777,
"learning_rate": 2.363659539928204e-05,
"loss": 0.5107,
"step": 4950
},
{
"epoch": 0.5777751865671642,
"grad_norm": 0.4984691122251259,
"learning_rate": 2.3593865393830766e-05,
"loss": 0.5007,
"step": 4955
},
{
"epoch": 0.5783582089552238,
"grad_norm": 0.504442137117383,
"learning_rate": 2.355114990920786e-05,
"loss": 0.5103,
"step": 4960
},
{
"epoch": 0.5789412313432836,
"grad_norm": 0.774789963927665,
"learning_rate": 2.3508449104205636e-05,
"loss": 0.516,
"step": 4965
},
{
"epoch": 0.5795242537313433,
"grad_norm": 0.49669070623006295,
"learning_rate": 2.3465763137561875e-05,
"loss": 0.4984,
"step": 4970
},
{
"epoch": 0.5801072761194029,
"grad_norm": 0.4915699683156642,
"learning_rate": 2.342309216795916e-05,
"loss": 0.5237,
"step": 4975
},
{
"epoch": 0.5806902985074627,
"grad_norm": 0.7919278813140391,
"learning_rate": 2.3380436354024338e-05,
"loss": 0.5327,
"step": 4980
},
{
"epoch": 0.5812733208955224,
"grad_norm": 0.513367233603117,
"learning_rate": 2.333779585432794e-05,
"loss": 0.545,
"step": 4985
},
{
"epoch": 0.581856343283582,
"grad_norm": 0.5185055850159029,
"learning_rate": 2.329517082738353e-05,
"loss": 0.5671,
"step": 4990
},
{
"epoch": 0.5824393656716418,
"grad_norm": 0.4535336329617379,
"learning_rate": 2.3252561431647158e-05,
"loss": 0.5072,
"step": 4995
},
{
"epoch": 0.5830223880597015,
"grad_norm": 0.4214639260752155,
"learning_rate": 2.32099678255168e-05,
"loss": 0.4765,
"step": 5000
},
{
"epoch": 0.5836054104477612,
"grad_norm": 0.4292230083522436,
"learning_rate": 2.316739016733169e-05,
"loss": 0.535,
"step": 5005
},
{
"epoch": 0.5841884328358209,
"grad_norm": 0.4500847118246923,
"learning_rate": 2.3124828615371817e-05,
"loss": 0.5579,
"step": 5010
},
{
"epoch": 0.5847714552238806,
"grad_norm": 0.4441260617873408,
"learning_rate": 2.3082283327857253e-05,
"loss": 0.502,
"step": 5015
},
{
"epoch": 0.5853544776119403,
"grad_norm": 0.4929193442519856,
"learning_rate": 2.3039754462947653e-05,
"loss": 0.5154,
"step": 5020
},
{
"epoch": 0.5859375,
"grad_norm": 0.4604596804207028,
"learning_rate": 2.2997242178741596e-05,
"loss": 0.5173,
"step": 5025
},
{
"epoch": 0.5865205223880597,
"grad_norm": 0.4671301724369313,
"learning_rate": 2.2954746633276016e-05,
"loss": 0.5022,
"step": 5030
},
{
"epoch": 0.5871035447761194,
"grad_norm": 0.4666266550085784,
"learning_rate": 2.2912267984525643e-05,
"loss": 0.5263,
"step": 5035
},
{
"epoch": 0.5876865671641791,
"grad_norm": 0.5663717691966537,
"learning_rate": 2.2869806390402384e-05,
"loss": 0.5126,
"step": 5040
},
{
"epoch": 0.5882695895522388,
"grad_norm": 0.4709782228008594,
"learning_rate": 2.2827362008754743e-05,
"loss": 0.5374,
"step": 5045
},
{
"epoch": 0.5888526119402985,
"grad_norm": 0.523861503593284,
"learning_rate": 2.278493499736724e-05,
"loss": 0.5313,
"step": 5050
},
{
"epoch": 0.5894356343283582,
"grad_norm": 0.4797723190933745,
"learning_rate": 2.2742525513959832e-05,
"loss": 0.5382,
"step": 5055
},
{
"epoch": 0.590018656716418,
"grad_norm": 0.43860453059040094,
"learning_rate": 2.2700133716187316e-05,
"loss": 0.5265,
"step": 5060
},
{
"epoch": 0.5906016791044776,
"grad_norm": 0.4740764262851301,
"learning_rate": 2.2657759761638707e-05,
"loss": 0.5148,
"step": 5065
},
{
"epoch": 0.5911847014925373,
"grad_norm": 0.48957608183212625,
"learning_rate": 2.261540380783675e-05,
"loss": 0.4952,
"step": 5070
},
{
"epoch": 0.5917677238805971,
"grad_norm": 0.5162045189958194,
"learning_rate": 2.257306601223722e-05,
"loss": 0.5331,
"step": 5075
},
{
"epoch": 0.5923507462686567,
"grad_norm": 0.5728273214706431,
"learning_rate": 2.2530746532228413e-05,
"loss": 0.5143,
"step": 5080
},
{
"epoch": 0.5929337686567164,
"grad_norm": 0.44294658485783917,
"learning_rate": 2.2488445525130557e-05,
"loss": 0.5385,
"step": 5085
},
{
"epoch": 0.5935167910447762,
"grad_norm": 0.4728473251887733,
"learning_rate": 2.2446163148195164e-05,
"loss": 0.5249,
"step": 5090
},
{
"epoch": 0.5940998134328358,
"grad_norm": 0.42056687179230184,
"learning_rate": 2.2403899558604525e-05,
"loss": 0.4896,
"step": 5095
},
{
"epoch": 0.5946828358208955,
"grad_norm": 0.42312832455359684,
"learning_rate": 2.2361654913471065e-05,
"loss": 0.4831,
"step": 5100
},
{
"epoch": 0.5952658582089553,
"grad_norm": 0.4310382361089622,
"learning_rate": 2.2319429369836815e-05,
"loss": 0.5038,
"step": 5105
},
{
"epoch": 0.5958488805970149,
"grad_norm": 0.43349085099461926,
"learning_rate": 2.2277223084672765e-05,
"loss": 0.5332,
"step": 5110
},
{
"epoch": 0.5964319029850746,
"grad_norm": 0.4358686772263049,
"learning_rate": 2.2235036214878325e-05,
"loss": 0.4716,
"step": 5115
},
{
"epoch": 0.5970149253731343,
"grad_norm": 0.4998061706788218,
"learning_rate": 2.2192868917280745e-05,
"loss": 0.5277,
"step": 5120
},
{
"epoch": 0.597597947761194,
"grad_norm": 0.4569209342643407,
"learning_rate": 2.21507213486345e-05,
"loss": 0.4995,
"step": 5125
},
{
"epoch": 0.5981809701492538,
"grad_norm": 0.45297207736047584,
"learning_rate": 2.2108593665620724e-05,
"loss": 0.5254,
"step": 5130
},
{
"epoch": 0.5987639925373134,
"grad_norm": 0.49176995249756655,
"learning_rate": 2.2066486024846615e-05,
"loss": 0.5312,
"step": 5135
},
{
"epoch": 0.5993470149253731,
"grad_norm": 0.45352488174846867,
"learning_rate": 2.2024398582844906e-05,
"loss": 0.5183,
"step": 5140
},
{
"epoch": 0.5999300373134329,
"grad_norm": 0.4686989486366428,
"learning_rate": 2.19823314960732e-05,
"loss": 0.5337,
"step": 5145
},
{
"epoch": 0.6005130597014925,
"grad_norm": 0.4989471659468385,
"learning_rate": 2.1940284920913445e-05,
"loss": 0.513,
"step": 5150
},
{
"epoch": 0.6010960820895522,
"grad_norm": 0.7737032679878307,
"learning_rate": 2.1898259013671357e-05,
"loss": 0.4948,
"step": 5155
},
{
"epoch": 0.601679104477612,
"grad_norm": 0.6456584381963478,
"learning_rate": 2.1856253930575787e-05,
"loss": 0.5506,
"step": 5160
},
{
"epoch": 0.6022621268656716,
"grad_norm": 0.49420633212169496,
"learning_rate": 2.18142698277782e-05,
"loss": 0.534,
"step": 5165
},
{
"epoch": 0.6028451492537313,
"grad_norm": 0.43079700140392413,
"learning_rate": 2.1772306861352044e-05,
"loss": 0.4994,
"step": 5170
},
{
"epoch": 0.6034281716417911,
"grad_norm": 0.4477340131354866,
"learning_rate": 2.1730365187292228e-05,
"loss": 0.4895,
"step": 5175
},
{
"epoch": 0.6040111940298507,
"grad_norm": 0.45974056791460993,
"learning_rate": 2.1688444961514476e-05,
"loss": 0.4963,
"step": 5180
},
{
"epoch": 0.6045942164179104,
"grad_norm": 0.4463763393224647,
"learning_rate": 2.1646546339854788e-05,
"loss": 0.5139,
"step": 5185
},
{
"epoch": 0.6051772388059702,
"grad_norm": 0.5136717833408538,
"learning_rate": 2.1604669478068863e-05,
"loss": 0.5195,
"step": 5190
},
{
"epoch": 0.6057602611940298,
"grad_norm": 0.4554979479736964,
"learning_rate": 2.1562814531831487e-05,
"loss": 0.5062,
"step": 5195
},
{
"epoch": 0.6063432835820896,
"grad_norm": 0.45967404846816146,
"learning_rate": 2.1520981656736e-05,
"loss": 0.4984,
"step": 5200
},
{
"epoch": 0.6069263059701493,
"grad_norm": 0.44666913701763594,
"learning_rate": 2.1479171008293686e-05,
"loss": 0.4939,
"step": 5205
},
{
"epoch": 0.6075093283582089,
"grad_norm": 1.738573008880425,
"learning_rate": 2.1437382741933204e-05,
"loss": 0.542,
"step": 5210
},
{
"epoch": 0.6080923507462687,
"grad_norm": 0.45746766067753797,
"learning_rate": 2.1395617012999993e-05,
"loss": 0.5014,
"step": 5215
},
{
"epoch": 0.6086753731343284,
"grad_norm": 0.4542305089948559,
"learning_rate": 2.1353873976755716e-05,
"loss": 0.4852,
"step": 5220
},
{
"epoch": 0.609258395522388,
"grad_norm": 0.4734086638694241,
"learning_rate": 2.131215378837771e-05,
"loss": 0.5218,
"step": 5225
},
{
"epoch": 0.6098414179104478,
"grad_norm": 0.4738565708182124,
"learning_rate": 2.1270456602958332e-05,
"loss": 0.5322,
"step": 5230
},
{
"epoch": 0.6104244402985075,
"grad_norm": 0.49359869399689404,
"learning_rate": 2.1228782575504447e-05,
"loss": 0.537,
"step": 5235
},
{
"epoch": 0.6110074626865671,
"grad_norm": 0.47548423763939335,
"learning_rate": 2.1187131860936845e-05,
"loss": 0.5183,
"step": 5240
},
{
"epoch": 0.6115904850746269,
"grad_norm": 0.6813270131879138,
"learning_rate": 2.114550461408961e-05,
"loss": 0.5097,
"step": 5245
},
{
"epoch": 0.6121735074626866,
"grad_norm": 0.4858514966177657,
"learning_rate": 2.1103900989709623e-05,
"loss": 0.5267,
"step": 5250
},
{
"epoch": 0.6127565298507462,
"grad_norm": 0.46679943200730223,
"learning_rate": 2.1062321142455953e-05,
"loss": 0.5187,
"step": 5255
},
{
"epoch": 0.613339552238806,
"grad_norm": 0.43680321948998807,
"learning_rate": 2.1020765226899257e-05,
"loss": 0.5072,
"step": 5260
},
{
"epoch": 0.6139225746268657,
"grad_norm": 0.48765723203673156,
"learning_rate": 2.0979233397521237e-05,
"loss": 0.5172,
"step": 5265
},
{
"epoch": 0.6145055970149254,
"grad_norm": 0.5421523270806327,
"learning_rate": 2.0937725808714037e-05,
"loss": 0.5173,
"step": 5270
},
{
"epoch": 0.6150886194029851,
"grad_norm": 0.4821842656660482,
"learning_rate": 2.089624261477973e-05,
"loss": 0.522,
"step": 5275
},
{
"epoch": 0.6156716417910447,
"grad_norm": 0.473993764265411,
"learning_rate": 2.0854783969929668e-05,
"loss": 0.5394,
"step": 5280
},
{
"epoch": 0.6162546641791045,
"grad_norm": 0.4419301919043876,
"learning_rate": 2.0813350028283958e-05,
"loss": 0.5152,
"step": 5285
},
{
"epoch": 0.6168376865671642,
"grad_norm": 0.4672525694912507,
"learning_rate": 2.0771940943870866e-05,
"loss": 0.5195,
"step": 5290
},
{
"epoch": 0.6174207089552238,
"grad_norm": 0.4420478828415689,
"learning_rate": 2.073055687062625e-05,
"loss": 0.5131,
"step": 5295
},
{
"epoch": 0.6180037313432836,
"grad_norm": 0.4626362183961214,
"learning_rate": 2.0689197962393007e-05,
"loss": 0.5162,
"step": 5300
},
{
"epoch": 0.6185867537313433,
"grad_norm": 0.49784203270276733,
"learning_rate": 2.0647864372920472e-05,
"loss": 0.5279,
"step": 5305
},
{
"epoch": 0.6191697761194029,
"grad_norm": 0.454384681268438,
"learning_rate": 2.0606556255863862e-05,
"loss": 0.5193,
"step": 5310
},
{
"epoch": 0.6197527985074627,
"grad_norm": 0.549168830900785,
"learning_rate": 2.05652737647837e-05,
"loss": 0.508,
"step": 5315
},
{
"epoch": 0.6203358208955224,
"grad_norm": 0.4790027663801494,
"learning_rate": 2.0524017053145238e-05,
"loss": 0.5009,
"step": 5320
},
{
"epoch": 0.620918843283582,
"grad_norm": 0.4858133125717372,
"learning_rate": 2.0482786274317923e-05,
"loss": 0.5257,
"step": 5325
},
{
"epoch": 0.6215018656716418,
"grad_norm": 0.4686160974815818,
"learning_rate": 2.0441581581574765e-05,
"loss": 0.5006,
"step": 5330
},
{
"epoch": 0.6220848880597015,
"grad_norm": 0.4623666153609126,
"learning_rate": 2.0400403128091812e-05,
"loss": 0.5169,
"step": 5335
},
{
"epoch": 0.6226679104477612,
"grad_norm": 0.45877847406013955,
"learning_rate": 2.0359251066947583e-05,
"loss": 0.5334,
"step": 5340
},
{
"epoch": 0.6232509328358209,
"grad_norm": 0.4835869467134031,
"learning_rate": 2.0318125551122468e-05,
"loss": 0.4976,
"step": 5345
},
{
"epoch": 0.6238339552238806,
"grad_norm": 0.4575233875191603,
"learning_rate": 2.027702673349818e-05,
"loss": 0.5078,
"step": 5350
},
{
"epoch": 0.6244169776119403,
"grad_norm": 0.42998314282841227,
"learning_rate": 2.023595476685718e-05,
"loss": 0.5217,
"step": 5355
},
{
"epoch": 0.625,
"grad_norm": 0.40746461264097383,
"learning_rate": 2.0194909803882128e-05,
"loss": 0.4901,
"step": 5360
},
{
"epoch": 0.6255830223880597,
"grad_norm": 0.4597436174722755,
"learning_rate": 2.0153891997155282e-05,
"loss": 0.5494,
"step": 5365
},
{
"epoch": 0.6261660447761194,
"grad_norm": 0.47770020364059385,
"learning_rate": 2.011290149915795e-05,
"loss": 0.5137,
"step": 5370
},
{
"epoch": 0.6267490671641791,
"grad_norm": 0.4282249414588399,
"learning_rate": 2.0071938462269936e-05,
"loss": 0.4793,
"step": 5375
},
{
"epoch": 0.6273320895522388,
"grad_norm": 0.4509595314631354,
"learning_rate": 2.0031003038768942e-05,
"loss": 0.498,
"step": 5380
},
{
"epoch": 0.6279151119402985,
"grad_norm": 0.47714728709347026,
"learning_rate": 1.999009538083003e-05,
"loss": 0.545,
"step": 5385
},
{
"epoch": 0.6284981343283582,
"grad_norm": 0.47392502530376346,
"learning_rate": 1.994921564052503e-05,
"loss": 0.5133,
"step": 5390
},
{
"epoch": 0.629081156716418,
"grad_norm": 0.44759229419435465,
"learning_rate": 1.990836396982202e-05,
"loss": 0.5262,
"step": 5395
},
{
"epoch": 0.6296641791044776,
"grad_norm": 0.4912367043364836,
"learning_rate": 1.9867540520584693e-05,
"loss": 0.5346,
"step": 5400
},
{
"epoch": 0.6302472014925373,
"grad_norm": 0.4501000804982757,
"learning_rate": 1.9826745444571853e-05,
"loss": 0.5019,
"step": 5405
},
{
"epoch": 0.6308302238805971,
"grad_norm": 0.48836277098183645,
"learning_rate": 1.978597889343683e-05,
"loss": 0.5289,
"step": 5410
},
{
"epoch": 0.6314132462686567,
"grad_norm": 0.4625223996087579,
"learning_rate": 1.97452410187269e-05,
"loss": 0.4958,
"step": 5415
},
{
"epoch": 0.6319962686567164,
"grad_norm": 0.48092056548731926,
"learning_rate": 1.970453197188275e-05,
"loss": 0.5015,
"step": 5420
},
{
"epoch": 0.6325792910447762,
"grad_norm": 0.41113077299689604,
"learning_rate": 1.9663851904237903e-05,
"loss": 0.4989,
"step": 5425
},
{
"epoch": 0.6331623134328358,
"grad_norm": 0.43567184997194336,
"learning_rate": 1.9623200967018134e-05,
"loss": 0.5066,
"step": 5430
},
{
"epoch": 0.6337453358208955,
"grad_norm": 0.480598507617135,
"learning_rate": 1.9582579311340943e-05,
"loss": 0.5476,
"step": 5435
},
{
"epoch": 0.6343283582089553,
"grad_norm": 0.4632420890807915,
"learning_rate": 1.9541987088214963e-05,
"loss": 0.514,
"step": 5440
},
{
"epoch": 0.6349113805970149,
"grad_norm": 0.461886176881567,
"learning_rate": 1.9501424448539445e-05,
"loss": 0.4993,
"step": 5445
},
{
"epoch": 0.6354944029850746,
"grad_norm": 0.4863359836422005,
"learning_rate": 1.946089154310364e-05,
"loss": 0.5349,
"step": 5450
},
{
"epoch": 0.6360774253731343,
"grad_norm": 0.44233037285547316,
"learning_rate": 1.9420388522586242e-05,
"loss": 0.5036,
"step": 5455
},
{
"epoch": 0.636660447761194,
"grad_norm": 0.4399036507814764,
"learning_rate": 1.937991553755491e-05,
"loss": 0.5198,
"step": 5460
},
{
"epoch": 0.6372434701492538,
"grad_norm": 0.4644842288483705,
"learning_rate": 1.9339472738465604e-05,
"loss": 0.468,
"step": 5465
},
{
"epoch": 0.6378264925373134,
"grad_norm": 0.5021801630911482,
"learning_rate": 1.929906027566207e-05,
"loss": 0.5046,
"step": 5470
},
{
"epoch": 0.6384095149253731,
"grad_norm": 0.4906231561063412,
"learning_rate": 1.9258678299375287e-05,
"loss": 0.5637,
"step": 5475
},
{
"epoch": 0.6389925373134329,
"grad_norm": 0.4411840007547938,
"learning_rate": 1.9218326959722915e-05,
"loss": 0.4941,
"step": 5480
},
{
"epoch": 0.6395755597014925,
"grad_norm": 0.44492116681266264,
"learning_rate": 1.9178006406708716e-05,
"loss": 0.4999,
"step": 5485
},
{
"epoch": 0.6401585820895522,
"grad_norm": 0.45899402350391216,
"learning_rate": 1.913771679022199e-05,
"loss": 0.5143,
"step": 5490
},
{
"epoch": 0.640741604477612,
"grad_norm": 0.5002130360013761,
"learning_rate": 1.9097458260037055e-05,
"loss": 0.5223,
"step": 5495
},
{
"epoch": 0.6413246268656716,
"grad_norm": 0.4445252627269499,
"learning_rate": 1.9057230965812652e-05,
"loss": 0.5277,
"step": 5500
},
{
"epoch": 0.6419076492537313,
"grad_norm": 0.4327961966702548,
"learning_rate": 1.901703505709141e-05,
"loss": 0.4689,
"step": 5505
},
{
"epoch": 0.6424906716417911,
"grad_norm": 0.47514425543756855,
"learning_rate": 1.897687068329928e-05,
"loss": 0.499,
"step": 5510
},
{
"epoch": 0.6430736940298507,
"grad_norm": 0.4544747659785093,
"learning_rate": 1.8936737993744996e-05,
"loss": 0.5026,
"step": 5515
},
{
"epoch": 0.6436567164179104,
"grad_norm": 0.45829680386820026,
"learning_rate": 1.8896637137619495e-05,
"loss": 0.5066,
"step": 5520
},
{
"epoch": 0.6442397388059702,
"grad_norm": 0.42928587523633865,
"learning_rate": 1.8856568263995373e-05,
"loss": 0.4651,
"step": 5525
},
{
"epoch": 0.6448227611940298,
"grad_norm": 0.4785710795086254,
"learning_rate": 1.8816531521826346e-05,
"loss": 0.5118,
"step": 5530
},
{
"epoch": 0.6454057835820896,
"grad_norm": 0.44984484126810564,
"learning_rate": 1.8776527059946676e-05,
"loss": 0.4979,
"step": 5535
},
{
"epoch": 0.6459888059701493,
"grad_norm": 0.5084605177342773,
"learning_rate": 1.8736555027070607e-05,
"loss": 0.513,
"step": 5540
},
{
"epoch": 0.6465718283582089,
"grad_norm": 0.4587926020250396,
"learning_rate": 1.8696615571791876e-05,
"loss": 0.5056,
"step": 5545
},
{
"epoch": 0.6471548507462687,
"grad_norm": 0.5265935639558714,
"learning_rate": 1.865670884258307e-05,
"loss": 0.5328,
"step": 5550
},
{
"epoch": 0.6477378731343284,
"grad_norm": 0.48393501644987547,
"learning_rate": 1.861683498779514e-05,
"loss": 0.5409,
"step": 5555
},
{
"epoch": 0.648320895522388,
"grad_norm": 0.42845129776042723,
"learning_rate": 1.8576994155656814e-05,
"loss": 0.5036,
"step": 5560
},
{
"epoch": 0.6489039179104478,
"grad_norm": 0.45750371585274724,
"learning_rate": 1.853718649427409e-05,
"loss": 0.5458,
"step": 5565
},
{
"epoch": 0.6494869402985075,
"grad_norm": 0.5071290594881412,
"learning_rate": 1.8497412151629617e-05,
"loss": 0.5574,
"step": 5570
},
{
"epoch": 0.6500699626865671,
"grad_norm": 0.4193141376087725,
"learning_rate": 1.8457671275582202e-05,
"loss": 0.501,
"step": 5575
},
{
"epoch": 0.6506529850746269,
"grad_norm": 0.5097091316108154,
"learning_rate": 1.841796401386626e-05,
"loss": 0.5424,
"step": 5580
},
{
"epoch": 0.6512360074626866,
"grad_norm": 0.4574079642887348,
"learning_rate": 1.8378290514091214e-05,
"loss": 0.5143,
"step": 5585
},
{
"epoch": 0.6518190298507462,
"grad_norm": 0.48232820582938274,
"learning_rate": 1.8338650923740984e-05,
"loss": 0.5068,
"step": 5590
},
{
"epoch": 0.652402052238806,
"grad_norm": 0.4295823610973414,
"learning_rate": 1.829904539017347e-05,
"loss": 0.5076,
"step": 5595
},
{
"epoch": 0.6529850746268657,
"grad_norm": 0.4893118320191642,
"learning_rate": 1.8259474060619925e-05,
"loss": 0.5219,
"step": 5600
},
{
"epoch": 0.6535680970149254,
"grad_norm": 0.44539776731173597,
"learning_rate": 1.8219937082184462e-05,
"loss": 0.4935,
"step": 5605
},
{
"epoch": 0.6541511194029851,
"grad_norm": 0.5129311723129513,
"learning_rate": 1.8180434601843505e-05,
"loss": 0.5179,
"step": 5610
},
{
"epoch": 0.6547341417910447,
"grad_norm": 0.4864225115581459,
"learning_rate": 1.8140966766445235e-05,
"loss": 0.4969,
"step": 5615
},
{
"epoch": 0.6553171641791045,
"grad_norm": 0.47007224857589525,
"learning_rate": 1.8101533722709036e-05,
"loss": 0.5005,
"step": 5620
},
{
"epoch": 0.6559001865671642,
"grad_norm": 0.4597974424746557,
"learning_rate": 1.8062135617224933e-05,
"loss": 0.5294,
"step": 5625
},
{
"epoch": 0.6564832089552238,
"grad_norm": 0.49917347832714665,
"learning_rate": 1.802277259645313e-05,
"loss": 0.5261,
"step": 5630
},
{
"epoch": 0.6570662313432836,
"grad_norm": 0.46525260879023533,
"learning_rate": 1.798344480672334e-05,
"loss": 0.5119,
"step": 5635
},
{
"epoch": 0.6576492537313433,
"grad_norm": 0.49319911982873976,
"learning_rate": 1.7944152394234354e-05,
"loss": 0.4884,
"step": 5640
},
{
"epoch": 0.6582322761194029,
"grad_norm": 0.5132851806324479,
"learning_rate": 1.7904895505053405e-05,
"loss": 0.5158,
"step": 5645
},
{
"epoch": 0.6588152985074627,
"grad_norm": 0.47569949694794605,
"learning_rate": 1.7865674285115735e-05,
"loss": 0.5121,
"step": 5650
},
{
"epoch": 0.6593983208955224,
"grad_norm": 0.44552199263042797,
"learning_rate": 1.7826488880223913e-05,
"loss": 0.4761,
"step": 5655
},
{
"epoch": 0.659981343283582,
"grad_norm": 0.47720149730201705,
"learning_rate": 1.778733943604743e-05,
"loss": 0.5124,
"step": 5660
},
{
"epoch": 0.6605643656716418,
"grad_norm": 0.4432509370944976,
"learning_rate": 1.774822609812205e-05,
"loss": 0.4711,
"step": 5665
},
{
"epoch": 0.6611473880597015,
"grad_norm": 0.4450343141290362,
"learning_rate": 1.7709149011849364e-05,
"loss": 0.4761,
"step": 5670
},
{
"epoch": 0.6617304104477612,
"grad_norm": 0.4797782997827445,
"learning_rate": 1.767010832249613e-05,
"loss": 0.5175,
"step": 5675
},
{
"epoch": 0.6623134328358209,
"grad_norm": 0.47463931278850086,
"learning_rate": 1.7631104175193863e-05,
"loss": 0.5208,
"step": 5680
},
{
"epoch": 0.6628964552238806,
"grad_norm": 0.49271476148028914,
"learning_rate": 1.7592136714938206e-05,
"loss": 0.522,
"step": 5685
},
{
"epoch": 0.6634794776119403,
"grad_norm": 0.4366587132398326,
"learning_rate": 1.755320608658844e-05,
"loss": 0.4709,
"step": 5690
},
{
"epoch": 0.6640625,
"grad_norm": 0.44579614307494786,
"learning_rate": 1.7514312434866904e-05,
"loss": 0.5267,
"step": 5695
},
{
"epoch": 0.6646455223880597,
"grad_norm": 0.4487339036982587,
"learning_rate": 1.747545590435848e-05,
"loss": 0.4991,
"step": 5700
},
{
"epoch": 0.6652285447761194,
"grad_norm": 0.4383027709136693,
"learning_rate": 1.7436636639510082e-05,
"loss": 0.5141,
"step": 5705
},
{
"epoch": 0.6658115671641791,
"grad_norm": 0.4458649709251083,
"learning_rate": 1.739785478463004e-05,
"loss": 0.4921,
"step": 5710
},
{
"epoch": 0.6663945895522388,
"grad_norm": 0.49612990301681104,
"learning_rate": 1.735911048388768e-05,
"loss": 0.5081,
"step": 5715
},
{
"epoch": 0.6669776119402985,
"grad_norm": 0.4717384137501139,
"learning_rate": 1.7320403881312665e-05,
"loss": 0.4909,
"step": 5720
},
{
"epoch": 0.6675606343283582,
"grad_norm": 0.5029540121644192,
"learning_rate": 1.7281735120794555e-05,
"loss": 0.5439,
"step": 5725
},
{
"epoch": 0.668143656716418,
"grad_norm": 0.4643667414061285,
"learning_rate": 1.7243104346082194e-05,
"loss": 0.4794,
"step": 5730
},
{
"epoch": 0.6687266791044776,
"grad_norm": 0.47417180490760813,
"learning_rate": 1.720451170078328e-05,
"loss": 0.4996,
"step": 5735
},
{
"epoch": 0.6693097014925373,
"grad_norm": 0.4814261614829366,
"learning_rate": 1.7165957328363703e-05,
"loss": 0.5027,
"step": 5740
},
{
"epoch": 0.6698927238805971,
"grad_norm": 0.45627638279169563,
"learning_rate": 1.71274413721471e-05,
"loss": 0.513,
"step": 5745
},
{
"epoch": 0.6704757462686567,
"grad_norm": 0.4334633223796316,
"learning_rate": 1.708896397531431e-05,
"loss": 0.4659,
"step": 5750
},
{
"epoch": 0.6710587686567164,
"grad_norm": 0.4579206868255898,
"learning_rate": 1.7050525280902824e-05,
"loss": 0.5158,
"step": 5755
},
{
"epoch": 0.6716417910447762,
"grad_norm": 0.4386284661436938,
"learning_rate": 1.701212543180623e-05,
"loss": 0.5206,
"step": 5760
},
{
"epoch": 0.6722248134328358,
"grad_norm": 0.45478097582209165,
"learning_rate": 1.6973764570773766e-05,
"loss": 0.4958,
"step": 5765
},
{
"epoch": 0.6728078358208955,
"grad_norm": 0.49448818817184126,
"learning_rate": 1.693544284040968e-05,
"loss": 0.532,
"step": 5770
},
{
"epoch": 0.6733908582089553,
"grad_norm": 0.44392640275278356,
"learning_rate": 1.6897160383172794e-05,
"loss": 0.5102,
"step": 5775
},
{
"epoch": 0.6739738805970149,
"grad_norm": 0.4681803651785721,
"learning_rate": 1.6858917341375893e-05,
"loss": 0.5033,
"step": 5780
},
{
"epoch": 0.6745569029850746,
"grad_norm": 0.5012783536207814,
"learning_rate": 1.6820713857185296e-05,
"loss": 0.5261,
"step": 5785
},
{
"epoch": 0.6751399253731343,
"grad_norm": 0.5021117339189587,
"learning_rate": 1.6782550072620208e-05,
"loss": 0.4987,
"step": 5790
},
{
"epoch": 0.675722947761194,
"grad_norm": 0.48367815094839645,
"learning_rate": 1.674442612955229e-05,
"loss": 0.5231,
"step": 5795
},
{
"epoch": 0.6763059701492538,
"grad_norm": 0.42292867586088734,
"learning_rate": 1.6706342169705092e-05,
"loss": 0.4833,
"step": 5800
},
{
"epoch": 0.6768889925373134,
"grad_norm": 0.5498079373194198,
"learning_rate": 1.6668298334653504e-05,
"loss": 0.5303,
"step": 5805
},
{
"epoch": 0.6774720149253731,
"grad_norm": 0.45613249161600566,
"learning_rate": 1.663029476582328e-05,
"loss": 0.4946,
"step": 5810
},
{
"epoch": 0.6780550373134329,
"grad_norm": 0.4300584687753966,
"learning_rate": 1.659233160449048e-05,
"loss": 0.4835,
"step": 5815
},
{
"epoch": 0.6786380597014925,
"grad_norm": 0.47672064051122565,
"learning_rate": 1.6554408991780958e-05,
"loss": 0.5239,
"step": 5820
},
{
"epoch": 0.6792210820895522,
"grad_norm": 0.4363678330399839,
"learning_rate": 1.65165270686698e-05,
"loss": 0.5095,
"step": 5825
},
{
"epoch": 0.679804104477612,
"grad_norm": 0.4386976710048956,
"learning_rate": 1.6478685975980867e-05,
"loss": 0.5037,
"step": 5830
},
{
"epoch": 0.6803871268656716,
"grad_norm": 0.4713692579068845,
"learning_rate": 1.6440885854386223e-05,
"loss": 0.4919,
"step": 5835
},
{
"epoch": 0.6809701492537313,
"grad_norm": 0.4590294577914819,
"learning_rate": 1.6403126844405627e-05,
"loss": 0.5001,
"step": 5840
},
{
"epoch": 0.6815531716417911,
"grad_norm": 0.48010469655748705,
"learning_rate": 1.6365409086405982e-05,
"loss": 0.523,
"step": 5845
},
{
"epoch": 0.6821361940298507,
"grad_norm": 0.4572102553531434,
"learning_rate": 1.6327732720600893e-05,
"loss": 0.4981,
"step": 5850
},
{
"epoch": 0.6827192164179104,
"grad_norm": 0.42279317577956194,
"learning_rate": 1.6290097887050037e-05,
"loss": 0.4824,
"step": 5855
},
{
"epoch": 0.6833022388059702,
"grad_norm": 0.4672338547985673,
"learning_rate": 1.6252504725658738e-05,
"loss": 0.5276,
"step": 5860
},
{
"epoch": 0.6838852611940298,
"grad_norm": 0.5057683989918992,
"learning_rate": 1.6214953376177355e-05,
"loss": 0.5108,
"step": 5865
},
{
"epoch": 0.6844682835820896,
"grad_norm": 0.4285897123749069,
"learning_rate": 1.617744397820089e-05,
"loss": 0.4943,
"step": 5870
},
{
"epoch": 0.6850513059701493,
"grad_norm": 0.47280662245843785,
"learning_rate": 1.613997667116832e-05,
"loss": 0.5127,
"step": 5875
},
{
"epoch": 0.6856343283582089,
"grad_norm": 0.467214604689162,
"learning_rate": 1.610255159436219e-05,
"loss": 0.4953,
"step": 5880
},
{
"epoch": 0.6862173507462687,
"grad_norm": 0.4452219391184069,
"learning_rate": 1.6065168886908046e-05,
"loss": 0.4957,
"step": 5885
},
{
"epoch": 0.6868003731343284,
"grad_norm": 0.44301870809154087,
"learning_rate": 1.6027828687773947e-05,
"loss": 0.4975,
"step": 5890
},
{
"epoch": 0.687383395522388,
"grad_norm": 0.4401277833271277,
"learning_rate": 1.5990531135769885e-05,
"loss": 0.5195,
"step": 5895
},
{
"epoch": 0.6879664179104478,
"grad_norm": 2.0394489198103662,
"learning_rate": 1.5953276369547356e-05,
"loss": 0.4956,
"step": 5900
},
{
"epoch": 0.6885494402985075,
"grad_norm": 0.46836588645884025,
"learning_rate": 1.591606452759879e-05,
"loss": 0.5327,
"step": 5905
},
{
"epoch": 0.6891324626865671,
"grad_norm": 0.44598757398103644,
"learning_rate": 1.587889574825705e-05,
"loss": 0.4764,
"step": 5910
},
{
"epoch": 0.6897154850746269,
"grad_norm": 0.49742861746687556,
"learning_rate": 1.5841770169694895e-05,
"loss": 0.5111,
"step": 5915
},
{
"epoch": 0.6902985074626866,
"grad_norm": 0.47331933429323847,
"learning_rate": 1.5804687929924522e-05,
"loss": 0.5047,
"step": 5920
},
{
"epoch": 0.6908815298507462,
"grad_norm": 0.4652913554765252,
"learning_rate": 1.5767649166796995e-05,
"loss": 0.5038,
"step": 5925
},
{
"epoch": 0.691464552238806,
"grad_norm": 0.453898586144349,
"learning_rate": 1.573065401800176e-05,
"loss": 0.4755,
"step": 5930
},
{
"epoch": 0.6920475746268657,
"grad_norm": 0.43523535343423403,
"learning_rate": 1.569370262106615e-05,
"loss": 0.553,
"step": 5935
},
{
"epoch": 0.6926305970149254,
"grad_norm": 0.46238104794971113,
"learning_rate": 1.5656795113354816e-05,
"loss": 0.4496,
"step": 5940
},
{
"epoch": 0.6932136194029851,
"grad_norm": 0.4612926915904917,
"learning_rate": 1.5619931632069284e-05,
"loss": 0.5039,
"step": 5945
},
{
"epoch": 0.6937966417910447,
"grad_norm": 0.46826494703709465,
"learning_rate": 1.5583112314247386e-05,
"loss": 0.5022,
"step": 5950
},
{
"epoch": 0.6943796641791045,
"grad_norm": 0.47987543937588256,
"learning_rate": 1.5546337296762826e-05,
"loss": 0.5097,
"step": 5955
},
{
"epoch": 0.6949626865671642,
"grad_norm": 0.43334758409327906,
"learning_rate": 1.5509606716324563e-05,
"loss": 0.4943,
"step": 5960
},
{
"epoch": 0.6955457089552238,
"grad_norm": 0.4681763383562353,
"learning_rate": 1.547292070947641e-05,
"loss": 0.5063,
"step": 5965
},
{
"epoch": 0.6961287313432836,
"grad_norm": 0.4924802920273831,
"learning_rate": 1.5436279412596466e-05,
"loss": 0.5047,
"step": 5970
},
{
"epoch": 0.6967117537313433,
"grad_norm": 0.5347314807147506,
"learning_rate": 1.5399682961896627e-05,
"loss": 0.5254,
"step": 5975
},
{
"epoch": 0.6972947761194029,
"grad_norm": 0.46161765208167127,
"learning_rate": 1.5363131493422045e-05,
"loss": 0.5527,
"step": 5980
},
{
"epoch": 0.6978777985074627,
"grad_norm": 0.481592995924184,
"learning_rate": 1.5326625143050717e-05,
"loss": 0.5093,
"step": 5985
},
{
"epoch": 0.6984608208955224,
"grad_norm": 0.43698502229598946,
"learning_rate": 1.5290164046492855e-05,
"loss": 0.5113,
"step": 5990
},
{
"epoch": 0.699043843283582,
"grad_norm": 0.45737599000837936,
"learning_rate": 1.5253748339290478e-05,
"loss": 0.5072,
"step": 5995
},
{
"epoch": 0.6996268656716418,
"grad_norm": 0.4632305738785457,
"learning_rate": 1.5217378156816836e-05,
"loss": 0.5092,
"step": 6000
},
{
"epoch": 0.7002098880597015,
"grad_norm": 0.5658113543744521,
"learning_rate": 1.5181053634276005e-05,
"loss": 0.4714,
"step": 6005
},
{
"epoch": 0.7007929104477612,
"grad_norm": 0.5024081912461719,
"learning_rate": 1.5144774906702261e-05,
"loss": 0.5587,
"step": 6010
},
{
"epoch": 0.7013759328358209,
"grad_norm": 0.4346747521893763,
"learning_rate": 1.5108542108959666e-05,
"loss": 0.4874,
"step": 6015
},
{
"epoch": 0.7019589552238806,
"grad_norm": 0.48751850242768197,
"learning_rate": 1.5072355375741564e-05,
"loss": 0.5152,
"step": 6020
},
{
"epoch": 0.7025419776119403,
"grad_norm": 0.4679030016933106,
"learning_rate": 1.5036214841570002e-05,
"loss": 0.5177,
"step": 6025
},
{
"epoch": 0.703125,
"grad_norm": 0.43192813973658906,
"learning_rate": 1.500012064079533e-05,
"loss": 0.5381,
"step": 6030
},
{
"epoch": 0.7037080223880597,
"grad_norm": 0.47334729350720534,
"learning_rate": 1.4964072907595633e-05,
"loss": 0.5428,
"step": 6035
},
{
"epoch": 0.7042910447761194,
"grad_norm": 0.4625335667175621,
"learning_rate": 1.4928071775976283e-05,
"loss": 0.4965,
"step": 6040
},
{
"epoch": 0.7048740671641791,
"grad_norm": 0.4837863609737646,
"learning_rate": 1.489211737976937e-05,
"loss": 0.5201,
"step": 6045
},
{
"epoch": 0.7054570895522388,
"grad_norm": 0.45448831055589556,
"learning_rate": 1.4856209852633282e-05,
"loss": 0.4988,
"step": 6050
},
{
"epoch": 0.7060401119402985,
"grad_norm": 0.4470633313561567,
"learning_rate": 1.482034932805217e-05,
"loss": 0.4923,
"step": 6055
},
{
"epoch": 0.7066231343283582,
"grad_norm": 0.46146957236249214,
"learning_rate": 1.478453593933545e-05,
"loss": 0.4966,
"step": 6060
},
{
"epoch": 0.707206156716418,
"grad_norm": 0.44846726296477046,
"learning_rate": 1.4748769819617291e-05,
"loss": 0.5121,
"step": 6065
},
{
"epoch": 0.7077891791044776,
"grad_norm": 0.445769696660708,
"learning_rate": 1.47130511018562e-05,
"loss": 0.49,
"step": 6070
},
{
"epoch": 0.7083722014925373,
"grad_norm": 0.44424913913430536,
"learning_rate": 1.4677379918834408e-05,
"loss": 0.5154,
"step": 6075
},
{
"epoch": 0.7089552238805971,
"grad_norm": 0.5025893980492081,
"learning_rate": 1.464175640315748e-05,
"loss": 0.4886,
"step": 6080
},
{
"epoch": 0.7095382462686567,
"grad_norm": 0.4287175183865977,
"learning_rate": 1.460618068725374e-05,
"loss": 0.4943,
"step": 6085
},
{
"epoch": 0.7101212686567164,
"grad_norm": 0.46657717095673207,
"learning_rate": 1.4570652903373877e-05,
"loss": 0.4866,
"step": 6090
},
{
"epoch": 0.7107042910447762,
"grad_norm": 0.44658175711235154,
"learning_rate": 1.453517318359034e-05,
"loss": 0.4952,
"step": 6095
},
{
"epoch": 0.7112873134328358,
"grad_norm": 0.5109217950300224,
"learning_rate": 1.4499741659796927e-05,
"loss": 0.5495,
"step": 6100
},
{
"epoch": 0.7118703358208955,
"grad_norm": 0.439976204898742,
"learning_rate": 1.4464358463708277e-05,
"loss": 0.4947,
"step": 6105
},
{
"epoch": 0.7124533582089553,
"grad_norm": 0.442606078185802,
"learning_rate": 1.442902372685937e-05,
"loss": 0.5125,
"step": 6110
},
{
"epoch": 0.7130363805970149,
"grad_norm": 0.5113814435199087,
"learning_rate": 1.4393737580605019e-05,
"loss": 0.5397,
"step": 6115
},
{
"epoch": 0.7136194029850746,
"grad_norm": 0.4292248926969336,
"learning_rate": 1.435850015611943e-05,
"loss": 0.5038,
"step": 6120
},
{
"epoch": 0.7142024253731343,
"grad_norm": 0.4436015934650438,
"learning_rate": 1.432331158439568e-05,
"loss": 0.5015,
"step": 6125
},
{
"epoch": 0.714785447761194,
"grad_norm": 0.4679393946330053,
"learning_rate": 1.4288171996245247e-05,
"loss": 0.4961,
"step": 6130
},
{
"epoch": 0.7153684701492538,
"grad_norm": 0.46822261278058663,
"learning_rate": 1.425308152229749e-05,
"loss": 0.5565,
"step": 6135
},
{
"epoch": 0.7159514925373134,
"grad_norm": 0.48866887191385927,
"learning_rate": 1.4218040292999221e-05,
"loss": 0.531,
"step": 6140
},
{
"epoch": 0.7165345149253731,
"grad_norm": 0.4487435192453521,
"learning_rate": 1.4183048438614166e-05,
"loss": 0.4994,
"step": 6145
},
{
"epoch": 0.7171175373134329,
"grad_norm": 0.5036947554976395,
"learning_rate": 1.4148106089222513e-05,
"loss": 0.519,
"step": 6150
},
{
"epoch": 0.7177005597014925,
"grad_norm": 0.49765078956448927,
"learning_rate": 1.4113213374720425e-05,
"loss": 0.4825,
"step": 6155
},
{
"epoch": 0.7182835820895522,
"grad_norm": 0.49796549587751004,
"learning_rate": 1.4078370424819515e-05,
"loss": 0.5142,
"step": 6160
},
{
"epoch": 0.718866604477612,
"grad_norm": 0.46661776742591504,
"learning_rate": 1.404357736904645e-05,
"loss": 0.5121,
"step": 6165
},
{
"epoch": 0.7194496268656716,
"grad_norm": 0.4206361496368435,
"learning_rate": 1.4008834336742366e-05,
"loss": 0.5043,
"step": 6170
},
{
"epoch": 0.7200326492537313,
"grad_norm": 0.49140533816343374,
"learning_rate": 1.3974141457062498e-05,
"loss": 0.5465,
"step": 6175
},
{
"epoch": 0.7206156716417911,
"grad_norm": 0.4730841055221398,
"learning_rate": 1.3939498858975584e-05,
"loss": 0.496,
"step": 6180
},
{
"epoch": 0.7211986940298507,
"grad_norm": 0.5165044754977881,
"learning_rate": 1.390490667126348e-05,
"loss": 0.5699,
"step": 6185
},
{
"epoch": 0.7217817164179104,
"grad_norm": 0.4574833927556042,
"learning_rate": 1.3870365022520627e-05,
"loss": 0.494,
"step": 6190
},
{
"epoch": 0.7223647388059702,
"grad_norm": 0.4797480787975331,
"learning_rate": 1.3835874041153607e-05,
"loss": 0.5,
"step": 6195
},
{
"epoch": 0.7229477611940298,
"grad_norm": 0.5266171413115511,
"learning_rate": 1.380143385538063e-05,
"loss": 0.5248,
"step": 6200
},
{
"epoch": 0.7235307835820896,
"grad_norm": 0.4787651159553734,
"learning_rate": 1.3767044593231082e-05,
"loss": 0.4999,
"step": 6205
},
{
"epoch": 0.7241138059701493,
"grad_norm": 0.42275543868072146,
"learning_rate": 1.3732706382545054e-05,
"loss": 0.4731,
"step": 6210
},
{
"epoch": 0.7246968283582089,
"grad_norm": 0.48666619671292066,
"learning_rate": 1.3698419350972851e-05,
"loss": 0.5242,
"step": 6215
},
{
"epoch": 0.7252798507462687,
"grad_norm": 0.41898376072456794,
"learning_rate": 1.3664183625974503e-05,
"loss": 0.5104,
"step": 6220
},
{
"epoch": 0.7258628731343284,
"grad_norm": 0.4929104768550004,
"learning_rate": 1.362999933481935e-05,
"loss": 0.5206,
"step": 6225
},
{
"epoch": 0.726445895522388,
"grad_norm": 0.4308013069303933,
"learning_rate": 1.3595866604585492e-05,
"loss": 0.5074,
"step": 6230
},
{
"epoch": 0.7270289179104478,
"grad_norm": 0.4421936343319682,
"learning_rate": 1.3561785562159374e-05,
"loss": 0.4808,
"step": 6235
},
{
"epoch": 0.7276119402985075,
"grad_norm": 0.4980669606106666,
"learning_rate": 1.3527756334235288e-05,
"loss": 0.4746,
"step": 6240
},
{
"epoch": 0.7281949626865671,
"grad_norm": 0.4364859749653744,
"learning_rate": 1.3493779047314925e-05,
"loss": 0.4967,
"step": 6245
},
{
"epoch": 0.7287779850746269,
"grad_norm": 0.4424531855538569,
"learning_rate": 1.3459853827706853e-05,
"loss": 0.4962,
"step": 6250
},
{
"epoch": 0.7293610074626866,
"grad_norm": 0.5043638790699465,
"learning_rate": 1.3425980801526118e-05,
"loss": 0.5095,
"step": 6255
},
{
"epoch": 0.7299440298507462,
"grad_norm": 0.464384451619953,
"learning_rate": 1.3392160094693724e-05,
"loss": 0.5008,
"step": 6260
},
{
"epoch": 0.730527052238806,
"grad_norm": 0.44362217621259836,
"learning_rate": 1.3358391832936174e-05,
"loss": 0.4965,
"step": 6265
},
{
"epoch": 0.7311100746268657,
"grad_norm": 0.4092013778366071,
"learning_rate": 1.3324676141785029e-05,
"loss": 0.5133,
"step": 6270
},
{
"epoch": 0.7316930970149254,
"grad_norm": 0.4350575485273892,
"learning_rate": 1.3291013146576403e-05,
"loss": 0.5128,
"step": 6275
},
{
"epoch": 0.7322761194029851,
"grad_norm": 0.4744197561421012,
"learning_rate": 1.3257402972450539e-05,
"loss": 0.4784,
"step": 6280
},
{
"epoch": 0.7328591417910447,
"grad_norm": 0.4633376651924879,
"learning_rate": 1.3223845744351287e-05,
"loss": 0.475,
"step": 6285
},
{
"epoch": 0.7334421641791045,
"grad_norm": 0.47064248305550177,
"learning_rate": 1.3190341587025698e-05,
"loss": 0.5147,
"step": 6290
},
{
"epoch": 0.7340251865671642,
"grad_norm": 0.4745007087563231,
"learning_rate": 1.3156890625023532e-05,
"loss": 0.5131,
"step": 6295
},
{
"epoch": 0.7346082089552238,
"grad_norm": 0.47910451161364465,
"learning_rate": 1.3123492982696806e-05,
"loss": 0.5125,
"step": 6300
},
{
"epoch": 0.7351912313432836,
"grad_norm": 0.45150153399392984,
"learning_rate": 1.3090148784199288e-05,
"loss": 0.5195,
"step": 6305
},
{
"epoch": 0.7357742537313433,
"grad_norm": 0.4666852788990984,
"learning_rate": 1.305685815348613e-05,
"loss": 0.5348,
"step": 6310
},
{
"epoch": 0.7363572761194029,
"grad_norm": 0.43969109543694684,
"learning_rate": 1.3023621214313289e-05,
"loss": 0.4887,
"step": 6315
},
{
"epoch": 0.7369402985074627,
"grad_norm": 0.4520730263820748,
"learning_rate": 1.2990438090237167e-05,
"loss": 0.5174,
"step": 6320
},
{
"epoch": 0.7375233208955224,
"grad_norm": 0.4531016913171499,
"learning_rate": 1.2957308904614099e-05,
"loss": 0.513,
"step": 6325
},
{
"epoch": 0.738106343283582,
"grad_norm": 0.43990578576591227,
"learning_rate": 1.2924233780599915e-05,
"loss": 0.469,
"step": 6330
},
{
"epoch": 0.7386893656716418,
"grad_norm": 0.44084700165618407,
"learning_rate": 1.2891212841149447e-05,
"loss": 0.4997,
"step": 6335
},
{
"epoch": 0.7392723880597015,
"grad_norm": 0.49100928132633287,
"learning_rate": 1.2858246209016128e-05,
"loss": 0.5187,
"step": 6340
},
{
"epoch": 0.7398554104477612,
"grad_norm": 1.5928599989218044,
"learning_rate": 1.2825334006751493e-05,
"loss": 0.4954,
"step": 6345
},
{
"epoch": 0.7404384328358209,
"grad_norm": 0.4278160829593741,
"learning_rate": 1.2792476356704759e-05,
"loss": 0.462,
"step": 6350
},
{
"epoch": 0.7410214552238806,
"grad_norm": 0.45173765932520604,
"learning_rate": 1.2759673381022305e-05,
"loss": 0.5198,
"step": 6355
},
{
"epoch": 0.7416044776119403,
"grad_norm": 0.46060933457902786,
"learning_rate": 1.27269252016473e-05,
"loss": 0.501,
"step": 6360
},
{
"epoch": 0.7421875,
"grad_norm": 0.5194412989177508,
"learning_rate": 1.2694231940319192e-05,
"loss": 0.4862,
"step": 6365
},
{
"epoch": 0.7427705223880597,
"grad_norm": 0.46637710292865725,
"learning_rate": 1.2661593718573294e-05,
"loss": 0.4913,
"step": 6370
},
{
"epoch": 0.7433535447761194,
"grad_norm": 0.49083348251147924,
"learning_rate": 1.2629010657740275e-05,
"loss": 0.5073,
"step": 6375
},
{
"epoch": 0.7439365671641791,
"grad_norm": 0.476732025190355,
"learning_rate": 1.2596482878945787e-05,
"loss": 0.5343,
"step": 6380
},
{
"epoch": 0.7445195895522388,
"grad_norm": 0.43400279583829604,
"learning_rate": 1.2564010503109952e-05,
"loss": 0.4952,
"step": 6385
},
{
"epoch": 0.7451026119402985,
"grad_norm": 0.4474931990549133,
"learning_rate": 1.2531593650946932e-05,
"loss": 0.4966,
"step": 6390
},
{
"epoch": 0.7456856343283582,
"grad_norm": 0.4746293119307018,
"learning_rate": 1.2499232442964506e-05,
"loss": 0.5241,
"step": 6395
},
{
"epoch": 0.746268656716418,
"grad_norm": 0.4596957874880727,
"learning_rate": 1.2466926999463575e-05,
"loss": 0.4931,
"step": 6400
},
{
"epoch": 0.7468516791044776,
"grad_norm": 0.48590988974018173,
"learning_rate": 1.2434677440537745e-05,
"loss": 0.5498,
"step": 6405
},
{
"epoch": 0.7474347014925373,
"grad_norm": 0.43144811701119584,
"learning_rate": 1.2402483886072883e-05,
"loss": 0.4673,
"step": 6410
},
{
"epoch": 0.7480177238805971,
"grad_norm": 0.4694370814956435,
"learning_rate": 1.237034645574666e-05,
"loss": 0.5173,
"step": 6415
},
{
"epoch": 0.7486007462686567,
"grad_norm": 0.5215126950510659,
"learning_rate": 1.233826526902809e-05,
"loss": 0.5413,
"step": 6420
},
{
"epoch": 0.7491837686567164,
"grad_norm": 0.4667853644040054,
"learning_rate": 1.230624044517713e-05,
"loss": 0.5206,
"step": 6425
},
{
"epoch": 0.7497667910447762,
"grad_norm": 0.49822013134556736,
"learning_rate": 1.2274272103244201e-05,
"loss": 0.5264,
"step": 6430
},
{
"epoch": 0.7503498134328358,
"grad_norm": 0.4924144531446238,
"learning_rate": 1.2242360362069763e-05,
"loss": 0.4843,
"step": 6435
},
{
"epoch": 0.7509328358208955,
"grad_norm": 0.47214741828129897,
"learning_rate": 1.2210505340283838e-05,
"loss": 0.529,
"step": 6440
},
{
"epoch": 0.7515158582089553,
"grad_norm": 0.4726759175626485,
"learning_rate": 1.2178707156305644e-05,
"loss": 0.4993,
"step": 6445
},
{
"epoch": 0.7520988805970149,
"grad_norm": 0.47497735759010773,
"learning_rate": 1.2146965928343062e-05,
"loss": 0.4923,
"step": 6450
},
{
"epoch": 0.7526819029850746,
"grad_norm": 0.5282112383115118,
"learning_rate": 1.2115281774392278e-05,
"loss": 0.5043,
"step": 6455
},
{
"epoch": 0.7532649253731343,
"grad_norm": 0.4289024052469388,
"learning_rate": 1.208365481223727e-05,
"loss": 0.4915,
"step": 6460
},
{
"epoch": 0.753847947761194,
"grad_norm": 0.4527587796867484,
"learning_rate": 1.2052085159449455e-05,
"loss": 0.491,
"step": 6465
},
{
"epoch": 0.7544309701492538,
"grad_norm": 0.46413997296204734,
"learning_rate": 1.202057293338717e-05,
"loss": 0.5207,
"step": 6470
},
{
"epoch": 0.7550139925373134,
"grad_norm": 0.47431690969030404,
"learning_rate": 1.1989118251195284e-05,
"loss": 0.4807,
"step": 6475
},
{
"epoch": 0.7555970149253731,
"grad_norm": 0.6124168305417785,
"learning_rate": 1.1957721229804761e-05,
"loss": 0.4909,
"step": 6480
},
{
"epoch": 0.7561800373134329,
"grad_norm": 0.4612146149691418,
"learning_rate": 1.1926381985932186e-05,
"loss": 0.4912,
"step": 6485
},
{
"epoch": 0.7567630597014925,
"grad_norm": 0.4602032665320846,
"learning_rate": 1.1895100636079387e-05,
"loss": 0.5287,
"step": 6490
},
{
"epoch": 0.7573460820895522,
"grad_norm": 0.44850719377837156,
"learning_rate": 1.186387729653296e-05,
"loss": 0.5147,
"step": 6495
},
{
"epoch": 0.757929104477612,
"grad_norm": 0.43547527328367247,
"learning_rate": 1.1832712083363865e-05,
"loss": 0.4774,
"step": 6500
},
{
"epoch": 0.7585121268656716,
"grad_norm": 0.48648908094016996,
"learning_rate": 1.1801605112426953e-05,
"loss": 0.4948,
"step": 6505
},
{
"epoch": 0.7590951492537313,
"grad_norm": 0.4362070385676467,
"learning_rate": 1.1770556499360593e-05,
"loss": 0.4768,
"step": 6510
},
{
"epoch": 0.7596781716417911,
"grad_norm": 0.4594448702342764,
"learning_rate": 1.1739566359586195e-05,
"loss": 0.5192,
"step": 6515
},
{
"epoch": 0.7602611940298507,
"grad_norm": 0.4895001977535846,
"learning_rate": 1.170863480830781e-05,
"loss": 0.5029,
"step": 6520
},
{
"epoch": 0.7608442164179104,
"grad_norm": 0.4811285447973769,
"learning_rate": 1.167776196051166e-05,
"loss": 0.5089,
"step": 6525
},
{
"epoch": 0.7614272388059702,
"grad_norm": 0.4169504765893255,
"learning_rate": 1.1646947930965795e-05,
"loss": 0.4477,
"step": 6530
},
{
"epoch": 0.7620102611940298,
"grad_norm": 0.4342791005744433,
"learning_rate": 1.1616192834219553e-05,
"loss": 0.4957,
"step": 6535
},
{
"epoch": 0.7625932835820896,
"grad_norm": 0.500766952555154,
"learning_rate": 1.1585496784603234e-05,
"loss": 0.474,
"step": 6540
},
{
"epoch": 0.7631763059701493,
"grad_norm": 0.459834987252376,
"learning_rate": 1.15548598962276e-05,
"loss": 0.4653,
"step": 6545
},
{
"epoch": 0.7637593283582089,
"grad_norm": 0.4634667523018127,
"learning_rate": 1.1524282282983526e-05,
"loss": 0.4952,
"step": 6550
},
{
"epoch": 0.7643423507462687,
"grad_norm": 0.48306231366385766,
"learning_rate": 1.1493764058541493e-05,
"loss": 0.5092,
"step": 6555
},
{
"epoch": 0.7649253731343284,
"grad_norm": 0.4412886363099704,
"learning_rate": 1.1463305336351233e-05,
"loss": 0.4836,
"step": 6560
},
{
"epoch": 0.765508395522388,
"grad_norm": 0.4594326939876668,
"learning_rate": 1.143290622964128e-05,
"loss": 0.4858,
"step": 6565
},
{
"epoch": 0.7660914179104478,
"grad_norm": 0.495560886941702,
"learning_rate": 1.1402566851418545e-05,
"loss": 0.484,
"step": 6570
},
{
"epoch": 0.7666744402985075,
"grad_norm": 0.44698231009276274,
"learning_rate": 1.1372287314467896e-05,
"loss": 0.4938,
"step": 6575
},
{
"epoch": 0.7672574626865671,
"grad_norm": 0.48489845254460334,
"learning_rate": 1.1342067731351754e-05,
"loss": 0.5349,
"step": 6580
},
{
"epoch": 0.7678404850746269,
"grad_norm": 0.3983706507650223,
"learning_rate": 1.1311908214409666e-05,
"loss": 0.4916,
"step": 6585
},
{
"epoch": 0.7684235074626866,
"grad_norm": 0.4829203676168014,
"learning_rate": 1.128180887575789e-05,
"loss": 0.5538,
"step": 6590
},
{
"epoch": 0.7690065298507462,
"grad_norm": 0.5181252743781259,
"learning_rate": 1.1251769827288953e-05,
"loss": 0.5103,
"step": 6595
},
{
"epoch": 0.769589552238806,
"grad_norm": 0.45245249797253495,
"learning_rate": 1.122179118067128e-05,
"loss": 0.4877,
"step": 6600
},
{
"epoch": 0.7701725746268657,
"grad_norm": 0.45213952635934074,
"learning_rate": 1.1191873047348743e-05,
"loss": 0.4699,
"step": 6605
},
{
"epoch": 0.7707555970149254,
"grad_norm": 0.4384946771009236,
"learning_rate": 1.1162015538540268e-05,
"loss": 0.5199,
"step": 6610
},
{
"epoch": 0.7713386194029851,
"grad_norm": 1.4634272550071448,
"learning_rate": 1.1132218765239417e-05,
"loss": 0.5111,
"step": 6615
},
{
"epoch": 0.7719216417910447,
"grad_norm": 0.4528179441810264,
"learning_rate": 1.1102482838213945e-05,
"loss": 0.5095,
"step": 6620
},
{
"epoch": 0.7725046641791045,
"grad_norm": 0.48790948502665343,
"learning_rate": 1.1072807868005438e-05,
"loss": 0.5375,
"step": 6625
},
{
"epoch": 0.7730876865671642,
"grad_norm": 0.43356020787990573,
"learning_rate": 1.104319396492888e-05,
"loss": 0.494,
"step": 6630
},
{
"epoch": 0.7736707089552238,
"grad_norm": 0.46568426433428856,
"learning_rate": 1.1013641239072233e-05,
"loss": 0.5089,
"step": 6635
},
{
"epoch": 0.7742537313432836,
"grad_norm": 0.5058210347479575,
"learning_rate": 1.098414980029603e-05,
"loss": 0.5278,
"step": 6640
},
{
"epoch": 0.7748367537313433,
"grad_norm": 0.46834473900137596,
"learning_rate": 1.0954719758232983e-05,
"loss": 0.5183,
"step": 6645
},
{
"epoch": 0.7754197761194029,
"grad_norm": 0.4485355699073728,
"learning_rate": 1.092535122228757e-05,
"loss": 0.5067,
"step": 6650
},
{
"epoch": 0.7760027985074627,
"grad_norm": 0.44029418480334825,
"learning_rate": 1.0896044301635616e-05,
"loss": 0.4902,
"step": 6655
},
{
"epoch": 0.7765858208955224,
"grad_norm": 0.46043398790080686,
"learning_rate": 1.0866799105223877e-05,
"loss": 0.4752,
"step": 6660
},
{
"epoch": 0.777168843283582,
"grad_norm": 0.45411148412679725,
"learning_rate": 1.0837615741769695e-05,
"loss": 0.5027,
"step": 6665
},
{
"epoch": 0.7777518656716418,
"grad_norm": 0.46002535963451235,
"learning_rate": 1.0808494319760511e-05,
"loss": 0.4818,
"step": 6670
},
{
"epoch": 0.7783348880597015,
"grad_norm": 0.4613160618534238,
"learning_rate": 1.0779434947453531e-05,
"loss": 0.5305,
"step": 6675
},
{
"epoch": 0.7789179104477612,
"grad_norm": 0.5192961001346327,
"learning_rate": 1.0750437732875265e-05,
"loss": 0.4909,
"step": 6680
},
{
"epoch": 0.7795009328358209,
"grad_norm": 0.47703840671844866,
"learning_rate": 1.0721502783821194e-05,
"loss": 0.5433,
"step": 6685
},
{
"epoch": 0.7800839552238806,
"grad_norm": 0.4027202848278968,
"learning_rate": 1.0692630207855296e-05,
"loss": 0.4795,
"step": 6690
},
{
"epoch": 0.7806669776119403,
"grad_norm": 0.5254895186437003,
"learning_rate": 1.0663820112309695e-05,
"loss": 0.5234,
"step": 6695
},
{
"epoch": 0.78125,
"grad_norm": 0.4714359353636831,
"learning_rate": 1.0635072604284254e-05,
"loss": 0.4837,
"step": 6700
},
{
"epoch": 0.7818330223880597,
"grad_norm": 0.4409933416168129,
"learning_rate": 1.0606387790646154e-05,
"loss": 0.5124,
"step": 6705
},
{
"epoch": 0.7824160447761194,
"grad_norm": 0.47666045358791953,
"learning_rate": 1.0577765778029525e-05,
"loss": 0.4762,
"step": 6710
},
{
"epoch": 0.7829990671641791,
"grad_norm": 0.48287842316913426,
"learning_rate": 1.0549206672835033e-05,
"loss": 0.4879,
"step": 6715
},
{
"epoch": 0.7835820895522388,
"grad_norm": 0.4494356289008418,
"learning_rate": 1.0520710581229507e-05,
"loss": 0.4816,
"step": 6720
},
{
"epoch": 0.7841651119402985,
"grad_norm": 0.46052908698109446,
"learning_rate": 1.049227760914549e-05,
"loss": 0.4516,
"step": 6725
},
{
"epoch": 0.7847481343283582,
"grad_norm": 0.4567314964584747,
"learning_rate": 1.0463907862280913e-05,
"loss": 0.4871,
"step": 6730
},
{
"epoch": 0.785331156716418,
"grad_norm": 0.44868088537045625,
"learning_rate": 1.043560144609866e-05,
"loss": 0.4955,
"step": 6735
},
{
"epoch": 0.7859141791044776,
"grad_norm": 0.4530134493607295,
"learning_rate": 1.0407358465826198e-05,
"loss": 0.5199,
"step": 6740
},
{
"epoch": 0.7864972014925373,
"grad_norm": 0.4463551169201282,
"learning_rate": 1.0379179026455136e-05,
"loss": 0.4913,
"step": 6745
},
{
"epoch": 0.7870802238805971,
"grad_norm": 0.47099610185569996,
"learning_rate": 1.0351063232740937e-05,
"loss": 0.5332,
"step": 6750
},
{
"epoch": 0.7876632462686567,
"grad_norm": 0.4402447824147342,
"learning_rate": 1.0323011189202408e-05,
"loss": 0.5053,
"step": 6755
},
{
"epoch": 0.7882462686567164,
"grad_norm": 0.45593057954899197,
"learning_rate": 1.0295023000121404e-05,
"loss": 0.474,
"step": 6760
},
{
"epoch": 0.7888292910447762,
"grad_norm": 0.4374300903516831,
"learning_rate": 1.0267098769542368e-05,
"loss": 0.5427,
"step": 6765
},
{
"epoch": 0.7894123134328358,
"grad_norm": 0.4216054919323253,
"learning_rate": 1.0239238601272036e-05,
"loss": 0.4862,
"step": 6770
},
{
"epoch": 0.7899953358208955,
"grad_norm": 0.43838979602724326,
"learning_rate": 1.0211442598878936e-05,
"loss": 0.4697,
"step": 6775
},
{
"epoch": 0.7905783582089553,
"grad_norm": 0.41927389582266983,
"learning_rate": 1.0183710865693105e-05,
"loss": 0.4731,
"step": 6780
},
{
"epoch": 0.7911613805970149,
"grad_norm": 0.4618407627786652,
"learning_rate": 1.0156043504805648e-05,
"loss": 0.4946,
"step": 6785
},
{
"epoch": 0.7917444029850746,
"grad_norm": 0.49476862297527385,
"learning_rate": 1.0128440619068379e-05,
"loss": 0.5218,
"step": 6790
},
{
"epoch": 0.7923274253731343,
"grad_norm": 0.4878473893991779,
"learning_rate": 1.0100902311093405e-05,
"loss": 0.5127,
"step": 6795
},
{
"epoch": 0.792910447761194,
"grad_norm": 0.44555429187543133,
"learning_rate": 1.0073428683252788e-05,
"loss": 0.4893,
"step": 6800
},
{
"epoch": 0.7934934701492538,
"grad_norm": 0.44977845000528854,
"learning_rate": 1.0046019837678153e-05,
"loss": 0.4687,
"step": 6805
},
{
"epoch": 0.7940764925373134,
"grad_norm": 0.447310487987635,
"learning_rate": 1.001867587626029e-05,
"loss": 0.484,
"step": 6810
},
{
"epoch": 0.7946595149253731,
"grad_norm": 0.48144807868357464,
"learning_rate": 9.991396900648774e-06,
"loss": 0.5395,
"step": 6815
},
{
"epoch": 0.7952425373134329,
"grad_norm": 0.46182249334458747,
"learning_rate": 9.964183012251619e-06,
"loss": 0.4914,
"step": 6820
},
{
"epoch": 0.7958255597014925,
"grad_norm": 0.4389736266429193,
"learning_rate": 9.937034312234872e-06,
"loss": 0.4966,
"step": 6825
},
{
"epoch": 0.7964085820895522,
"grad_norm": 0.4683635632125285,
"learning_rate": 9.90995090152225e-06,
"loss": 0.4872,
"step": 6830
},
{
"epoch": 0.796991604477612,
"grad_norm": 0.449577983637824,
"learning_rate": 9.88293288079476e-06,
"loss": 0.4775,
"step": 6835
},
{
"epoch": 0.7975746268656716,
"grad_norm": 0.4679257203680657,
"learning_rate": 9.855980350490315e-06,
"loss": 0.4628,
"step": 6840
},
{
"epoch": 0.7981576492537313,
"grad_norm": 0.4910686168202571,
"learning_rate": 9.82909341080339e-06,
"loss": 0.516,
"step": 6845
},
{
"epoch": 0.7987406716417911,
"grad_norm": 0.4328561795827866,
"learning_rate": 9.802272161684601e-06,
"loss": 0.5062,
"step": 6850
},
{
"epoch": 0.7993236940298507,
"grad_norm": 0.48810830150277185,
"learning_rate": 9.775516702840411e-06,
"loss": 0.542,
"step": 6855
},
{
"epoch": 0.7999067164179104,
"grad_norm": 0.4326359263896414,
"learning_rate": 9.748827133732665e-06,
"loss": 0.491,
"step": 6860
},
{
"epoch": 0.8004897388059702,
"grad_norm": 0.45895357662154007,
"learning_rate": 9.722203553578288e-06,
"loss": 0.5017,
"step": 6865
},
{
"epoch": 0.8010727611940298,
"grad_norm": 0.4201551976753478,
"learning_rate": 9.695646061348892e-06,
"loss": 0.5233,
"step": 6870
},
{
"epoch": 0.8016557835820896,
"grad_norm": 0.4261865295483104,
"learning_rate": 9.669154755770415e-06,
"loss": 0.4615,
"step": 6875
},
{
"epoch": 0.8022388059701493,
"grad_norm": 0.7140092170328386,
"learning_rate": 9.642729735322733e-06,
"loss": 0.5097,
"step": 6880
},
{
"epoch": 0.8028218283582089,
"grad_norm": 0.4990283440949626,
"learning_rate": 9.616371098239346e-06,
"loss": 0.5716,
"step": 6885
},
{
"epoch": 0.8034048507462687,
"grad_norm": 0.4481559993333328,
"learning_rate": 9.590078942506933e-06,
"loss": 0.4856,
"step": 6890
},
{
"epoch": 0.8039878731343284,
"grad_norm": 0.4607158596942868,
"learning_rate": 9.56385336586507e-06,
"loss": 0.5107,
"step": 6895
},
{
"epoch": 0.804570895522388,
"grad_norm": 0.476387734456041,
"learning_rate": 9.537694465805797e-06,
"loss": 0.4998,
"step": 6900
},
{
"epoch": 0.8051539179104478,
"grad_norm": 0.46770623541362477,
"learning_rate": 9.511602339573324e-06,
"loss": 0.4998,
"step": 6905
},
{
"epoch": 0.8057369402985075,
"grad_norm": 0.5086514392993742,
"learning_rate": 9.485577084163604e-06,
"loss": 0.5105,
"step": 6910
},
{
"epoch": 0.8063199626865671,
"grad_norm": 0.5077908007862963,
"learning_rate": 9.45961879632401e-06,
"loss": 0.5116,
"step": 6915
},
{
"epoch": 0.8069029850746269,
"grad_norm": 0.5463748621976762,
"learning_rate": 9.43372757255297e-06,
"loss": 0.4687,
"step": 6920
},
{
"epoch": 0.8074860074626866,
"grad_norm": 0.4191317799403366,
"learning_rate": 9.40790350909961e-06,
"loss": 0.498,
"step": 6925
},
{
"epoch": 0.8080690298507462,
"grad_norm": 0.4865303702117012,
"learning_rate": 9.382146701963373e-06,
"loss": 0.456,
"step": 6930
},
{
"epoch": 0.808652052238806,
"grad_norm": 0.42674988798148156,
"learning_rate": 9.356457246893695e-06,
"loss": 0.5227,
"step": 6935
},
{
"epoch": 0.8092350746268657,
"grad_norm": 0.46859664111243005,
"learning_rate": 9.330835239389645e-06,
"loss": 0.5018,
"step": 6940
},
{
"epoch": 0.8098180970149254,
"grad_norm": 0.5581605519629456,
"learning_rate": 9.305280774699531e-06,
"loss": 0.4893,
"step": 6945
},
{
"epoch": 0.8104011194029851,
"grad_norm": 0.4648844812652044,
"learning_rate": 9.279793947820596e-06,
"loss": 0.5034,
"step": 6950
},
{
"epoch": 0.8109841417910447,
"grad_norm": 0.4348511386801257,
"learning_rate": 9.254374853498636e-06,
"loss": 0.476,
"step": 6955
},
{
"epoch": 0.8115671641791045,
"grad_norm": 0.4471748979975722,
"learning_rate": 9.229023586227666e-06,
"loss": 0.4746,
"step": 6960
},
{
"epoch": 0.8121501865671642,
"grad_norm": 0.49028467746657334,
"learning_rate": 9.203740240249527e-06,
"loss": 0.5075,
"step": 6965
},
{
"epoch": 0.8127332089552238,
"grad_norm": 0.45156139381243476,
"learning_rate": 9.178524909553617e-06,
"loss": 0.4997,
"step": 6970
},
{
"epoch": 0.8133162313432836,
"grad_norm": 0.4580984480401016,
"learning_rate": 9.153377687876439e-06,
"loss": 0.5098,
"step": 6975
},
{
"epoch": 0.8138992537313433,
"grad_norm": 0.4417064208682211,
"learning_rate": 9.128298668701341e-06,
"loss": 0.4866,
"step": 6980
},
{
"epoch": 0.8144822761194029,
"grad_norm": 0.4396784431419953,
"learning_rate": 9.103287945258104e-06,
"loss": 0.4843,
"step": 6985
},
{
"epoch": 0.8150652985074627,
"grad_norm": 0.42781740173572586,
"learning_rate": 9.078345610522662e-06,
"loss": 0.4864,
"step": 6990
},
{
"epoch": 0.8156483208955224,
"grad_norm": 0.4706194313899779,
"learning_rate": 9.053471757216675e-06,
"loss": 0.4829,
"step": 6995
},
{
"epoch": 0.816231343283582,
"grad_norm": 0.4373005186110652,
"learning_rate": 9.028666477807253e-06,
"loss": 0.4946,
"step": 7000
},
{
"epoch": 0.8168143656716418,
"grad_norm": 0.45822848787961307,
"learning_rate": 9.003929864506583e-06,
"loss": 0.4747,
"step": 7005
},
{
"epoch": 0.8173973880597015,
"grad_norm": 0.4376578697732612,
"learning_rate": 8.979262009271589e-06,
"loss": 0.4982,
"step": 7010
},
{
"epoch": 0.8179804104477612,
"grad_norm": 0.50115571632827,
"learning_rate": 8.954663003803579e-06,
"loss": 0.4934,
"step": 7015
},
{
"epoch": 0.8185634328358209,
"grad_norm": 0.44219752315988364,
"learning_rate": 8.930132939547932e-06,
"loss": 0.4663,
"step": 7020
},
{
"epoch": 0.8191464552238806,
"grad_norm": 0.440108685069607,
"learning_rate": 8.905671907693738e-06,
"loss": 0.4856,
"step": 7025
},
{
"epoch": 0.8197294776119403,
"grad_norm": 0.5246570100006753,
"learning_rate": 8.881279999173466e-06,
"loss": 0.5021,
"step": 7030
},
{
"epoch": 0.8203125,
"grad_norm": 0.4319239074767847,
"learning_rate": 8.856957304662602e-06,
"loss": 0.5123,
"step": 7035
},
{
"epoch": 0.8208955223880597,
"grad_norm": 0.4326829341676435,
"learning_rate": 8.832703914579363e-06,
"loss": 0.5021,
"step": 7040
},
{
"epoch": 0.8214785447761194,
"grad_norm": 0.4380817083829351,
"learning_rate": 8.80851991908431e-06,
"loss": 0.5044,
"step": 7045
},
{
"epoch": 0.8220615671641791,
"grad_norm": 0.4637369898587162,
"learning_rate": 8.784405408080046e-06,
"loss": 0.4852,
"step": 7050
},
{
"epoch": 0.8226445895522388,
"grad_norm": 0.4796910879347511,
"learning_rate": 8.760360471210865e-06,
"loss": 0.5338,
"step": 7055
},
{
"epoch": 0.8232276119402985,
"grad_norm": 0.4135280946993472,
"learning_rate": 8.736385197862415e-06,
"loss": 0.4824,
"step": 7060
},
{
"epoch": 0.8238106343283582,
"grad_norm": 0.4287858690307368,
"learning_rate": 8.712479677161388e-06,
"loss": 0.4869,
"step": 7065
},
{
"epoch": 0.824393656716418,
"grad_norm": 0.4714198385972294,
"learning_rate": 8.688643997975156e-06,
"loss": 0.5143,
"step": 7070
},
{
"epoch": 0.8249766791044776,
"grad_norm": 0.44395978208148723,
"learning_rate": 8.66487824891149e-06,
"loss": 0.4922,
"step": 7075
},
{
"epoch": 0.8255597014925373,
"grad_norm": 0.508754740613158,
"learning_rate": 8.641182518318162e-06,
"loss": 0.5184,
"step": 7080
},
{
"epoch": 0.8261427238805971,
"grad_norm": 0.4026422778012762,
"learning_rate": 8.617556894282683e-06,
"loss": 0.4886,
"step": 7085
},
{
"epoch": 0.8267257462686567,
"grad_norm": 0.47446388322469524,
"learning_rate": 8.594001464631938e-06,
"loss": 0.5189,
"step": 7090
},
{
"epoch": 0.8273087686567164,
"grad_norm": 0.47024774444454565,
"learning_rate": 8.570516316931869e-06,
"loss": 0.5266,
"step": 7095
},
{
"epoch": 0.8278917910447762,
"grad_norm": 0.44748825048810753,
"learning_rate": 8.547101538487136e-06,
"loss": 0.4845,
"step": 7100
},
{
"epoch": 0.8284748134328358,
"grad_norm": 0.4806024012846129,
"learning_rate": 8.52375721634083e-06,
"loss": 0.5031,
"step": 7105
},
{
"epoch": 0.8290578358208955,
"grad_norm": 0.5413375895981546,
"learning_rate": 8.5004834372741e-06,
"loss": 0.5106,
"step": 7110
},
{
"epoch": 0.8296408582089553,
"grad_norm": 0.4337024823163042,
"learning_rate": 8.477280287805883e-06,
"loss": 0.4954,
"step": 7115
},
{
"epoch": 0.8302238805970149,
"grad_norm": 0.4344528368728774,
"learning_rate": 8.454147854192515e-06,
"loss": 0.5021,
"step": 7120
},
{
"epoch": 0.8308069029850746,
"grad_norm": 0.45520925847409455,
"learning_rate": 8.4310862224275e-06,
"loss": 0.4828,
"step": 7125
},
{
"epoch": 0.8313899253731343,
"grad_norm": 0.4734386762743593,
"learning_rate": 8.408095478241099e-06,
"loss": 0.5243,
"step": 7130
},
{
"epoch": 0.831972947761194,
"grad_norm": 0.4717328221940539,
"learning_rate": 8.385175707100064e-06,
"loss": 0.4907,
"step": 7135
},
{
"epoch": 0.8325559701492538,
"grad_norm": 0.41240208092851716,
"learning_rate": 8.36232699420732e-06,
"loss": 0.5031,
"step": 7140
},
{
"epoch": 0.8331389925373134,
"grad_norm": 0.41514081930390884,
"learning_rate": 8.33954942450163e-06,
"loss": 0.4896,
"step": 7145
},
{
"epoch": 0.8337220149253731,
"grad_norm": 0.4675941109997365,
"learning_rate": 8.316843082657277e-06,
"loss": 0.5009,
"step": 7150
},
{
"epoch": 0.8343050373134329,
"grad_norm": 0.4427930397053809,
"learning_rate": 8.294208053083771e-06,
"loss": 0.511,
"step": 7155
},
{
"epoch": 0.8348880597014925,
"grad_norm": 0.45949248005156934,
"learning_rate": 8.271644419925526e-06,
"loss": 0.4719,
"step": 7160
},
{
"epoch": 0.8354710820895522,
"grad_norm": 0.471413986453529,
"learning_rate": 8.249152267061524e-06,
"loss": 0.4994,
"step": 7165
},
{
"epoch": 0.836054104477612,
"grad_norm": 0.44813478705837095,
"learning_rate": 8.226731678105045e-06,
"loss": 0.4804,
"step": 7170
},
{
"epoch": 0.8366371268656716,
"grad_norm": 0.4426301509227603,
"learning_rate": 8.20438273640332e-06,
"loss": 0.5143,
"step": 7175
},
{
"epoch": 0.8372201492537313,
"grad_norm": 0.4465652598526728,
"learning_rate": 8.18210552503725e-06,
"loss": 0.4871,
"step": 7180
},
{
"epoch": 0.8378031716417911,
"grad_norm": 0.47244655194156704,
"learning_rate": 8.159900126821062e-06,
"loss": 0.5003,
"step": 7185
},
{
"epoch": 0.8383861940298507,
"grad_norm": 0.4471968512251493,
"learning_rate": 8.137766624302036e-06,
"loss": 0.5142,
"step": 7190
},
{
"epoch": 0.8389692164179104,
"grad_norm": 0.4709917176706898,
"learning_rate": 8.115705099760184e-06,
"loss": 0.5195,
"step": 7195
},
{
"epoch": 0.8395522388059702,
"grad_norm": 0.4380879394783618,
"learning_rate": 8.093715635207948e-06,
"loss": 0.4885,
"step": 7200
},
{
"epoch": 0.8401352611940298,
"grad_norm": 0.41712410093981933,
"learning_rate": 8.071798312389863e-06,
"loss": 0.5116,
"step": 7205
},
{
"epoch": 0.8407182835820896,
"grad_norm": 0.4325869434630372,
"learning_rate": 8.049953212782329e-06,
"loss": 0.4478,
"step": 7210
},
{
"epoch": 0.8413013059701493,
"grad_norm": 0.4166574600617677,
"learning_rate": 8.028180417593215e-06,
"loss": 0.4824,
"step": 7215
},
{
"epoch": 0.8418843283582089,
"grad_norm": 0.43830478937394324,
"learning_rate": 8.006480007761628e-06,
"loss": 0.4674,
"step": 7220
},
{
"epoch": 0.8424673507462687,
"grad_norm": 0.4817669338294806,
"learning_rate": 7.98485206395758e-06,
"loss": 0.5077,
"step": 7225
},
{
"epoch": 0.8430503731343284,
"grad_norm": 0.4647045998674549,
"learning_rate": 7.963296666581703e-06,
"loss": 0.4702,
"step": 7230
},
{
"epoch": 0.843633395522388,
"grad_norm": 0.49174994550827694,
"learning_rate": 7.941813895764919e-06,
"loss": 0.5212,
"step": 7235
},
{
"epoch": 0.8442164179104478,
"grad_norm": 0.41757383756848515,
"learning_rate": 7.920403831368189e-06,
"loss": 0.5016,
"step": 7240
},
{
"epoch": 0.8447994402985075,
"grad_norm": 0.44263407474750815,
"learning_rate": 7.899066552982179e-06,
"loss": 0.4994,
"step": 7245
},
{
"epoch": 0.8453824626865671,
"grad_norm": 0.46533860910272895,
"learning_rate": 7.87780213992699e-06,
"loss": 0.4973,
"step": 7250
},
{
"epoch": 0.8459654850746269,
"grad_norm": 0.4396469203604777,
"learning_rate": 7.856610671251826e-06,
"loss": 0.4962,
"step": 7255
},
{
"epoch": 0.8465485074626866,
"grad_norm": 0.461412303324744,
"learning_rate": 7.835492225734753e-06,
"loss": 0.4848,
"step": 7260
},
{
"epoch": 0.8471315298507462,
"grad_norm": 0.5184880522572166,
"learning_rate": 7.81444688188236e-06,
"loss": 0.5133,
"step": 7265
},
{
"epoch": 0.847714552238806,
"grad_norm": 0.47001007740037093,
"learning_rate": 7.793474717929495e-06,
"loss": 0.4852,
"step": 7270
},
{
"epoch": 0.8482975746268657,
"grad_norm": 0.47109511656808234,
"learning_rate": 7.772575811838948e-06,
"loss": 0.4961,
"step": 7275
},
{
"epoch": 0.8488805970149254,
"grad_norm": 0.44116994424682127,
"learning_rate": 7.751750241301192e-06,
"loss": 0.4972,
"step": 7280
},
{
"epoch": 0.8494636194029851,
"grad_norm": 0.4501932519308333,
"learning_rate": 7.730998083734083e-06,
"loss": 0.4859,
"step": 7285
},
{
"epoch": 0.8500466417910447,
"grad_norm": 0.48144234778210115,
"learning_rate": 7.710319416282543e-06,
"loss": 0.4984,
"step": 7290
},
{
"epoch": 0.8506296641791045,
"grad_norm": 0.456988520365441,
"learning_rate": 7.689714315818339e-06,
"loss": 0.5232,
"step": 7295
},
{
"epoch": 0.8512126865671642,
"grad_norm": 0.5275485527755961,
"learning_rate": 7.669182858939715e-06,
"loss": 0.494,
"step": 7300
},
{
"epoch": 0.8517957089552238,
"grad_norm": 0.4358686155004703,
"learning_rate": 7.648725121971178e-06,
"loss": 0.4652,
"step": 7305
},
{
"epoch": 0.8523787313432836,
"grad_norm": 0.5241681742015615,
"learning_rate": 7.628341180963175e-06,
"loss": 0.5107,
"step": 7310
},
{
"epoch": 0.8529617537313433,
"grad_norm": 0.4446315316066114,
"learning_rate": 7.608031111691826e-06,
"loss": 0.4736,
"step": 7315
},
{
"epoch": 0.8535447761194029,
"grad_norm": 0.4586742881823037,
"learning_rate": 7.587794989658621e-06,
"loss": 0.4789,
"step": 7320
},
{
"epoch": 0.8541277985074627,
"grad_norm": 0.43393817483796693,
"learning_rate": 7.567632890090176e-06,
"loss": 0.4517,
"step": 7325
},
{
"epoch": 0.8547108208955224,
"grad_norm": 0.4530263088196386,
"learning_rate": 7.5475448879379255e-06,
"loss": 0.5204,
"step": 7330
},
{
"epoch": 0.855293843283582,
"grad_norm": 0.47232925084661864,
"learning_rate": 7.527531057877849e-06,
"loss": 0.5212,
"step": 7335
},
{
"epoch": 0.8558768656716418,
"grad_norm": 0.4413601041808957,
"learning_rate": 7.507591474310185e-06,
"loss": 0.4907,
"step": 7340
},
{
"epoch": 0.8564598880597015,
"grad_norm": 0.5105056148052758,
"learning_rate": 7.487726211359198e-06,
"loss": 0.5465,
"step": 7345
},
{
"epoch": 0.8570429104477612,
"grad_norm": 0.6715647626166893,
"learning_rate": 7.46793534287283e-06,
"loss": 0.5069,
"step": 7350
},
{
"epoch": 0.8576259328358209,
"grad_norm": 0.4588809773198785,
"learning_rate": 7.448218942422498e-06,
"loss": 0.5474,
"step": 7355
},
{
"epoch": 0.8582089552238806,
"grad_norm": 0.4831576147260091,
"learning_rate": 7.428577083302757e-06,
"loss": 0.5093,
"step": 7360
},
{
"epoch": 0.8587919776119403,
"grad_norm": 0.5042721886700091,
"learning_rate": 7.409009838531095e-06,
"loss": 0.5375,
"step": 7365
},
{
"epoch": 0.859375,
"grad_norm": 0.46284073380005986,
"learning_rate": 7.389517280847598e-06,
"loss": 0.5159,
"step": 7370
},
{
"epoch": 0.8599580223880597,
"grad_norm": 0.4661662360141043,
"learning_rate": 7.370099482714715e-06,
"loss": 0.5315,
"step": 7375
},
{
"epoch": 0.8605410447761194,
"grad_norm": 0.4476959939432871,
"learning_rate": 7.35075651631699e-06,
"loss": 0.4901,
"step": 7380
},
{
"epoch": 0.8611240671641791,
"grad_norm": 0.44717566109177675,
"learning_rate": 7.331488453560767e-06,
"loss": 0.4921,
"step": 7385
},
{
"epoch": 0.8617070895522388,
"grad_norm": 0.4476851149766431,
"learning_rate": 7.312295366073952e-06,
"loss": 0.4839,
"step": 7390
},
{
"epoch": 0.8622901119402985,
"grad_norm": 0.4656768131692163,
"learning_rate": 7.293177325205734e-06,
"loss": 0.5089,
"step": 7395
},
{
"epoch": 0.8628731343283582,
"grad_norm": 0.47867044879346754,
"learning_rate": 7.274134402026321e-06,
"loss": 0.5204,
"step": 7400
},
{
"epoch": 0.863456156716418,
"grad_norm": 0.4495376624711596,
"learning_rate": 7.255166667326668e-06,
"loss": 0.4864,
"step": 7405
},
{
"epoch": 0.8640391791044776,
"grad_norm": 0.45540868280070557,
"learning_rate": 7.236274191618228e-06,
"loss": 0.4969,
"step": 7410
},
{
"epoch": 0.8646222014925373,
"grad_norm": 0.5275767191951953,
"learning_rate": 7.217457045132682e-06,
"loss": 0.5086,
"step": 7415
},
{
"epoch": 0.8652052238805971,
"grad_norm": 0.42863000174753835,
"learning_rate": 7.198715297821681e-06,
"loss": 0.5033,
"step": 7420
},
{
"epoch": 0.8657882462686567,
"grad_norm": 0.4834849241582333,
"learning_rate": 7.18004901935657e-06,
"loss": 0.5057,
"step": 7425
},
{
"epoch": 0.8663712686567164,
"grad_norm": 0.46483441625060457,
"learning_rate": 7.161458279128172e-06,
"loss": 0.5058,
"step": 7430
},
{
"epoch": 0.8669542910447762,
"grad_norm": 0.4699147577232594,
"learning_rate": 7.142943146246471e-06,
"loss": 0.5052,
"step": 7435
},
{
"epoch": 0.8675373134328358,
"grad_norm": 0.4728336401428503,
"learning_rate": 7.124503689540403e-06,
"loss": 0.4945,
"step": 7440
},
{
"epoch": 0.8681203358208955,
"grad_norm": 0.42756247469245834,
"learning_rate": 7.106139977557563e-06,
"loss": 0.4868,
"step": 7445
},
{
"epoch": 0.8687033582089553,
"grad_norm": 0.5508454557117077,
"learning_rate": 7.087852078564006e-06,
"loss": 0.5078,
"step": 7450
},
{
"epoch": 0.8692863805970149,
"grad_norm": 0.42365611783227247,
"learning_rate": 7.069640060543914e-06,
"loss": 0.4795,
"step": 7455
},
{
"epoch": 0.8698694029850746,
"grad_norm": 0.4619373613002068,
"learning_rate": 7.051503991199415e-06,
"loss": 0.5093,
"step": 7460
},
{
"epoch": 0.8704524253731343,
"grad_norm": 0.46430333389633666,
"learning_rate": 7.03344393795029e-06,
"loss": 0.4628,
"step": 7465
},
{
"epoch": 0.871035447761194,
"grad_norm": 0.46350593595683315,
"learning_rate": 7.0154599679337405e-06,
"loss": 0.4966,
"step": 7470
},
{
"epoch": 0.8716184701492538,
"grad_norm": 0.46997441642678967,
"learning_rate": 6.997552148004124e-06,
"loss": 0.4619,
"step": 7475
},
{
"epoch": 0.8722014925373134,
"grad_norm": 0.4940809056382957,
"learning_rate": 6.9797205447327236e-06,
"loss": 0.5207,
"step": 7480
},
{
"epoch": 0.8727845149253731,
"grad_norm": 0.42737497657553214,
"learning_rate": 6.961965224407487e-06,
"loss": 0.4879,
"step": 7485
},
{
"epoch": 0.8733675373134329,
"grad_norm": 0.4420504460833519,
"learning_rate": 6.944286253032789e-06,
"loss": 0.4519,
"step": 7490
},
{
"epoch": 0.8739505597014925,
"grad_norm": 0.4482189949894458,
"learning_rate": 6.9266836963291725e-06,
"loss": 0.5216,
"step": 7495
},
{
"epoch": 0.8745335820895522,
"grad_norm": 0.48507550585616305,
"learning_rate": 6.90915761973312e-06,
"loss": 0.4904,
"step": 7500
},
{
"epoch": 0.875116604477612,
"grad_norm": 0.4845741351343996,
"learning_rate": 6.891708088396803e-06,
"loss": 0.5031,
"step": 7505
},
{
"epoch": 0.8756996268656716,
"grad_norm": 0.45661614078718604,
"learning_rate": 6.874335167187844e-06,
"loss": 0.4694,
"step": 7510
},
{
"epoch": 0.8762826492537313,
"grad_norm": 0.4378046368918111,
"learning_rate": 6.857038920689068e-06,
"loss": 0.473,
"step": 7515
},
{
"epoch": 0.8768656716417911,
"grad_norm": 0.4351278013857574,
"learning_rate": 6.839819413198259e-06,
"loss": 0.4686,
"step": 7520
},
{
"epoch": 0.8774486940298507,
"grad_norm": 0.4598040793979348,
"learning_rate": 6.822676708727941e-06,
"loss": 0.5058,
"step": 7525
},
{
"epoch": 0.8780317164179104,
"grad_norm": 0.46734999477037775,
"learning_rate": 6.805610871005115e-06,
"loss": 0.5142,
"step": 7530
},
{
"epoch": 0.8786147388059702,
"grad_norm": 0.42358187968622485,
"learning_rate": 6.788621963471055e-06,
"loss": 0.4656,
"step": 7535
},
{
"epoch": 0.8791977611940298,
"grad_norm": 0.42943564345062557,
"learning_rate": 6.771710049281019e-06,
"loss": 0.4866,
"step": 7540
},
{
"epoch": 0.8797807835820896,
"grad_norm": 0.46090527920896157,
"learning_rate": 6.754875191304076e-06,
"loss": 0.5283,
"step": 7545
},
{
"epoch": 0.8803638059701493,
"grad_norm": 0.46574541087337956,
"learning_rate": 6.73811745212283e-06,
"loss": 0.5072,
"step": 7550
},
{
"epoch": 0.8809468283582089,
"grad_norm": 0.5402513744522806,
"learning_rate": 6.721436894033206e-06,
"loss": 0.4474,
"step": 7555
},
{
"epoch": 0.8815298507462687,
"grad_norm": 0.4887426580865765,
"learning_rate": 6.704833579044198e-06,
"loss": 0.4945,
"step": 7560
},
{
"epoch": 0.8821128731343284,
"grad_norm": 0.4720599639216187,
"learning_rate": 6.688307568877681e-06,
"loss": 0.4757,
"step": 7565
},
{
"epoch": 0.882695895522388,
"grad_norm": 0.44242282186933635,
"learning_rate": 6.6718589249681215e-06,
"loss": 0.5141,
"step": 7570
},
{
"epoch": 0.8832789179104478,
"grad_norm": 0.47443100760410045,
"learning_rate": 6.655487708462407e-06,
"loss": 0.479,
"step": 7575
},
{
"epoch": 0.8838619402985075,
"grad_norm": 0.4484432447267727,
"learning_rate": 6.639193980219574e-06,
"loss": 0.503,
"step": 7580
},
{
"epoch": 0.8844449626865671,
"grad_norm": 0.4544364445156426,
"learning_rate": 6.622977800810626e-06,
"loss": 0.4757,
"step": 7585
},
{
"epoch": 0.8850279850746269,
"grad_norm": 0.470997939385338,
"learning_rate": 6.60683923051825e-06,
"loss": 0.4791,
"step": 7590
},
{
"epoch": 0.8856110074626866,
"grad_norm": 0.4213706332803066,
"learning_rate": 6.5907783293366525e-06,
"loss": 0.468,
"step": 7595
},
{
"epoch": 0.8861940298507462,
"grad_norm": 0.4352892615766016,
"learning_rate": 6.574795156971298e-06,
"loss": 0.4843,
"step": 7600
},
{
"epoch": 0.886777052238806,
"grad_norm": 0.45036633475522897,
"learning_rate": 6.5588897728387055e-06,
"loss": 0.4705,
"step": 7605
},
{
"epoch": 0.8873600746268657,
"grad_norm": 0.5077288646140908,
"learning_rate": 6.543062236066208e-06,
"loss": 0.4791,
"step": 7610
},
{
"epoch": 0.8879430970149254,
"grad_norm": 0.4513500980150284,
"learning_rate": 6.527312605491758e-06,
"loss": 0.5178,
"step": 7615
},
{
"epoch": 0.8885261194029851,
"grad_norm": 0.4274182074103898,
"learning_rate": 6.5116409396636935e-06,
"loss": 0.4626,
"step": 7620
},
{
"epoch": 0.8891091417910447,
"grad_norm": 0.4667801241554799,
"learning_rate": 6.496047296840513e-06,
"loss": 0.5071,
"step": 7625
},
{
"epoch": 0.8896921641791045,
"grad_norm": 0.4123902058374272,
"learning_rate": 6.480531734990686e-06,
"loss": 0.4992,
"step": 7630
},
{
"epoch": 0.8902751865671642,
"grad_norm": 0.47897565830736566,
"learning_rate": 6.4650943117924065e-06,
"loss": 0.5153,
"step": 7635
},
{
"epoch": 0.8908582089552238,
"grad_norm": 0.47530139972165525,
"learning_rate": 6.449735084633407e-06,
"loss": 0.4857,
"step": 7640
},
{
"epoch": 0.8914412313432836,
"grad_norm": 0.4311055331542708,
"learning_rate": 6.4344541106107046e-06,
"loss": 0.4877,
"step": 7645
},
{
"epoch": 0.8920242537313433,
"grad_norm": 0.4668273679980216,
"learning_rate": 6.419251446530451e-06,
"loss": 0.5164,
"step": 7650
},
{
"epoch": 0.8926072761194029,
"grad_norm": 0.44140094368731336,
"learning_rate": 6.404127148907656e-06,
"loss": 0.4784,
"step": 7655
},
{
"epoch": 0.8931902985074627,
"grad_norm": 0.4544488440695652,
"learning_rate": 6.38908127396602e-06,
"loss": 0.4988,
"step": 7660
},
{
"epoch": 0.8937733208955224,
"grad_norm": 0.4557410652922852,
"learning_rate": 6.374113877637701e-06,
"loss": 0.4987,
"step": 7665
},
{
"epoch": 0.894356343283582,
"grad_norm": 0.4444789436043567,
"learning_rate": 6.359225015563138e-06,
"loss": 0.5243,
"step": 7670
},
{
"epoch": 0.8949393656716418,
"grad_norm": 0.4698237483504685,
"learning_rate": 6.3444147430908015e-06,
"loss": 0.4691,
"step": 7675
},
{
"epoch": 0.8955223880597015,
"grad_norm": 0.4859998755530479,
"learning_rate": 6.329683115277018e-06,
"loss": 0.4916,
"step": 7680
},
{
"epoch": 0.8961054104477612,
"grad_norm": 0.515441325427795,
"learning_rate": 6.315030186885763e-06,
"loss": 0.5116,
"step": 7685
},
{
"epoch": 0.8966884328358209,
"grad_norm": 0.5283029015600534,
"learning_rate": 6.300456012388446e-06,
"loss": 0.4934,
"step": 7690
},
{
"epoch": 0.8972714552238806,
"grad_norm": 0.4417646004083893,
"learning_rate": 6.285960645963708e-06,
"loss": 0.5026,
"step": 7695
},
{
"epoch": 0.8978544776119403,
"grad_norm": 0.4605167879082956,
"learning_rate": 6.271544141497232e-06,
"loss": 0.4901,
"step": 7700
},
{
"epoch": 0.8984375,
"grad_norm": 0.4792028373581772,
"learning_rate": 6.257206552581541e-06,
"loss": 0.5118,
"step": 7705
},
{
"epoch": 0.8990205223880597,
"grad_norm": 0.4644526939230792,
"learning_rate": 6.242947932515786e-06,
"loss": 0.5282,
"step": 7710
},
{
"epoch": 0.8996035447761194,
"grad_norm": 0.5046663803144333,
"learning_rate": 6.228768334305555e-06,
"loss": 0.4946,
"step": 7715
},
{
"epoch": 0.9001865671641791,
"grad_norm": 0.43155084823278395,
"learning_rate": 6.214667810662682e-06,
"loss": 0.4794,
"step": 7720
},
{
"epoch": 0.9007695895522388,
"grad_norm": 0.4640789279302754,
"learning_rate": 6.200646414005046e-06,
"loss": 0.5239,
"step": 7725
},
{
"epoch": 0.9013526119402985,
"grad_norm": 0.45218047321306765,
"learning_rate": 6.18670419645637e-06,
"loss": 0.5015,
"step": 7730
},
{
"epoch": 0.9019356343283582,
"grad_norm": 0.4588478078182171,
"learning_rate": 6.172841209846046e-06,
"loss": 0.4913,
"step": 7735
},
{
"epoch": 0.902518656716418,
"grad_norm": 0.4433138530597733,
"learning_rate": 6.159057505708912e-06,
"loss": 0.4594,
"step": 7740
},
{
"epoch": 0.9031016791044776,
"grad_norm": 0.42245796632484117,
"learning_rate": 6.145353135285091e-06,
"loss": 0.4945,
"step": 7745
},
{
"epoch": 0.9036847014925373,
"grad_norm": 0.4756088628902291,
"learning_rate": 6.131728149519778e-06,
"loss": 0.4932,
"step": 7750
},
{
"epoch": 0.9042677238805971,
"grad_norm": 0.4597129158786207,
"learning_rate": 6.118182599063075e-06,
"loss": 0.5354,
"step": 7755
},
{
"epoch": 0.9048507462686567,
"grad_norm": 0.503389587802853,
"learning_rate": 6.104716534269772e-06,
"loss": 0.5077,
"step": 7760
},
{
"epoch": 0.9054337686567164,
"grad_norm": 0.4249365343444884,
"learning_rate": 6.091330005199183e-06,
"loss": 0.4868,
"step": 7765
},
{
"epoch": 0.9060167910447762,
"grad_norm": 0.43851878310292347,
"learning_rate": 6.078023061614953e-06,
"loss": 0.5015,
"step": 7770
},
{
"epoch": 0.9065998134328358,
"grad_norm": 0.43139132012318054,
"learning_rate": 6.064795752984875e-06,
"loss": 0.4832,
"step": 7775
},
{
"epoch": 0.9071828358208955,
"grad_norm": 0.45409545061847195,
"learning_rate": 6.0516481284806885e-06,
"loss": 0.4794,
"step": 7780
},
{
"epoch": 0.9077658582089553,
"grad_norm": 0.4411595137434583,
"learning_rate": 6.0385802369779385e-06,
"loss": 0.5183,
"step": 7785
},
{
"epoch": 0.9083488805970149,
"grad_norm": 0.45410062982495236,
"learning_rate": 6.025592127055741e-06,
"loss": 0.4736,
"step": 7790
},
{
"epoch": 0.9089319029850746,
"grad_norm": 0.42835954681482424,
"learning_rate": 6.012683846996645e-06,
"loss": 0.4541,
"step": 7795
},
{
"epoch": 0.9095149253731343,
"grad_norm": 0.42047696751886715,
"learning_rate": 5.999855444786425e-06,
"loss": 0.4973,
"step": 7800
},
{
"epoch": 0.910097947761194,
"grad_norm": 0.44914336580430964,
"learning_rate": 5.987106968113928e-06,
"loss": 0.5061,
"step": 7805
},
{
"epoch": 0.9106809701492538,
"grad_norm": 0.460206776855078,
"learning_rate": 5.974438464370872e-06,
"loss": 0.4888,
"step": 7810
},
{
"epoch": 0.9112639925373134,
"grad_norm": 0.4470862201000093,
"learning_rate": 5.961849980651684e-06,
"loss": 0.4659,
"step": 7815
},
{
"epoch": 0.9118470149253731,
"grad_norm": 0.47720200193566115,
"learning_rate": 5.949341563753319e-06,
"loss": 0.4992,
"step": 7820
},
{
"epoch": 0.9124300373134329,
"grad_norm": 0.4676169159668866,
"learning_rate": 5.936913260175094e-06,
"loss": 0.5179,
"step": 7825
},
{
"epoch": 0.9130130597014925,
"grad_norm": 0.4584704293118159,
"learning_rate": 5.924565116118499e-06,
"loss": 0.4797,
"step": 7830
},
{
"epoch": 0.9135960820895522,
"grad_norm": 0.4776754416693919,
"learning_rate": 5.9122971774870435e-06,
"loss": 0.4678,
"step": 7835
},
{
"epoch": 0.914179104477612,
"grad_norm": 0.42982811564868484,
"learning_rate": 5.900109489886081e-06,
"loss": 0.4804,
"step": 7840
},
{
"epoch": 0.9147621268656716,
"grad_norm": 0.41729830382315986,
"learning_rate": 5.8880020986226285e-06,
"loss": 0.468,
"step": 7845
},
{
"epoch": 0.9153451492537313,
"grad_norm": 0.4582133802824904,
"learning_rate": 5.875975048705206e-06,
"loss": 0.4939,
"step": 7850
},
{
"epoch": 0.9159281716417911,
"grad_norm": 0.4934433745443256,
"learning_rate": 5.864028384843678e-06,
"loss": 0.512,
"step": 7855
},
{
"epoch": 0.9165111940298507,
"grad_norm": 0.4461718822871272,
"learning_rate": 5.8521621514490715e-06,
"loss": 0.5226,
"step": 7860
},
{
"epoch": 0.9170942164179104,
"grad_norm": 0.43580261924917685,
"learning_rate": 5.8403763926334146e-06,
"loss": 0.4712,
"step": 7865
},
{
"epoch": 0.9176772388059702,
"grad_norm": 0.4289025060075553,
"learning_rate": 5.82867115220959e-06,
"loss": 0.4857,
"step": 7870
},
{
"epoch": 0.9182602611940298,
"grad_norm": 0.4471506171646922,
"learning_rate": 5.81704647369114e-06,
"loss": 0.5177,
"step": 7875
},
{
"epoch": 0.9188432835820896,
"grad_norm": 0.4349776664841292,
"learning_rate": 5.805502400292137e-06,
"loss": 0.4925,
"step": 7880
},
{
"epoch": 0.9194263059701493,
"grad_norm": 0.4524645377436028,
"learning_rate": 5.794038974926995e-06,
"loss": 0.4807,
"step": 7885
},
{
"epoch": 0.9200093283582089,
"grad_norm": 0.45618145888853556,
"learning_rate": 5.782656240210343e-06,
"loss": 0.5078,
"step": 7890
},
{
"epoch": 0.9205923507462687,
"grad_norm": 0.4316155039975515,
"learning_rate": 5.771354238456828e-06,
"loss": 0.4795,
"step": 7895
},
{
"epoch": 0.9211753731343284,
"grad_norm": 0.46008183460748864,
"learning_rate": 5.760133011680985e-06,
"loss": 0.4788,
"step": 7900
},
{
"epoch": 0.921758395522388,
"grad_norm": 0.44423836815839907,
"learning_rate": 5.748992601597076e-06,
"loss": 0.4949,
"step": 7905
},
{
"epoch": 0.9223414179104478,
"grad_norm": 0.45890602021310334,
"learning_rate": 5.737933049618925e-06,
"loss": 0.4811,
"step": 7910
},
{
"epoch": 0.9229244402985075,
"grad_norm": 0.424972332822205,
"learning_rate": 5.726954396859773e-06,
"loss": 0.4854,
"step": 7915
},
{
"epoch": 0.9235074626865671,
"grad_norm": 0.42910016132492174,
"learning_rate": 5.7160566841321255e-06,
"loss": 0.4512,
"step": 7920
},
{
"epoch": 0.9240904850746269,
"grad_norm": 0.46334797979450737,
"learning_rate": 5.705239951947597e-06,
"loss": 0.4834,
"step": 7925
},
{
"epoch": 0.9246735074626866,
"grad_norm": 0.4603535009443525,
"learning_rate": 5.694504240516759e-06,
"loss": 0.4837,
"step": 7930
},
{
"epoch": 0.9252565298507462,
"grad_norm": 0.4662682403880509,
"learning_rate": 5.683849589748994e-06,
"loss": 0.4823,
"step": 7935
},
{
"epoch": 0.925839552238806,
"grad_norm": 0.44330683887009237,
"learning_rate": 5.673276039252347e-06,
"loss": 0.5095,
"step": 7940
},
{
"epoch": 0.9264225746268657,
"grad_norm": 0.43601918566724357,
"learning_rate": 5.662783628333379e-06,
"loss": 0.5165,
"step": 7945
},
{
"epoch": 0.9270055970149254,
"grad_norm": 0.41717748530531923,
"learning_rate": 5.652372395997015e-06,
"loss": 0.4911,
"step": 7950
},
{
"epoch": 0.9275886194029851,
"grad_norm": 2.401919501282761,
"learning_rate": 5.642042380946412e-06,
"loss": 0.4687,
"step": 7955
},
{
"epoch": 0.9281716417910447,
"grad_norm": 0.41624865614532475,
"learning_rate": 5.631793621582793e-06,
"loss": 0.4945,
"step": 7960
},
{
"epoch": 0.9287546641791045,
"grad_norm": 0.5068445026903259,
"learning_rate": 5.621626156005335e-06,
"loss": 0.4786,
"step": 7965
},
{
"epoch": 0.9293376865671642,
"grad_norm": 0.45630828056750805,
"learning_rate": 5.611540022010985e-06,
"loss": 0.4951,
"step": 7970
},
{
"epoch": 0.9299207089552238,
"grad_norm": 0.4956233980595794,
"learning_rate": 5.6015352570943755e-06,
"loss": 0.4843,
"step": 7975
},
{
"epoch": 0.9305037313432836,
"grad_norm": 0.4329791152614355,
"learning_rate": 5.591611898447632e-06,
"loss": 0.4634,
"step": 7980
},
{
"epoch": 0.9310867537313433,
"grad_norm": 0.4663064517760106,
"learning_rate": 5.581769982960261e-06,
"loss": 0.5264,
"step": 7985
},
{
"epoch": 0.9316697761194029,
"grad_norm": 0.4836205261061597,
"learning_rate": 5.572009547219013e-06,
"loss": 0.5156,
"step": 7990
},
{
"epoch": 0.9322527985074627,
"grad_norm": 0.4277290525866144,
"learning_rate": 5.5623306275077475e-06,
"loss": 0.481,
"step": 7995
},
{
"epoch": 0.9328358208955224,
"grad_norm": 0.43341519970296755,
"learning_rate": 5.552733259807276e-06,
"loss": 0.498,
"step": 8000
},
{
"epoch": 0.933418843283582,
"grad_norm": 0.43071532419905234,
"learning_rate": 5.543217479795256e-06,
"loss": 0.4958,
"step": 8005
},
{
"epoch": 0.9340018656716418,
"grad_norm": 0.46622528251802137,
"learning_rate": 5.533783322846053e-06,
"loss": 0.4814,
"step": 8010
},
{
"epoch": 0.9345848880597015,
"grad_norm": 0.5249920187972582,
"learning_rate": 5.524430824030594e-06,
"loss": 0.4907,
"step": 8015
},
{
"epoch": 0.9351679104477612,
"grad_norm": 0.45135919668531826,
"learning_rate": 5.515160018116247e-06,
"loss": 0.492,
"step": 8020
},
{
"epoch": 0.9357509328358209,
"grad_norm": 0.4928051727444886,
"learning_rate": 5.505970939566699e-06,
"loss": 0.5035,
"step": 8025
},
{
"epoch": 0.9363339552238806,
"grad_norm": 0.44235862805439663,
"learning_rate": 5.4968636225418125e-06,
"loss": 0.4783,
"step": 8030
},
{
"epoch": 0.9369169776119403,
"grad_norm": 0.41958300409071425,
"learning_rate": 5.487838100897508e-06,
"loss": 0.4789,
"step": 8035
},
{
"epoch": 0.9375,
"grad_norm": 0.4570459794031607,
"learning_rate": 5.478894408185641e-06,
"loss": 0.4661,
"step": 8040
},
{
"epoch": 0.9380830223880597,
"grad_norm": 0.4783472459507973,
"learning_rate": 5.470032577653869e-06,
"loss": 0.4758,
"step": 8045
},
{
"epoch": 0.9386660447761194,
"grad_norm": 0.4549507857184879,
"learning_rate": 5.4612526422455265e-06,
"loss": 0.4961,
"step": 8050
},
{
"epoch": 0.9392490671641791,
"grad_norm": 0.4650558485006105,
"learning_rate": 5.452554634599519e-06,
"loss": 0.4891,
"step": 8055
},
{
"epoch": 0.9398320895522388,
"grad_norm": 0.580533723917234,
"learning_rate": 5.443938587050186e-06,
"loss": 0.5172,
"step": 8060
},
{
"epoch": 0.9404151119402985,
"grad_norm": 0.42296595646423346,
"learning_rate": 5.435404531627176e-06,
"loss": 0.4853,
"step": 8065
},
{
"epoch": 0.9409981343283582,
"grad_norm": 0.4573887758625825,
"learning_rate": 5.426952500055348e-06,
"loss": 0.4614,
"step": 8070
},
{
"epoch": 0.941581156716418,
"grad_norm": 0.4745039842823071,
"learning_rate": 5.41858252375464e-06,
"loss": 0.5061,
"step": 8075
},
{
"epoch": 0.9421641791044776,
"grad_norm": 0.47805447519039157,
"learning_rate": 5.410294633839949e-06,
"loss": 0.4735,
"step": 8080
},
{
"epoch": 0.9427472014925373,
"grad_norm": 0.43867449428861033,
"learning_rate": 5.402088861121025e-06,
"loss": 0.4718,
"step": 8085
},
{
"epoch": 0.9433302238805971,
"grad_norm": 0.4467252451572471,
"learning_rate": 5.393965236102353e-06,
"loss": 0.4798,
"step": 8090
},
{
"epoch": 0.9439132462686567,
"grad_norm": 0.5133360727759508,
"learning_rate": 5.385923788983034e-06,
"loss": 0.4894,
"step": 8095
},
{
"epoch": 0.9444962686567164,
"grad_norm": 0.4747210863179483,
"learning_rate": 5.377964549656685e-06,
"loss": 0.5098,
"step": 8100
},
{
"epoch": 0.9450792910447762,
"grad_norm": 0.44347953398302503,
"learning_rate": 5.370087547711307e-06,
"loss": 0.5105,
"step": 8105
},
{
"epoch": 0.9456623134328358,
"grad_norm": 0.4387472897639345,
"learning_rate": 5.362292812429207e-06,
"loss": 0.4815,
"step": 8110
},
{
"epoch": 0.9462453358208955,
"grad_norm": 0.4722026626305582,
"learning_rate": 5.354580372786854e-06,
"loss": 0.4776,
"step": 8115
},
{
"epoch": 0.9468283582089553,
"grad_norm": 0.4566856777345421,
"learning_rate": 5.346950257454792e-06,
"loss": 0.5002,
"step": 8120
},
{
"epoch": 0.9474113805970149,
"grad_norm": 0.4601768931007766,
"learning_rate": 5.339402494797539e-06,
"loss": 0.4725,
"step": 8125
},
{
"epoch": 0.9479944029850746,
"grad_norm": 0.4422563608916944,
"learning_rate": 5.331937112873462e-06,
"loss": 0.4411,
"step": 8130
},
{
"epoch": 0.9485774253731343,
"grad_norm": 0.44855925098469346,
"learning_rate": 5.324554139434679e-06,
"loss": 0.4941,
"step": 8135
},
{
"epoch": 0.949160447761194,
"grad_norm": 0.4437725309777671,
"learning_rate": 5.317253601926967e-06,
"loss": 0.4836,
"step": 8140
},
{
"epoch": 0.9497434701492538,
"grad_norm": 0.4488409563199269,
"learning_rate": 5.310035527489651e-06,
"loss": 0.4958,
"step": 8145
},
{
"epoch": 0.9503264925373134,
"grad_norm": 0.46309779837786297,
"learning_rate": 5.3028999429555045e-06,
"loss": 0.493,
"step": 8150
},
{
"epoch": 0.9509095149253731,
"grad_norm": 0.45673001540720787,
"learning_rate": 5.295846874850646e-06,
"loss": 0.4945,
"step": 8155
},
{
"epoch": 0.9514925373134329,
"grad_norm": 0.4481344375463158,
"learning_rate": 5.288876349394448e-06,
"loss": 0.4957,
"step": 8160
},
{
"epoch": 0.9520755597014925,
"grad_norm": 0.45392704978409076,
"learning_rate": 5.281988392499431e-06,
"loss": 0.4721,
"step": 8165
},
{
"epoch": 0.9526585820895522,
"grad_norm": 0.45349195761765704,
"learning_rate": 5.275183029771177e-06,
"loss": 0.4741,
"step": 8170
},
{
"epoch": 0.953241604477612,
"grad_norm": 0.4411901617730967,
"learning_rate": 5.2684602865082255e-06,
"loss": 0.4966,
"step": 8175
},
{
"epoch": 0.9538246268656716,
"grad_norm": 0.4585915951637137,
"learning_rate": 5.261820187701984e-06,
"loss": 0.5453,
"step": 8180
},
{
"epoch": 0.9544076492537313,
"grad_norm": 0.4593205732104069,
"learning_rate": 5.2552627580366334e-06,
"loss": 0.5013,
"step": 8185
},
{
"epoch": 0.9549906716417911,
"grad_norm": 0.42592095115888917,
"learning_rate": 5.248788021889036e-06,
"loss": 0.4797,
"step": 8190
},
{
"epoch": 0.9555736940298507,
"grad_norm": 0.46962819902389563,
"learning_rate": 5.2423960033286505e-06,
"loss": 0.4763,
"step": 8195
},
{
"epoch": 0.9561567164179104,
"grad_norm": 0.40855238448922176,
"learning_rate": 5.236086726117433e-06,
"loss": 0.4743,
"step": 8200
},
{
"epoch": 0.9567397388059702,
"grad_norm": 0.43575271699567064,
"learning_rate": 5.229860213709753e-06,
"loss": 0.4773,
"step": 8205
},
{
"epoch": 0.9573227611940298,
"grad_norm": 0.4688749371758774,
"learning_rate": 5.223716489252311e-06,
"loss": 0.5166,
"step": 8210
},
{
"epoch": 0.9579057835820896,
"grad_norm": 0.7670542201127664,
"learning_rate": 5.217655575584045e-06,
"loss": 0.493,
"step": 8215
},
{
"epoch": 0.9584888059701493,
"grad_norm": 0.46421302891934385,
"learning_rate": 5.211677495236046e-06,
"loss": 0.5096,
"step": 8220
},
{
"epoch": 0.9590718283582089,
"grad_norm": 0.4771862479599736,
"learning_rate": 5.205782270431484e-06,
"loss": 0.4693,
"step": 8225
},
{
"epoch": 0.9596548507462687,
"grad_norm": 0.44226860049129957,
"learning_rate": 5.199969923085515e-06,
"loss": 0.4459,
"step": 8230
},
{
"epoch": 0.9602378731343284,
"grad_norm": 0.4294703525768145,
"learning_rate": 5.194240474805201e-06,
"loss": 0.481,
"step": 8235
},
{
"epoch": 0.960820895522388,
"grad_norm": 0.46086183427389743,
"learning_rate": 5.188593946889429e-06,
"loss": 0.4973,
"step": 8240
},
{
"epoch": 0.9614039179104478,
"grad_norm": 0.4280169626175964,
"learning_rate": 5.183030360328846e-06,
"loss": 0.4698,
"step": 8245
},
{
"epoch": 0.9619869402985075,
"grad_norm": 0.4753252130786709,
"learning_rate": 5.177549735805758e-06,
"loss": 0.5015,
"step": 8250
},
{
"epoch": 0.9625699626865671,
"grad_norm": 0.4620963248632002,
"learning_rate": 5.172152093694067e-06,
"loss": 0.4802,
"step": 8255
},
{
"epoch": 0.9631529850746269,
"grad_norm": 0.5332881434749811,
"learning_rate": 5.166837454059193e-06,
"loss": 0.5106,
"step": 8260
},
{
"epoch": 0.9637360074626866,
"grad_norm": 0.5025697757795641,
"learning_rate": 5.161605836658004e-06,
"loss": 0.4986,
"step": 8265
},
{
"epoch": 0.9643190298507462,
"grad_norm": 0.46525870817249937,
"learning_rate": 5.156457260938732e-06,
"loss": 0.4925,
"step": 8270
},
{
"epoch": 0.964902052238806,
"grad_norm": 0.4466358224458004,
"learning_rate": 5.151391746040905e-06,
"loss": 0.4896,
"step": 8275
},
{
"epoch": 0.9654850746268657,
"grad_norm": 0.4722232462608345,
"learning_rate": 5.146409310795282e-06,
"loss": 0.5116,
"step": 8280
},
{
"epoch": 0.9660680970149254,
"grad_norm": 0.4839383338646517,
"learning_rate": 5.14150997372378e-06,
"loss": 0.4704,
"step": 8285
},
{
"epoch": 0.9666511194029851,
"grad_norm": 0.4598082477182381,
"learning_rate": 5.1366937530393955e-06,
"loss": 0.4675,
"step": 8290
},
{
"epoch": 0.9672341417910447,
"grad_norm": 0.4571533268091363,
"learning_rate": 5.131960666646149e-06,
"loss": 0.4886,
"step": 8295
},
{
"epoch": 0.9678171641791045,
"grad_norm": 0.4826350517333677,
"learning_rate": 5.127310732139018e-06,
"loss": 0.5239,
"step": 8300
},
{
"epoch": 0.9684001865671642,
"grad_norm": 0.501035586719562,
"learning_rate": 5.122743966803858e-06,
"loss": 0.4845,
"step": 8305
},
{
"epoch": 0.9689832089552238,
"grad_norm": 0.45689428681207833,
"learning_rate": 5.118260387617359e-06,
"loss": 0.4802,
"step": 8310
},
{
"epoch": 0.9695662313432836,
"grad_norm": 0.41839841626319924,
"learning_rate": 5.113860011246964e-06,
"loss": 0.4759,
"step": 8315
},
{
"epoch": 0.9701492537313433,
"grad_norm": 0.4623224498637685,
"learning_rate": 5.109542854050814e-06,
"loss": 0.5191,
"step": 8320
},
{
"epoch": 0.9707322761194029,
"grad_norm": 0.4400774444581771,
"learning_rate": 5.105308932077693e-06,
"loss": 0.4751,
"step": 8325
},
{
"epoch": 0.9713152985074627,
"grad_norm": 0.5458461815433695,
"learning_rate": 5.101158261066959e-06,
"loss": 0.4934,
"step": 8330
},
{
"epoch": 0.9718983208955224,
"grad_norm": 0.45228504839890776,
"learning_rate": 5.097090856448492e-06,
"loss": 0.5007,
"step": 8335
},
{
"epoch": 0.972481343283582,
"grad_norm": 0.467798541042544,
"learning_rate": 5.0931067333426275e-06,
"loss": 0.5082,
"step": 8340
},
{
"epoch": 0.9730643656716418,
"grad_norm": 0.4266615826249801,
"learning_rate": 5.0892059065601145e-06,
"loss": 0.4974,
"step": 8345
},
{
"epoch": 0.9736473880597015,
"grad_norm": 0.4621554435159516,
"learning_rate": 5.0853883906020525e-06,
"loss": 0.4965,
"step": 8350
},
{
"epoch": 0.9742304104477612,
"grad_norm": 0.46073201489628657,
"learning_rate": 5.081654199659831e-06,
"loss": 0.5071,
"step": 8355
},
{
"epoch": 0.9748134328358209,
"grad_norm": 0.5462060706132782,
"learning_rate": 5.07800334761509e-06,
"loss": 0.4938,
"step": 8360
},
{
"epoch": 0.9753964552238806,
"grad_norm": 0.48250851908785264,
"learning_rate": 5.074435848039658e-06,
"loss": 0.5091,
"step": 8365
},
{
"epoch": 0.9759794776119403,
"grad_norm": 0.44847740844622386,
"learning_rate": 5.070951714195508e-06,
"loss": 0.4957,
"step": 8370
},
{
"epoch": 0.9765625,
"grad_norm": 0.4467178102848771,
"learning_rate": 5.067550959034707e-06,
"loss": 0.49,
"step": 8375
},
{
"epoch": 0.9771455223880597,
"grad_norm": 0.46636700776310513,
"learning_rate": 5.064233595199362e-06,
"loss": 0.4884,
"step": 8380
},
{
"epoch": 0.9777285447761194,
"grad_norm": 0.4615425374049883,
"learning_rate": 5.060999635021583e-06,
"loss": 0.4771,
"step": 8385
},
{
"epoch": 0.9783115671641791,
"grad_norm": 0.45286825679857706,
"learning_rate": 5.057849090523426e-06,
"loss": 0.4959,
"step": 8390
},
{
"epoch": 0.9788945895522388,
"grad_norm": 0.4569055516084358,
"learning_rate": 5.054781973416858e-06,
"loss": 0.4955,
"step": 8395
},
{
"epoch": 0.9794776119402985,
"grad_norm": 0.4455395860822749,
"learning_rate": 5.051798295103711e-06,
"loss": 0.4752,
"step": 8400
},
{
"epoch": 0.9800606343283582,
"grad_norm": 0.4183019666197293,
"learning_rate": 5.048898066675631e-06,
"loss": 0.4552,
"step": 8405
},
{
"epoch": 0.980643656716418,
"grad_norm": 0.43323200908430703,
"learning_rate": 5.046081298914053e-06,
"loss": 0.5092,
"step": 8410
},
{
"epoch": 0.9812266791044776,
"grad_norm": 0.46606214824527303,
"learning_rate": 5.043348002290145e-06,
"loss": 0.4639,
"step": 8415
},
{
"epoch": 0.9818097014925373,
"grad_norm": 0.4620950107319596,
"learning_rate": 5.0406981869647805e-06,
"loss": 0.5072,
"step": 8420
},
{
"epoch": 0.9823927238805971,
"grad_norm": 0.44313034547356717,
"learning_rate": 5.038131862788491e-06,
"loss": 0.4765,
"step": 8425
},
{
"epoch": 0.9829757462686567,
"grad_norm": 0.44702546793927656,
"learning_rate": 5.035649039301438e-06,
"loss": 0.4612,
"step": 8430
},
{
"epoch": 0.9835587686567164,
"grad_norm": 0.45947101038805505,
"learning_rate": 5.033249725733377e-06,
"loss": 0.4967,
"step": 8435
},
{
"epoch": 0.9841417910447762,
"grad_norm": 0.42282849418109447,
"learning_rate": 5.0309339310036125e-06,
"loss": 0.507,
"step": 8440
},
{
"epoch": 0.9847248134328358,
"grad_norm": 0.44692770292599415,
"learning_rate": 5.02870166372098e-06,
"loss": 0.4808,
"step": 8445
},
{
"epoch": 0.9853078358208955,
"grad_norm": 0.5582818670824954,
"learning_rate": 5.0265529321838004e-06,
"loss": 0.5405,
"step": 8450
},
{
"epoch": 0.9858908582089553,
"grad_norm": 0.4665732954055579,
"learning_rate": 5.02448774437986e-06,
"loss": 0.5157,
"step": 8455
},
{
"epoch": 0.9864738805970149,
"grad_norm": 0.5202669052061701,
"learning_rate": 5.022506107986374e-06,
"loss": 0.5146,
"step": 8460
},
{
"epoch": 0.9870569029850746,
"grad_norm": 0.4563878919577009,
"learning_rate": 5.020608030369962e-06,
"loss": 0.4614,
"step": 8465
},
{
"epoch": 0.9876399253731343,
"grad_norm": 0.45761657701588687,
"learning_rate": 5.018793518586616e-06,
"loss": 0.5007,
"step": 8470
},
{
"epoch": 0.988222947761194,
"grad_norm": 0.46136936074859125,
"learning_rate": 5.017062579381676e-06,
"loss": 0.5068,
"step": 8475
},
{
"epoch": 0.9888059701492538,
"grad_norm": 0.48777910519871126,
"learning_rate": 5.015415219189812e-06,
"loss": 0.4889,
"step": 8480
},
{
"epoch": 0.9893889925373134,
"grad_norm": 0.47084471147391416,
"learning_rate": 5.013851444134987e-06,
"loss": 0.5022,
"step": 8485
},
{
"epoch": 0.9899720149253731,
"grad_norm": 0.432844624719787,
"learning_rate": 5.012371260030445e-06,
"loss": 0.491,
"step": 8490
},
{
"epoch": 0.9905550373134329,
"grad_norm": 0.4649632477223049,
"learning_rate": 5.010974672378682e-06,
"loss": 0.4741,
"step": 8495
},
{
"epoch": 0.9911380597014925,
"grad_norm": 0.4764372199483869,
"learning_rate": 5.009661686371434e-06,
"loss": 0.4772,
"step": 8500
},
{
"epoch": 0.9917210820895522,
"grad_norm": 0.4868588188840104,
"learning_rate": 5.008432306889652e-06,
"loss": 0.5214,
"step": 8505
},
{
"epoch": 0.992304104477612,
"grad_norm": 0.438360158275501,
"learning_rate": 5.0072865385034785e-06,
"loss": 0.4905,
"step": 8510
},
{
"epoch": 0.9928871268656716,
"grad_norm": 0.48340344564902316,
"learning_rate": 5.006224385472242e-06,
"loss": 0.4927,
"step": 8515
},
{
"epoch": 0.9934701492537313,
"grad_norm": 0.45531955111325645,
"learning_rate": 5.0052458517444364e-06,
"loss": 0.4888,
"step": 8520
},
{
"epoch": 0.9940531716417911,
"grad_norm": 0.4362698593831642,
"learning_rate": 5.004350940957703e-06,
"loss": 0.4818,
"step": 8525
},
{
"epoch": 0.9946361940298507,
"grad_norm": 0.46810128935885154,
"learning_rate": 5.0035396564388184e-06,
"loss": 0.5101,
"step": 8530
},
{
"epoch": 0.9952192164179104,
"grad_norm": 0.8386528644620809,
"learning_rate": 5.00281200120369e-06,
"loss": 0.4791,
"step": 8535
},
{
"epoch": 0.9958022388059702,
"grad_norm": 0.4834359055420526,
"learning_rate": 5.00216797795733e-06,
"loss": 0.4913,
"step": 8540
},
{
"epoch": 0.9963852611940298,
"grad_norm": 0.4925810326491755,
"learning_rate": 5.001607589093861e-06,
"loss": 0.5076,
"step": 8545
},
{
"epoch": 0.9969682835820896,
"grad_norm": 0.44173971825287517,
"learning_rate": 5.001130836696491e-06,
"loss": 0.4809,
"step": 8550
},
{
"epoch": 0.9975513059701493,
"grad_norm": 0.48077397748529965,
"learning_rate": 5.000737722537526e-06,
"loss": 0.4974,
"step": 8555
},
{
"epoch": 0.9981343283582089,
"grad_norm": 0.6138381620178532,
"learning_rate": 5.00042824807834e-06,
"loss": 0.4925,
"step": 8560
},
{
"epoch": 0.9987173507462687,
"grad_norm": 0.43738249534157836,
"learning_rate": 5.000202414469386e-06,
"loss": 0.4573,
"step": 8565
},
{
"epoch": 0.9993003731343284,
"grad_norm": 0.4562639975448534,
"learning_rate": 5.0000602225501925e-06,
"loss": 0.4941,
"step": 8570
},
{
"epoch": 0.999883395522388,
"grad_norm": 0.4592908023435152,
"learning_rate": 5.0000016728493425e-06,
"loss": 0.4962,
"step": 8575
},
{
"epoch": 1.0004664179104477,
"grad_norm": 0.47426058627957746,
"learning_rate": 4.728485497099385e-05,
"loss": 0.4391,
"step": 8580
},
{
"epoch": 1.0010494402985075,
"grad_norm": 0.5320004656516157,
"learning_rate": 4.728072178445397e-05,
"loss": 0.4522,
"step": 8585
},
{
"epoch": 1.0016324626865671,
"grad_norm": 0.5024183427100246,
"learning_rate": 4.7276585656710295e-05,
"loss": 0.4439,
"step": 8590
},
{
"epoch": 1.0022154850746268,
"grad_norm": 0.5574190311460309,
"learning_rate": 4.7272446588377794e-05,
"loss": 0.4286,
"step": 8595
},
{
"epoch": 1.0027985074626866,
"grad_norm": 0.48789901216879533,
"learning_rate": 4.726830458007194e-05,
"loss": 0.4493,
"step": 8600
},
{
"epoch": 1.0033815298507462,
"grad_norm": 0.5860716293937468,
"learning_rate": 4.7264159632408603e-05,
"loss": 0.4677,
"step": 8605
},
{
"epoch": 1.0039645522388059,
"grad_norm": 0.5517100367748053,
"learning_rate": 4.72600117460041e-05,
"loss": 0.4521,
"step": 8610
},
{
"epoch": 1.0045475746268657,
"grad_norm": 0.486171634292782,
"learning_rate": 4.7255860921475156e-05,
"loss": 0.485,
"step": 8615
},
{
"epoch": 1.0051305970149254,
"grad_norm": 0.531279693662836,
"learning_rate": 4.725170715943898e-05,
"loss": 0.4816,
"step": 8620
},
{
"epoch": 1.005713619402985,
"grad_norm": 0.5180505587047434,
"learning_rate": 4.724755046051321e-05,
"loss": 0.4304,
"step": 8625
},
{
"epoch": 1.0062966417910448,
"grad_norm": 0.5284778037290939,
"learning_rate": 4.724339082531588e-05,
"loss": 0.4632,
"step": 8630
},
{
"epoch": 1.0068796641791045,
"grad_norm": 0.5374534908810564,
"learning_rate": 4.72392282544655e-05,
"loss": 0.4742,
"step": 8635
},
{
"epoch": 1.007462686567164,
"grad_norm": 0.49462912725198666,
"learning_rate": 4.723506274858101e-05,
"loss": 0.4861,
"step": 8640
},
{
"epoch": 1.008045708955224,
"grad_norm": 0.487854302513415,
"learning_rate": 4.723089430828177e-05,
"loss": 0.4928,
"step": 8645
},
{
"epoch": 1.0086287313432836,
"grad_norm": 0.5019200031208862,
"learning_rate": 4.722672293418759e-05,
"loss": 0.493,
"step": 8650
},
{
"epoch": 1.0092117537313432,
"grad_norm": 0.582688345784807,
"learning_rate": 4.722254862691871e-05,
"loss": 0.5059,
"step": 8655
},
{
"epoch": 1.009794776119403,
"grad_norm": 0.5238946658768917,
"learning_rate": 4.721837138709582e-05,
"loss": 0.5027,
"step": 8660
},
{
"epoch": 1.0103777985074627,
"grad_norm": 0.5185530471749461,
"learning_rate": 4.7214191215340036e-05,
"loss": 0.5167,
"step": 8665
},
{
"epoch": 1.0109608208955223,
"grad_norm": 0.5974306068998476,
"learning_rate": 4.7210008112272895e-05,
"loss": 0.4869,
"step": 8670
},
{
"epoch": 1.0115438432835822,
"grad_norm": 0.6107926371037933,
"learning_rate": 4.72058220785164e-05,
"loss": 0.5024,
"step": 8675
},
{
"epoch": 1.0121268656716418,
"grad_norm": 0.5282921823444569,
"learning_rate": 4.720163311469296e-05,
"loss": 0.4619,
"step": 8680
},
{
"epoch": 1.0127098880597014,
"grad_norm": 0.6003339223461499,
"learning_rate": 4.7197441221425446e-05,
"loss": 0.5173,
"step": 8685
},
{
"epoch": 1.0132929104477613,
"grad_norm": 0.4730173461537816,
"learning_rate": 4.7193246399337146e-05,
"loss": 0.4813,
"step": 8690
},
{
"epoch": 1.013875932835821,
"grad_norm": 0.5071456563461645,
"learning_rate": 4.7189048649051786e-05,
"loss": 0.4793,
"step": 8695
},
{
"epoch": 1.0144589552238805,
"grad_norm": 0.5015048651689884,
"learning_rate": 4.718484797119355e-05,
"loss": 0.5245,
"step": 8700
},
{
"epoch": 1.0150419776119404,
"grad_norm": 0.526074227917695,
"learning_rate": 4.718064436638701e-05,
"loss": 0.475,
"step": 8705
},
{
"epoch": 1.015625,
"grad_norm": 0.5009454385549627,
"learning_rate": 4.717643783525722e-05,
"loss": 0.4524,
"step": 8710
},
{
"epoch": 1.0162080223880596,
"grad_norm": 0.5373569510563955,
"learning_rate": 4.717222837842965e-05,
"loss": 0.4908,
"step": 8715
},
{
"epoch": 1.0167910447761195,
"grad_norm": 0.4843490024297434,
"learning_rate": 4.7168015996530204e-05,
"loss": 0.4695,
"step": 8720
},
{
"epoch": 1.017374067164179,
"grad_norm": 0.4665154044423995,
"learning_rate": 4.716380069018523e-05,
"loss": 0.4744,
"step": 8725
},
{
"epoch": 1.0179570895522387,
"grad_norm": 0.5304849338647405,
"learning_rate": 4.715958246002148e-05,
"loss": 0.4814,
"step": 8730
},
{
"epoch": 1.0185401119402986,
"grad_norm": 0.48442401715883604,
"learning_rate": 4.715536130666619e-05,
"loss": 0.4741,
"step": 8735
},
{
"epoch": 1.0191231343283582,
"grad_norm": 0.5382707553704985,
"learning_rate": 4.715113723074699e-05,
"loss": 0.4884,
"step": 8740
},
{
"epoch": 1.0197061567164178,
"grad_norm": 0.5264939747473608,
"learning_rate": 4.7146910232891975e-05,
"loss": 0.4908,
"step": 8745
},
{
"epoch": 1.0202891791044777,
"grad_norm": 0.46111084467153807,
"learning_rate": 4.714268031372964e-05,
"loss": 0.4963,
"step": 8750
},
{
"epoch": 1.0208722014925373,
"grad_norm": 0.5123460394242938,
"learning_rate": 4.7138447473888945e-05,
"loss": 0.452,
"step": 8755
},
{
"epoch": 1.021455223880597,
"grad_norm": 0.468514834125539,
"learning_rate": 4.7134211713999264e-05,
"loss": 0.4578,
"step": 8760
},
{
"epoch": 1.0220382462686568,
"grad_norm": 0.4684210782948509,
"learning_rate": 4.712997303469043e-05,
"loss": 0.4705,
"step": 8765
},
{
"epoch": 1.0226212686567164,
"grad_norm": 0.5073812307209864,
"learning_rate": 4.712573143659268e-05,
"loss": 0.5215,
"step": 8770
},
{
"epoch": 1.023204291044776,
"grad_norm": 0.5322751538831147,
"learning_rate": 4.71214869203367e-05,
"loss": 0.5065,
"step": 8775
},
{
"epoch": 1.023787313432836,
"grad_norm": 0.49987343777690785,
"learning_rate": 4.711723948655362e-05,
"loss": 0.4676,
"step": 8780
},
{
"epoch": 1.0243703358208955,
"grad_norm": 0.5284446960377239,
"learning_rate": 4.711298913587497e-05,
"loss": 0.4509,
"step": 8785
},
{
"epoch": 1.0249533582089552,
"grad_norm": 0.47657330330882275,
"learning_rate": 4.710873586893276e-05,
"loss": 0.4871,
"step": 8790
},
{
"epoch": 1.025536380597015,
"grad_norm": 0.4842840362473736,
"learning_rate": 4.710447968635941e-05,
"loss": 0.4951,
"step": 8795
},
{
"epoch": 1.0261194029850746,
"grad_norm": 0.4510225196671863,
"learning_rate": 4.7100220588787755e-05,
"loss": 0.4686,
"step": 8800
},
{
"epoch": 1.0267024253731343,
"grad_norm": 0.49756046747429755,
"learning_rate": 4.709595857685109e-05,
"loss": 0.5167,
"step": 8805
},
{
"epoch": 1.0272854477611941,
"grad_norm": 0.4744972653076067,
"learning_rate": 4.7091693651183144e-05,
"loss": 0.487,
"step": 8810
},
{
"epoch": 1.0278684701492538,
"grad_norm": 0.5150713800815724,
"learning_rate": 4.708742581241806e-05,
"loss": 0.5044,
"step": 8815
},
{
"epoch": 1.0284514925373134,
"grad_norm": 0.557518317918137,
"learning_rate": 4.7083155061190426e-05,
"loss": 0.5088,
"step": 8820
},
{
"epoch": 1.0290345149253732,
"grad_norm": 0.5309722610343153,
"learning_rate": 4.707888139813527e-05,
"loss": 0.4847,
"step": 8825
},
{
"epoch": 1.0296175373134329,
"grad_norm": 0.5018724058089924,
"learning_rate": 4.707460482388804e-05,
"loss": 0.4847,
"step": 8830
},
{
"epoch": 1.0302005597014925,
"grad_norm": 0.5097255324921017,
"learning_rate": 4.707032533908461e-05,
"loss": 0.4737,
"step": 8835
},
{
"epoch": 1.0307835820895523,
"grad_norm": 0.6170849116718103,
"learning_rate": 4.706604294436132e-05,
"loss": 0.4815,
"step": 8840
},
{
"epoch": 1.031366604477612,
"grad_norm": 0.5081805654355479,
"learning_rate": 4.706175764035491e-05,
"loss": 0.519,
"step": 8845
},
{
"epoch": 1.0319496268656716,
"grad_norm": 0.5130221033518372,
"learning_rate": 4.705746942770255e-05,
"loss": 0.4659,
"step": 8850
},
{
"epoch": 1.0325326492537314,
"grad_norm": 0.5066243868822757,
"learning_rate": 4.7053178307041886e-05,
"loss": 0.4911,
"step": 8855
},
{
"epoch": 1.033115671641791,
"grad_norm": 0.4661575061349955,
"learning_rate": 4.704888427901094e-05,
"loss": 0.4657,
"step": 8860
},
{
"epoch": 1.0336986940298507,
"grad_norm": 0.47132915772418305,
"learning_rate": 4.704458734424821e-05,
"loss": 0.512,
"step": 8865
},
{
"epoch": 1.0342817164179103,
"grad_norm": 0.4675671747669863,
"learning_rate": 4.70402875033926e-05,
"loss": 0.4833,
"step": 8870
},
{
"epoch": 1.0348647388059702,
"grad_norm": 0.5619087684145831,
"learning_rate": 4.7035984757083454e-05,
"loss": 0.5305,
"step": 8875
},
{
"epoch": 1.0354477611940298,
"grad_norm": 0.49373872507593697,
"learning_rate": 4.703167910596055e-05,
"loss": 0.5055,
"step": 8880
},
{
"epoch": 1.0360307835820897,
"grad_norm": 0.47489827739888235,
"learning_rate": 4.70273705506641e-05,
"loss": 0.4775,
"step": 8885
},
{
"epoch": 1.0366138059701493,
"grad_norm": 0.4971757998829849,
"learning_rate": 4.702305909183475e-05,
"loss": 0.4826,
"step": 8890
},
{
"epoch": 1.037196828358209,
"grad_norm": 0.5422128300971927,
"learning_rate": 4.7018744730113565e-05,
"loss": 0.5111,
"step": 8895
},
{
"epoch": 1.0377798507462686,
"grad_norm": 0.5012066261927423,
"learning_rate": 4.701442746614206e-05,
"loss": 0.5076,
"step": 8900
},
{
"epoch": 1.0383628731343284,
"grad_norm": 0.5258269384569444,
"learning_rate": 4.701010730056216e-05,
"loss": 0.4817,
"step": 8905
},
{
"epoch": 1.038945895522388,
"grad_norm": 0.5029440378162809,
"learning_rate": 4.700578423401622e-05,
"loss": 0.4841,
"step": 8910
},
{
"epoch": 1.0395289179104477,
"grad_norm": 0.5032231245192578,
"learning_rate": 4.7001458267147055e-05,
"loss": 0.4596,
"step": 8915
},
{
"epoch": 1.0401119402985075,
"grad_norm": 0.5668474501893467,
"learning_rate": 4.699712940059791e-05,
"loss": 0.5245,
"step": 8920
},
{
"epoch": 1.0406949626865671,
"grad_norm": 0.7205300131824032,
"learning_rate": 4.6992797635012415e-05,
"loss": 0.4943,
"step": 8925
},
{
"epoch": 1.0412779850746268,
"grad_norm": 0.46430225487702537,
"learning_rate": 4.6988462971034676e-05,
"loss": 0.4969,
"step": 8930
},
{
"epoch": 1.0418610074626866,
"grad_norm": 0.508570057790338,
"learning_rate": 4.698412540930922e-05,
"loss": 0.4993,
"step": 8935
},
{
"epoch": 1.0424440298507462,
"grad_norm": 0.4725765308686987,
"learning_rate": 4.697978495048099e-05,
"loss": 0.4678,
"step": 8940
},
{
"epoch": 1.0430270522388059,
"grad_norm": 0.5665033300096499,
"learning_rate": 4.697544159519539e-05,
"loss": 0.4983,
"step": 8945
},
{
"epoch": 1.0436100746268657,
"grad_norm": 0.5501878853405641,
"learning_rate": 4.697109534409821e-05,
"loss": 0.5027,
"step": 8950
},
{
"epoch": 1.0441930970149254,
"grad_norm": 0.5139618047452844,
"learning_rate": 4.696674619783572e-05,
"loss": 0.4927,
"step": 8955
},
{
"epoch": 1.044776119402985,
"grad_norm": 0.5540540968653056,
"learning_rate": 4.696239415705458e-05,
"loss": 0.5114,
"step": 8960
},
{
"epoch": 1.0453591417910448,
"grad_norm": 0.5196349914754745,
"learning_rate": 4.69580392224019e-05,
"loss": 0.4993,
"step": 8965
},
{
"epoch": 1.0459421641791045,
"grad_norm": 0.6616451345728891,
"learning_rate": 4.695368139452521e-05,
"loss": 0.472,
"step": 8970
},
{
"epoch": 1.046525186567164,
"grad_norm": 0.4858247139067617,
"learning_rate": 4.69493206740725e-05,
"loss": 0.515,
"step": 8975
},
{
"epoch": 1.047108208955224,
"grad_norm": 0.48350960828960377,
"learning_rate": 4.694495706169214e-05,
"loss": 0.4762,
"step": 8980
},
{
"epoch": 1.0476912313432836,
"grad_norm": 0.5122511024079021,
"learning_rate": 4.6940590558032985e-05,
"loss": 0.4952,
"step": 8985
},
{
"epoch": 1.0482742537313432,
"grad_norm": 1.3959823952147061,
"learning_rate": 4.693622116374427e-05,
"loss": 0.5212,
"step": 8990
},
{
"epoch": 1.048857276119403,
"grad_norm": 0.5401382717722655,
"learning_rate": 4.693184887947569e-05,
"loss": 0.4941,
"step": 8995
},
{
"epoch": 1.0494402985074627,
"grad_norm": 0.4507778073049574,
"learning_rate": 4.692747370587737e-05,
"loss": 0.499,
"step": 9000
},
{
"epoch": 1.0500233208955223,
"grad_norm": 0.48797343069389976,
"learning_rate": 4.692309564359984e-05,
"loss": 0.461,
"step": 9005
},
{
"epoch": 1.0506063432835822,
"grad_norm": 0.47561161106654043,
"learning_rate": 4.691871469329408e-05,
"loss": 0.5224,
"step": 9010
},
{
"epoch": 1.0511893656716418,
"grad_norm": 0.5634009367566019,
"learning_rate": 4.69143308556115e-05,
"loss": 0.4768,
"step": 9015
},
{
"epoch": 1.0517723880597014,
"grad_norm": 0.49734215708222496,
"learning_rate": 4.690994413120394e-05,
"loss": 0.4823,
"step": 9020
},
{
"epoch": 1.0523554104477613,
"grad_norm": 0.4665279232609275,
"learning_rate": 4.690555452072366e-05,
"loss": 0.4763,
"step": 9025
},
{
"epoch": 1.052938432835821,
"grad_norm": 0.5190852875616855,
"learning_rate": 4.690116202482335e-05,
"loss": 0.4974,
"step": 9030
},
{
"epoch": 1.0535214552238805,
"grad_norm": 0.46566803498581916,
"learning_rate": 4.689676664415613e-05,
"loss": 0.4779,
"step": 9035
},
{
"epoch": 1.0541044776119404,
"grad_norm": 0.5319093271984696,
"learning_rate": 4.689236837937556e-05,
"loss": 0.4909,
"step": 9040
},
{
"epoch": 1.0546875,
"grad_norm": 0.4767808134962894,
"learning_rate": 4.6887967231135605e-05,
"loss": 0.5057,
"step": 9045
},
{
"epoch": 1.0552705223880596,
"grad_norm": 0.47198420277762565,
"learning_rate": 4.688356320009069e-05,
"loss": 0.5153,
"step": 9050
},
{
"epoch": 1.0558535447761195,
"grad_norm": 0.4972543814636995,
"learning_rate": 4.687915628689564e-05,
"loss": 0.5068,
"step": 9055
},
{
"epoch": 1.056436567164179,
"grad_norm": 0.5146050784315456,
"learning_rate": 4.687474649220573e-05,
"loss": 0.4925,
"step": 9060
},
{
"epoch": 1.0570195895522387,
"grad_norm": 0.5259065163440027,
"learning_rate": 4.687033381667664e-05,
"loss": 0.5216,
"step": 9065
},
{
"epoch": 1.0576026119402986,
"grad_norm": 0.4843993142938887,
"learning_rate": 4.6865918260964506e-05,
"loss": 0.4815,
"step": 9070
},
{
"epoch": 1.0581856343283582,
"grad_norm": 0.48264634391304173,
"learning_rate": 4.6861499825725877e-05,
"loss": 0.472,
"step": 9075
},
{
"epoch": 1.0587686567164178,
"grad_norm": 0.4916707467335896,
"learning_rate": 4.685707851161773e-05,
"loss": 0.5039,
"step": 9080
},
{
"epoch": 1.0593516791044777,
"grad_norm": 0.5395782527592068,
"learning_rate": 4.685265431929747e-05,
"loss": 0.5192,
"step": 9085
},
{
"epoch": 1.0599347014925373,
"grad_norm": 0.48955754808478563,
"learning_rate": 4.6848227249422936e-05,
"loss": 0.4976,
"step": 9090
},
{
"epoch": 1.060517723880597,
"grad_norm": 0.5173966026954939,
"learning_rate": 4.684379730265239e-05,
"loss": 0.4904,
"step": 9095
},
{
"epoch": 1.0611007462686568,
"grad_norm": 0.5586831591167815,
"learning_rate": 4.683936447964452e-05,
"loss": 0.5159,
"step": 9100
},
{
"epoch": 1.0616837686567164,
"grad_norm": 0.5306411920252676,
"learning_rate": 4.683492878105844e-05,
"loss": 0.5023,
"step": 9105
},
{
"epoch": 1.062266791044776,
"grad_norm": 0.5394987090128364,
"learning_rate": 4.683049020755372e-05,
"loss": 0.5145,
"step": 9110
},
{
"epoch": 1.062849813432836,
"grad_norm": 0.49914789057311915,
"learning_rate": 4.6826048759790295e-05,
"loss": 0.4957,
"step": 9115
},
{
"epoch": 1.0634328358208955,
"grad_norm": 0.5095562857323067,
"learning_rate": 4.6821604438428594e-05,
"loss": 0.52,
"step": 9120
},
{
"epoch": 1.0640158582089552,
"grad_norm": 0.4947345455878514,
"learning_rate": 4.6817157244129436e-05,
"loss": 0.5119,
"step": 9125
},
{
"epoch": 1.064598880597015,
"grad_norm": 0.49090098726653636,
"learning_rate": 4.681270717755409e-05,
"loss": 0.5096,
"step": 9130
},
{
"epoch": 1.0651819029850746,
"grad_norm": 0.5338327678334841,
"learning_rate": 4.680825423936421e-05,
"loss": 0.4926,
"step": 9135
},
{
"epoch": 1.0657649253731343,
"grad_norm": 0.5090081376265945,
"learning_rate": 4.680379843022192e-05,
"loss": 0.4686,
"step": 9140
},
{
"epoch": 1.0663479477611941,
"grad_norm": 0.44239826305309765,
"learning_rate": 4.679933975078976e-05,
"loss": 0.4555,
"step": 9145
},
{
"epoch": 1.0669309701492538,
"grad_norm": 0.45188824230008673,
"learning_rate": 4.679487820173069e-05,
"loss": 0.4937,
"step": 9150
},
{
"epoch": 1.0675139925373134,
"grad_norm": 0.49566573297809374,
"learning_rate": 4.67904137837081e-05,
"loss": 0.452,
"step": 9155
},
{
"epoch": 1.0680970149253732,
"grad_norm": 0.5076839916250646,
"learning_rate": 4.678594649738581e-05,
"loss": 0.4913,
"step": 9160
},
{
"epoch": 1.0686800373134329,
"grad_norm": 0.4373870137902814,
"learning_rate": 4.678147634342805e-05,
"loss": 0.4681,
"step": 9165
},
{
"epoch": 1.0692630597014925,
"grad_norm": 0.4800654095382689,
"learning_rate": 4.67770033224995e-05,
"loss": 0.4948,
"step": 9170
},
{
"epoch": 1.0698460820895523,
"grad_norm": 0.46964119819112704,
"learning_rate": 4.677252743526525e-05,
"loss": 0.4559,
"step": 9175
},
{
"epoch": 1.070429104477612,
"grad_norm": 0.49650644344484723,
"learning_rate": 4.676804868239083e-05,
"loss": 0.4652,
"step": 9180
},
{
"epoch": 1.0710121268656716,
"grad_norm": 0.8449820819897319,
"learning_rate": 4.676356706454217e-05,
"loss": 0.4713,
"step": 9185
},
{
"epoch": 1.0715951492537314,
"grad_norm": 0.48019854470117596,
"learning_rate": 4.675908258238567e-05,
"loss": 0.501,
"step": 9190
},
{
"epoch": 1.072178171641791,
"grad_norm": 0.510626881014733,
"learning_rate": 4.6754595236588114e-05,
"loss": 0.4853,
"step": 9195
},
{
"epoch": 1.0727611940298507,
"grad_norm": 0.4874317455515729,
"learning_rate": 4.6750105027816716e-05,
"loss": 0.5114,
"step": 9200
},
{
"epoch": 1.0733442164179103,
"grad_norm": 0.49008478006262846,
"learning_rate": 4.674561195673914e-05,
"loss": 0.5026,
"step": 9205
},
{
"epoch": 1.0739272388059702,
"grad_norm": 0.45606704402648035,
"learning_rate": 4.6741116024023476e-05,
"loss": 0.4815,
"step": 9210
},
{
"epoch": 1.0745102611940298,
"grad_norm": 0.5470864580462397,
"learning_rate": 4.6736617230338205e-05,
"loss": 0.5029,
"step": 9215
},
{
"epoch": 1.0750932835820897,
"grad_norm": 1.0582793581481065,
"learning_rate": 4.673211557635225e-05,
"loss": 0.4717,
"step": 9220
},
{
"epoch": 1.0756763059701493,
"grad_norm": 0.5869186280878189,
"learning_rate": 4.6727611062734994e-05,
"loss": 0.5123,
"step": 9225
},
{
"epoch": 1.076259328358209,
"grad_norm": 0.48811171212689763,
"learning_rate": 4.672310369015619e-05,
"loss": 0.5022,
"step": 9230
},
{
"epoch": 1.0768423507462686,
"grad_norm": 0.4379874815657095,
"learning_rate": 4.6718593459286036e-05,
"loss": 0.4868,
"step": 9235
},
{
"epoch": 1.0774253731343284,
"grad_norm": 0.536314202527189,
"learning_rate": 4.671408037079519e-05,
"loss": 0.5014,
"step": 9240
},
{
"epoch": 1.078008395522388,
"grad_norm": 0.455274704274001,
"learning_rate": 4.670956442535467e-05,
"loss": 0.4843,
"step": 9245
},
{
"epoch": 1.0785914179104477,
"grad_norm": 0.4838644013521913,
"learning_rate": 4.670504562363598e-05,
"loss": 0.4902,
"step": 9250
},
{
"epoch": 1.0791744402985075,
"grad_norm": 0.4735942629468596,
"learning_rate": 4.6700523966311e-05,
"loss": 0.536,
"step": 9255
},
{
"epoch": 1.0797574626865671,
"grad_norm": 0.5403926324710048,
"learning_rate": 4.669599945405208e-05,
"loss": 0.5086,
"step": 9260
},
{
"epoch": 1.0803404850746268,
"grad_norm": 0.43208196517726205,
"learning_rate": 4.6691472087531967e-05,
"loss": 0.4789,
"step": 9265
},
{
"epoch": 1.0809235074626866,
"grad_norm": 0.4463915398795724,
"learning_rate": 4.668694186742383e-05,
"loss": 0.4841,
"step": 9270
},
{
"epoch": 1.0815065298507462,
"grad_norm": 0.4637201156142629,
"learning_rate": 4.668240879440127e-05,
"loss": 0.4747,
"step": 9275
},
{
"epoch": 1.0820895522388059,
"grad_norm": 0.5083155084770778,
"learning_rate": 4.6677872869138304e-05,
"loss": 0.5122,
"step": 9280
},
{
"epoch": 1.0826725746268657,
"grad_norm": 0.4567415229261271,
"learning_rate": 4.66733340923094e-05,
"loss": 0.4732,
"step": 9285
},
{
"epoch": 1.0832555970149254,
"grad_norm": 0.46826732560954215,
"learning_rate": 4.666879246458941e-05,
"loss": 0.5041,
"step": 9290
},
{
"epoch": 1.083838619402985,
"grad_norm": 0.4764779152646372,
"learning_rate": 4.6664247986653645e-05,
"loss": 0.5059,
"step": 9295
},
{
"epoch": 1.0844216417910448,
"grad_norm": 0.46620112639686195,
"learning_rate": 4.6659700659177814e-05,
"loss": 0.4608,
"step": 9300
},
{
"epoch": 1.0850046641791045,
"grad_norm": 0.4915813706267885,
"learning_rate": 4.665515048283808e-05,
"loss": 0.4997,
"step": 9305
},
{
"epoch": 1.085587686567164,
"grad_norm": 0.43796257819202183,
"learning_rate": 4.665059745831098e-05,
"loss": 0.4994,
"step": 9310
},
{
"epoch": 1.086170708955224,
"grad_norm": 0.5151808827501788,
"learning_rate": 4.664604158627355e-05,
"loss": 0.4713,
"step": 9315
},
{
"epoch": 1.0867537313432836,
"grad_norm": 0.49502837394742893,
"learning_rate": 4.6641482867403156e-05,
"loss": 0.5048,
"step": 9320
},
{
"epoch": 1.0873367537313432,
"grad_norm": 0.4620786512707241,
"learning_rate": 4.663692130237766e-05,
"loss": 0.4756,
"step": 9325
},
{
"epoch": 1.087919776119403,
"grad_norm": 0.4758291599811604,
"learning_rate": 4.6632356891875336e-05,
"loss": 0.5135,
"step": 9330
},
{
"epoch": 1.0885027985074627,
"grad_norm": 0.47571548818133175,
"learning_rate": 4.6627789636574836e-05,
"loss": 0.4952,
"step": 9335
},
{
"epoch": 1.0890858208955223,
"grad_norm": 0.474514825567827,
"learning_rate": 4.662321953715529e-05,
"loss": 0.4858,
"step": 9340
},
{
"epoch": 1.0896688432835822,
"grad_norm": 0.469298753664266,
"learning_rate": 4.6618646594296226e-05,
"loss": 0.4772,
"step": 9345
},
{
"epoch": 1.0902518656716418,
"grad_norm": 0.46481163188836494,
"learning_rate": 4.661407080867759e-05,
"loss": 0.4887,
"step": 9350
},
{
"epoch": 1.0908348880597014,
"grad_norm": 0.5020235935240444,
"learning_rate": 4.6609492180979756e-05,
"loss": 0.4904,
"step": 9355
},
{
"epoch": 1.0914179104477613,
"grad_norm": 0.47503692296447025,
"learning_rate": 4.660491071188353e-05,
"loss": 0.5197,
"step": 9360
},
{
"epoch": 1.092000932835821,
"grad_norm": 0.5448342002718358,
"learning_rate": 4.660032640207013e-05,
"loss": 0.5347,
"step": 9365
},
{
"epoch": 1.0925839552238805,
"grad_norm": 0.472936178069635,
"learning_rate": 4.6595739252221196e-05,
"loss": 0.5326,
"step": 9370
},
{
"epoch": 1.0931669776119404,
"grad_norm": 0.48811140194669395,
"learning_rate": 4.6591149263018796e-05,
"loss": 0.5094,
"step": 9375
},
{
"epoch": 1.09375,
"grad_norm": 0.46926160934689126,
"learning_rate": 4.658655643514541e-05,
"loss": 0.4928,
"step": 9380
},
{
"epoch": 1.0943330223880596,
"grad_norm": 0.4603620397682986,
"learning_rate": 4.6581960769283955e-05,
"loss": 0.4901,
"step": 9385
},
{
"epoch": 1.0949160447761195,
"grad_norm": 0.4775351626053154,
"learning_rate": 4.657736226611778e-05,
"loss": 0.5341,
"step": 9390
},
{
"epoch": 1.095499067164179,
"grad_norm": 0.5354532994408574,
"learning_rate": 4.657276092633059e-05,
"loss": 0.5089,
"step": 9395
},
{
"epoch": 1.0960820895522387,
"grad_norm": 0.46464617476535186,
"learning_rate": 4.656815675060662e-05,
"loss": 0.485,
"step": 9400
},
{
"epoch": 1.0966651119402986,
"grad_norm": 0.4646395350154372,
"learning_rate": 4.656354973963041e-05,
"loss": 0.4913,
"step": 9405
},
{
"epoch": 1.0972481343283582,
"grad_norm": 0.5303126801182483,
"learning_rate": 4.655893989408702e-05,
"loss": 0.5052,
"step": 9410
},
{
"epoch": 1.0978311567164178,
"grad_norm": 0.4792925399666974,
"learning_rate": 4.6554327214661876e-05,
"loss": 0.4784,
"step": 9415
},
{
"epoch": 1.0984141791044777,
"grad_norm": 0.48310458072243645,
"learning_rate": 4.654971170204083e-05,
"loss": 0.4818,
"step": 9420
},
{
"epoch": 1.0989972014925373,
"grad_norm": 0.49198574089775615,
"learning_rate": 4.654509335691018e-05,
"loss": 0.478,
"step": 9425
},
{
"epoch": 1.099580223880597,
"grad_norm": 0.4933335405522241,
"learning_rate": 4.6540472179956625e-05,
"loss": 0.5031,
"step": 9430
},
{
"epoch": 1.1001632462686568,
"grad_norm": 0.45486513478535884,
"learning_rate": 4.6535848171867275e-05,
"loss": 0.5033,
"step": 9435
},
{
"epoch": 1.1007462686567164,
"grad_norm": 0.49315534436056047,
"learning_rate": 4.6531221333329694e-05,
"loss": 0.5081,
"step": 9440
},
{
"epoch": 1.101329291044776,
"grad_norm": 0.4587755290744345,
"learning_rate": 4.652659166503184e-05,
"loss": 0.481,
"step": 9445
},
{
"epoch": 1.101912313432836,
"grad_norm": 0.4697565098870484,
"learning_rate": 4.652195916766211e-05,
"loss": 0.4809,
"step": 9450
},
{
"epoch": 1.1024953358208955,
"grad_norm": 0.5226619085445289,
"learning_rate": 4.651732384190929e-05,
"loss": 0.4955,
"step": 9455
},
{
"epoch": 1.1030783582089552,
"grad_norm": 0.48377912133133427,
"learning_rate": 4.6512685688462645e-05,
"loss": 0.5489,
"step": 9460
},
{
"epoch": 1.103661380597015,
"grad_norm": 0.5086592264829859,
"learning_rate": 4.6508044708011784e-05,
"loss": 0.4849,
"step": 9465
},
{
"epoch": 1.1042444029850746,
"grad_norm": 0.4609475650380094,
"learning_rate": 4.65034009012468e-05,
"loss": 0.4928,
"step": 9470
},
{
"epoch": 1.1048274253731343,
"grad_norm": 0.4959771192749763,
"learning_rate": 4.649875426885818e-05,
"loss": 0.5119,
"step": 9475
},
{
"epoch": 1.1054104477611941,
"grad_norm": 0.48799886078119564,
"learning_rate": 4.649410481153683e-05,
"loss": 0.4918,
"step": 9480
},
{
"epoch": 1.1059934701492538,
"grad_norm": 0.6184568571875035,
"learning_rate": 4.648945252997407e-05,
"loss": 0.4846,
"step": 9485
},
{
"epoch": 1.1065764925373134,
"grad_norm": 0.5193725706342754,
"learning_rate": 4.6484797424861675e-05,
"loss": 0.4931,
"step": 9490
},
{
"epoch": 1.1071595149253732,
"grad_norm": 0.44442788847116255,
"learning_rate": 4.6480139496891796e-05,
"loss": 0.4562,
"step": 9495
},
{
"epoch": 1.1077425373134329,
"grad_norm": 0.4449516559335734,
"learning_rate": 4.6475478746757025e-05,
"loss": 0.5007,
"step": 9500
},
{
"epoch": 1.1083255597014925,
"grad_norm": 0.5289263524242726,
"learning_rate": 4.6470815175150364e-05,
"loss": 0.5147,
"step": 9505
},
{
"epoch": 1.1089085820895523,
"grad_norm": 0.4972928998457614,
"learning_rate": 4.646614878276526e-05,
"loss": 0.5089,
"step": 9510
},
{
"epoch": 1.109491604477612,
"grad_norm": 0.446913491949988,
"learning_rate": 4.646147957029555e-05,
"loss": 0.4915,
"step": 9515
},
{
"epoch": 1.1100746268656716,
"grad_norm": 0.4775093930913278,
"learning_rate": 4.64568075384355e-05,
"loss": 0.4999,
"step": 9520
},
{
"epoch": 1.1106576492537314,
"grad_norm": 0.44866621767628573,
"learning_rate": 4.6452132687879796e-05,
"loss": 0.4638,
"step": 9525
},
{
"epoch": 1.111240671641791,
"grad_norm": 0.46156410846716345,
"learning_rate": 4.644745501932355e-05,
"loss": 0.48,
"step": 9530
},
{
"epoch": 1.1118236940298507,
"grad_norm": 0.4564165396246152,
"learning_rate": 4.644277453346227e-05,
"loss": 0.4955,
"step": 9535
},
{
"epoch": 1.1124067164179103,
"grad_norm": 0.48881473046306984,
"learning_rate": 4.643809123099192e-05,
"loss": 0.525,
"step": 9540
},
{
"epoch": 1.1129897388059702,
"grad_norm": 0.5063410142886016,
"learning_rate": 4.6433405112608845e-05,
"loss": 0.5017,
"step": 9545
},
{
"epoch": 1.1135727611940298,
"grad_norm": 0.4634913419143995,
"learning_rate": 4.6428716179009844e-05,
"loss": 0.4923,
"step": 9550
},
{
"epoch": 1.1141557835820897,
"grad_norm": 0.44747230770129015,
"learning_rate": 4.6424024430892105e-05,
"loss": 0.4753,
"step": 9555
},
{
"epoch": 1.1147388059701493,
"grad_norm": 0.5267843304204154,
"learning_rate": 4.641932986895325e-05,
"loss": 0.4864,
"step": 9560
},
{
"epoch": 1.115321828358209,
"grad_norm": 0.49303473319964075,
"learning_rate": 4.64146324938913e-05,
"loss": 0.5239,
"step": 9565
},
{
"epoch": 1.1159048507462686,
"grad_norm": 0.4536781686999443,
"learning_rate": 4.6409932306404735e-05,
"loss": 0.4804,
"step": 9570
},
{
"epoch": 1.1164878731343284,
"grad_norm": 0.5465837228080015,
"learning_rate": 4.640522930719241e-05,
"loss": 0.5159,
"step": 9575
},
{
"epoch": 1.117070895522388,
"grad_norm": 0.4206690194245236,
"learning_rate": 4.640052349695363e-05,
"loss": 0.44,
"step": 9580
},
{
"epoch": 1.1176539179104477,
"grad_norm": 0.46474965912968363,
"learning_rate": 4.6395814876388086e-05,
"loss": 0.483,
"step": 9585
},
{
"epoch": 1.1182369402985075,
"grad_norm": 0.49977594667791597,
"learning_rate": 4.6391103446195915e-05,
"loss": 0.4804,
"step": 9590
},
{
"epoch": 1.1188199626865671,
"grad_norm": 0.4470192727431772,
"learning_rate": 4.6386389207077665e-05,
"loss": 0.482,
"step": 9595
},
{
"epoch": 1.1194029850746268,
"grad_norm": 0.4619398096288415,
"learning_rate": 4.6381672159734287e-05,
"loss": 0.4891,
"step": 9600
},
{
"epoch": 1.1199860074626866,
"grad_norm": 0.45123039183594194,
"learning_rate": 4.637695230486718e-05,
"loss": 0.5042,
"step": 9605
},
{
"epoch": 1.1205690298507462,
"grad_norm": 0.7988979042706097,
"learning_rate": 4.637222964317811e-05,
"loss": 0.5056,
"step": 9610
},
{
"epoch": 1.1211520522388059,
"grad_norm": 0.48769226712085023,
"learning_rate": 4.6367504175369326e-05,
"loss": 0.4817,
"step": 9615
},
{
"epoch": 1.1217350746268657,
"grad_norm": 0.45837775913119466,
"learning_rate": 4.636277590214344e-05,
"loss": 0.4797,
"step": 9620
},
{
"epoch": 1.1223180970149254,
"grad_norm": 0.4736479089789474,
"learning_rate": 4.63580448242035e-05,
"loss": 0.4854,
"step": 9625
},
{
"epoch": 1.122901119402985,
"grad_norm": 0.4734355258184259,
"learning_rate": 4.6353310942252986e-05,
"loss": 0.5026,
"step": 9630
},
{
"epoch": 1.1234841417910448,
"grad_norm": 0.4531022842477336,
"learning_rate": 4.634857425699577e-05,
"loss": 0.4939,
"step": 9635
},
{
"epoch": 1.1240671641791045,
"grad_norm": 0.45650809373210965,
"learning_rate": 4.634383476913615e-05,
"loss": 0.4905,
"step": 9640
},
{
"epoch": 1.124650186567164,
"grad_norm": 0.5049363292458079,
"learning_rate": 4.633909247937884e-05,
"loss": 0.5124,
"step": 9645
},
{
"epoch": 1.125233208955224,
"grad_norm": 0.4666273322582206,
"learning_rate": 4.6334347388429e-05,
"loss": 0.444,
"step": 9650
},
{
"epoch": 1.1258162313432836,
"grad_norm": 0.44415022743546667,
"learning_rate": 4.6329599496992145e-05,
"loss": 0.4767,
"step": 9655
},
{
"epoch": 1.1263992537313432,
"grad_norm": 0.4194373685239034,
"learning_rate": 4.632484880577425e-05,
"loss": 0.4825,
"step": 9660
},
{
"epoch": 1.126982276119403,
"grad_norm": 0.46274489368091026,
"learning_rate": 4.632009531548171e-05,
"loss": 0.4918,
"step": 9665
},
{
"epoch": 1.1275652985074627,
"grad_norm": 0.46087571366096425,
"learning_rate": 4.6315339026821305e-05,
"loss": 0.5047,
"step": 9670
},
{
"epoch": 1.1281483208955223,
"grad_norm": 0.4785351679933754,
"learning_rate": 4.631057994050027e-05,
"loss": 0.4671,
"step": 9675
},
{
"epoch": 1.1287313432835822,
"grad_norm": 0.4536972049315929,
"learning_rate": 4.6305818057226226e-05,
"loss": 0.5097,
"step": 9680
},
{
"epoch": 1.1293143656716418,
"grad_norm": 0.4963284888796665,
"learning_rate": 4.630105337770722e-05,
"loss": 0.5098,
"step": 9685
},
{
"epoch": 1.1298973880597014,
"grad_norm": 0.4814699797414588,
"learning_rate": 4.62962859026517e-05,
"loss": 0.5175,
"step": 9690
},
{
"epoch": 1.1304804104477613,
"grad_norm": 0.4545111543543466,
"learning_rate": 4.629151563276857e-05,
"loss": 0.4986,
"step": 9695
},
{
"epoch": 1.131063432835821,
"grad_norm": 0.5441314432893252,
"learning_rate": 4.62867425687671e-05,
"loss": 0.5291,
"step": 9700
},
{
"epoch": 1.1316464552238805,
"grad_norm": 0.465387680919809,
"learning_rate": 4.6281966711357014e-05,
"loss": 0.5149,
"step": 9705
},
{
"epoch": 1.1322294776119404,
"grad_norm": 0.4594632439309518,
"learning_rate": 4.6277188061248436e-05,
"loss": 0.4979,
"step": 9710
},
{
"epoch": 1.1328125,
"grad_norm": 0.4946851483949794,
"learning_rate": 4.6272406619151896e-05,
"loss": 0.491,
"step": 9715
},
{
"epoch": 1.1333955223880596,
"grad_norm": 0.47943186970197044,
"learning_rate": 4.626762238577836e-05,
"loss": 0.4894,
"step": 9720
},
{
"epoch": 1.1339785447761195,
"grad_norm": 0.44710867224321554,
"learning_rate": 4.626283536183918e-05,
"loss": 0.5072,
"step": 9725
},
{
"epoch": 1.134561567164179,
"grad_norm": 0.462289128611941,
"learning_rate": 4.6258045548046166e-05,
"loss": 0.4788,
"step": 9730
},
{
"epoch": 1.1351445895522387,
"grad_norm": 0.4868105685200524,
"learning_rate": 4.625325294511149e-05,
"loss": 0.4868,
"step": 9735
},
{
"epoch": 1.1357276119402986,
"grad_norm": 0.4790346195426802,
"learning_rate": 4.624845755374779e-05,
"loss": 0.5232,
"step": 9740
},
{
"epoch": 1.1363106343283582,
"grad_norm": 0.5622037868349276,
"learning_rate": 4.624365937466808e-05,
"loss": 0.4915,
"step": 9745
},
{
"epoch": 1.1368936567164178,
"grad_norm": 0.551869800635871,
"learning_rate": 4.6238858408585804e-05,
"loss": 0.4657,
"step": 9750
},
{
"epoch": 1.1374766791044777,
"grad_norm": 0.550117149877168,
"learning_rate": 4.623405465621483e-05,
"loss": 0.493,
"step": 9755
},
{
"epoch": 1.1380597014925373,
"grad_norm": 0.4748486725342485,
"learning_rate": 4.622924811826942e-05,
"loss": 0.5158,
"step": 9760
},
{
"epoch": 1.138642723880597,
"grad_norm": 0.5332978068300538,
"learning_rate": 4.622443879546426e-05,
"loss": 0.5166,
"step": 9765
},
{
"epoch": 1.1392257462686568,
"grad_norm": 0.47769296938995587,
"learning_rate": 4.6219626688514456e-05,
"loss": 0.5003,
"step": 9770
},
{
"epoch": 1.1398087686567164,
"grad_norm": 0.4560542794837011,
"learning_rate": 4.621481179813552e-05,
"loss": 0.4695,
"step": 9775
},
{
"epoch": 1.140391791044776,
"grad_norm": 0.5074775456178793,
"learning_rate": 4.620999412504338e-05,
"loss": 0.4987,
"step": 9780
},
{
"epoch": 1.140974813432836,
"grad_norm": 0.4837252945781173,
"learning_rate": 4.620517366995437e-05,
"loss": 0.4963,
"step": 9785
},
{
"epoch": 1.1415578358208955,
"grad_norm": 0.45856568740720954,
"learning_rate": 4.620035043358526e-05,
"loss": 0.4961,
"step": 9790
},
{
"epoch": 1.1421408582089552,
"grad_norm": 0.4698757059244029,
"learning_rate": 4.619552441665322e-05,
"loss": 0.5111,
"step": 9795
},
{
"epoch": 1.142723880597015,
"grad_norm": 0.4499050961202368,
"learning_rate": 4.619069561987581e-05,
"loss": 0.4926,
"step": 9800
},
{
"epoch": 1.1433069029850746,
"grad_norm": 0.45156172108012554,
"learning_rate": 4.618586404397104e-05,
"loss": 0.4798,
"step": 9805
},
{
"epoch": 1.1438899253731343,
"grad_norm": 0.45580586613281165,
"learning_rate": 4.618102968965733e-05,
"loss": 0.4725,
"step": 9810
},
{
"epoch": 1.1444729477611941,
"grad_norm": 0.43980063096394745,
"learning_rate": 4.617619255765349e-05,
"loss": 0.4752,
"step": 9815
},
{
"epoch": 1.1450559701492538,
"grad_norm": 0.48147377861958357,
"learning_rate": 4.6171352648678755e-05,
"loss": 0.4981,
"step": 9820
},
{
"epoch": 1.1456389925373134,
"grad_norm": 0.5069422112555739,
"learning_rate": 4.616650996345277e-05,
"loss": 0.4734,
"step": 9825
},
{
"epoch": 1.1462220149253732,
"grad_norm": 0.5087637858166896,
"learning_rate": 4.6161664502695606e-05,
"loss": 0.5172,
"step": 9830
},
{
"epoch": 1.1468050373134329,
"grad_norm": 0.42833596634640814,
"learning_rate": 4.6156816267127726e-05,
"loss": 0.4649,
"step": 9835
},
{
"epoch": 1.1473880597014925,
"grad_norm": 0.4598225831428788,
"learning_rate": 4.615196525747003e-05,
"loss": 0.46,
"step": 9840
},
{
"epoch": 1.1479710820895521,
"grad_norm": 0.5264734272354958,
"learning_rate": 4.61471114744438e-05,
"loss": 0.5014,
"step": 9845
},
{
"epoch": 1.148554104477612,
"grad_norm": 0.48798600824327,
"learning_rate": 4.6142254918770764e-05,
"loss": 0.5134,
"step": 9850
},
{
"epoch": 1.1491371268656716,
"grad_norm": 0.48672018339280493,
"learning_rate": 4.6137395591173035e-05,
"loss": 0.5069,
"step": 9855
},
{
"epoch": 1.1497201492537314,
"grad_norm": 0.4470801393970488,
"learning_rate": 4.613253349237314e-05,
"loss": 0.4745,
"step": 9860
},
{
"epoch": 1.150303171641791,
"grad_norm": 0.6393897239521419,
"learning_rate": 4.612766862309404e-05,
"loss": 0.5135,
"step": 9865
},
{
"epoch": 1.1508861940298507,
"grad_norm": 0.4702345469985217,
"learning_rate": 4.612280098405909e-05,
"loss": 0.4825,
"step": 9870
},
{
"epoch": 1.1514692164179103,
"grad_norm": 0.45536375824284325,
"learning_rate": 4.611793057599208e-05,
"loss": 0.4509,
"step": 9875
},
{
"epoch": 1.1520522388059702,
"grad_norm": 0.6847591734638865,
"learning_rate": 4.611305739961715e-05,
"loss": 0.4968,
"step": 9880
},
{
"epoch": 1.1526352611940298,
"grad_norm": 0.4881444154577869,
"learning_rate": 4.6108181455658936e-05,
"loss": 0.5217,
"step": 9885
},
{
"epoch": 1.1532182835820897,
"grad_norm": 0.47170189984374566,
"learning_rate": 4.610330274484242e-05,
"loss": 0.4984,
"step": 9890
},
{
"epoch": 1.1538013059701493,
"grad_norm": 0.47449564973164,
"learning_rate": 4.6098421267893024e-05,
"loss": 0.4729,
"step": 9895
},
{
"epoch": 1.154384328358209,
"grad_norm": 0.45892838721809553,
"learning_rate": 4.609353702553659e-05,
"loss": 0.4877,
"step": 9900
},
{
"epoch": 1.1549673507462686,
"grad_norm": 0.49238188160505747,
"learning_rate": 4.608865001849935e-05,
"loss": 0.4879,
"step": 9905
},
{
"epoch": 1.1555503731343284,
"grad_norm": 0.6389954991876162,
"learning_rate": 4.6083760247507945e-05,
"loss": 0.5487,
"step": 9910
},
{
"epoch": 1.156133395522388,
"grad_norm": 0.48889944206681957,
"learning_rate": 4.607886771328945e-05,
"loss": 0.4958,
"step": 9915
},
{
"epoch": 1.1567164179104479,
"grad_norm": 0.5675028453045466,
"learning_rate": 4.607397241657133e-05,
"loss": 0.4861,
"step": 9920
},
{
"epoch": 1.1572994402985075,
"grad_norm": 0.45399241888001685,
"learning_rate": 4.6069074358081476e-05,
"loss": 0.4797,
"step": 9925
},
{
"epoch": 1.1578824626865671,
"grad_norm": 0.5737213260956343,
"learning_rate": 4.606417353854818e-05,
"loss": 0.484,
"step": 9930
},
{
"epoch": 1.1584654850746268,
"grad_norm": 0.47210844957751424,
"learning_rate": 4.6059269958700136e-05,
"loss": 0.4956,
"step": 9935
},
{
"epoch": 1.1590485074626866,
"grad_norm": 0.4452252202160641,
"learning_rate": 4.605436361926648e-05,
"loss": 0.4561,
"step": 9940
},
{
"epoch": 1.1596315298507462,
"grad_norm": 0.48156847504108874,
"learning_rate": 4.604945452097672e-05,
"loss": 0.5153,
"step": 9945
},
{
"epoch": 1.1602145522388059,
"grad_norm": 0.44527665917724885,
"learning_rate": 4.6044542664560804e-05,
"loss": 0.4955,
"step": 9950
},
{
"epoch": 1.1607975746268657,
"grad_norm": 0.4377330122208343,
"learning_rate": 4.6039628050749066e-05,
"loss": 0.4412,
"step": 9955
},
{
"epoch": 1.1613805970149254,
"grad_norm": 0.44907976595838806,
"learning_rate": 4.6034710680272274e-05,
"loss": 0.4628,
"step": 9960
},
{
"epoch": 1.161963619402985,
"grad_norm": 0.4848892879121351,
"learning_rate": 4.6029790553861594e-05,
"loss": 0.5327,
"step": 9965
},
{
"epoch": 1.1625466417910448,
"grad_norm": 0.47865209843468254,
"learning_rate": 4.602486767224858e-05,
"loss": 0.4881,
"step": 9970
},
{
"epoch": 1.1631296641791045,
"grad_norm": 0.5065935109536762,
"learning_rate": 4.601994203616525e-05,
"loss": 0.5038,
"step": 9975
},
{
"epoch": 1.163712686567164,
"grad_norm": 0.5058923277966703,
"learning_rate": 4.601501364634397e-05,
"loss": 0.4909,
"step": 9980
},
{
"epoch": 1.164295708955224,
"grad_norm": 0.5358765833523091,
"learning_rate": 4.601008250351756e-05,
"loss": 0.54,
"step": 9985
},
{
"epoch": 1.1648787313432836,
"grad_norm": 0.44762389969899724,
"learning_rate": 4.600514860841923e-05,
"loss": 0.503,
"step": 9990
},
{
"epoch": 1.1654617537313432,
"grad_norm": 0.5131406173451447,
"learning_rate": 4.6000211961782605e-05,
"loss": 0.5139,
"step": 9995
},
{
"epoch": 1.166044776119403,
"grad_norm": 0.5104173496537336,
"learning_rate": 4.599527256434171e-05,
"loss": 0.48,
"step": 10000
},
{
"epoch": 1.1666277985074627,
"grad_norm": 0.4886514119195971,
"learning_rate": 4.599033041683099e-05,
"loss": 0.5176,
"step": 10005
},
{
"epoch": 1.1672108208955223,
"grad_norm": 0.42725696967219473,
"learning_rate": 4.598538551998531e-05,
"loss": 0.4679,
"step": 10010
},
{
"epoch": 1.1677938432835822,
"grad_norm": 0.43566282015581653,
"learning_rate": 4.59804378745399e-05,
"loss": 0.4853,
"step": 10015
},
{
"epoch": 1.1683768656716418,
"grad_norm": 0.5137803755067021,
"learning_rate": 4.597548748123046e-05,
"loss": 0.5042,
"step": 10020
},
{
"epoch": 1.1689598880597014,
"grad_norm": 0.4609910866389386,
"learning_rate": 4.597053434079303e-05,
"loss": 0.4835,
"step": 10025
},
{
"epoch": 1.1695429104477613,
"grad_norm": 0.48248991604882147,
"learning_rate": 4.596557845396412e-05,
"loss": 0.509,
"step": 10030
},
{
"epoch": 1.170125932835821,
"grad_norm": 0.4857742343614341,
"learning_rate": 4.596061982148062e-05,
"loss": 0.478,
"step": 10035
},
{
"epoch": 1.1707089552238805,
"grad_norm": 0.4685134256969988,
"learning_rate": 4.595565844407982e-05,
"loss": 0.5073,
"step": 10040
},
{
"epoch": 1.1712919776119404,
"grad_norm": 0.4552557913062714,
"learning_rate": 4.5950694322499444e-05,
"loss": 0.5152,
"step": 10045
},
{
"epoch": 1.171875,
"grad_norm": 0.4901456457078471,
"learning_rate": 4.59457274574776e-05,
"loss": 0.4881,
"step": 10050
},
{
"epoch": 1.1724580223880596,
"grad_norm": 0.4775643180270164,
"learning_rate": 4.5940757849752805e-05,
"loss": 0.4607,
"step": 10055
},
{
"epoch": 1.1730410447761195,
"grad_norm": 0.46063185695387576,
"learning_rate": 4.5935785500064014e-05,
"loss": 0.487,
"step": 10060
},
{
"epoch": 1.173624067164179,
"grad_norm": 0.4746823163282632,
"learning_rate": 4.5930810409150556e-05,
"loss": 0.5182,
"step": 10065
},
{
"epoch": 1.1742070895522387,
"grad_norm": 0.43843195262337814,
"learning_rate": 4.5925832577752175e-05,
"loss": 0.4881,
"step": 10070
},
{
"epoch": 1.1747901119402986,
"grad_norm": 0.4685303294788509,
"learning_rate": 4.592085200660903e-05,
"loss": 0.4951,
"step": 10075
},
{
"epoch": 1.1753731343283582,
"grad_norm": 0.4749093447944963,
"learning_rate": 4.5915868696461685e-05,
"loss": 0.5261,
"step": 10080
},
{
"epoch": 1.1759561567164178,
"grad_norm": 0.4795736672954712,
"learning_rate": 4.591088264805111e-05,
"loss": 0.5027,
"step": 10085
},
{
"epoch": 1.1765391791044777,
"grad_norm": 0.48217126639017427,
"learning_rate": 4.590589386211869e-05,
"loss": 0.4937,
"step": 10090
},
{
"epoch": 1.1771222014925373,
"grad_norm": 0.420400842266455,
"learning_rate": 4.5900902339406195e-05,
"loss": 0.4879,
"step": 10095
},
{
"epoch": 1.177705223880597,
"grad_norm": 0.4889175318908055,
"learning_rate": 4.589590808065583e-05,
"loss": 0.4891,
"step": 10100
},
{
"epoch": 1.1782882462686568,
"grad_norm": 0.4699220474921735,
"learning_rate": 4.5890911086610184e-05,
"loss": 0.4997,
"step": 10105
},
{
"epoch": 1.1788712686567164,
"grad_norm": 0.4499121851481335,
"learning_rate": 4.588591135801227e-05,
"loss": 0.4942,
"step": 10110
},
{
"epoch": 1.179454291044776,
"grad_norm": 0.48039361463042335,
"learning_rate": 4.588090889560549e-05,
"loss": 0.5262,
"step": 10115
},
{
"epoch": 1.180037313432836,
"grad_norm": 0.4870244514079405,
"learning_rate": 4.587590370013367e-05,
"loss": 0.5012,
"step": 10120
},
{
"epoch": 1.1806203358208955,
"grad_norm": 0.4359608570010937,
"learning_rate": 4.587089577234104e-05,
"loss": 0.4856,
"step": 10125
},
{
"epoch": 1.1812033582089552,
"grad_norm": 0.466445375839257,
"learning_rate": 4.5865885112972216e-05,
"loss": 0.4906,
"step": 10130
},
{
"epoch": 1.181786380597015,
"grad_norm": 0.42268739888448525,
"learning_rate": 4.5860871722772246e-05,
"loss": 0.4782,
"step": 10135
},
{
"epoch": 1.1823694029850746,
"grad_norm": 0.4966303201182526,
"learning_rate": 4.585585560248657e-05,
"loss": 0.4948,
"step": 10140
},
{
"epoch": 1.1829524253731343,
"grad_norm": 0.49723599768519844,
"learning_rate": 4.5850836752861034e-05,
"loss": 0.4967,
"step": 10145
},
{
"epoch": 1.1835354477611941,
"grad_norm": 0.47220426148758693,
"learning_rate": 4.58458151746419e-05,
"loss": 0.5122,
"step": 10150
},
{
"epoch": 1.1841184701492538,
"grad_norm": 0.44873656844364723,
"learning_rate": 4.584079086857582e-05,
"loss": 0.4919,
"step": 10155
},
{
"epoch": 1.1847014925373134,
"grad_norm": 0.7909442158986997,
"learning_rate": 4.5835763835409864e-05,
"loss": 0.512,
"step": 10160
},
{
"epoch": 1.1852845149253732,
"grad_norm": 0.4579010837343018,
"learning_rate": 4.583073407589151e-05,
"loss": 0.4753,
"step": 10165
},
{
"epoch": 1.1858675373134329,
"grad_norm": 0.47469471616317205,
"learning_rate": 4.5825701590768625e-05,
"loss": 0.4933,
"step": 10170
},
{
"epoch": 1.1864505597014925,
"grad_norm": 0.4495999689572246,
"learning_rate": 4.582066638078949e-05,
"loss": 0.4934,
"step": 10175
},
{
"epoch": 1.1870335820895521,
"grad_norm": 0.4796448933384789,
"learning_rate": 4.58156284467028e-05,
"loss": 0.4762,
"step": 10180
},
{
"epoch": 1.187616604477612,
"grad_norm": 0.46714972366630153,
"learning_rate": 4.5810587789257646e-05,
"loss": 0.4858,
"step": 10185
},
{
"epoch": 1.1881996268656716,
"grad_norm": 0.4889030838131866,
"learning_rate": 4.5805544409203535e-05,
"loss": 0.5097,
"step": 10190
},
{
"epoch": 1.1887826492537314,
"grad_norm": 0.47705359892658183,
"learning_rate": 4.5800498307290344e-05,
"loss": 0.509,
"step": 10195
},
{
"epoch": 1.189365671641791,
"grad_norm": 0.4612757329558082,
"learning_rate": 4.579544948426841e-05,
"loss": 0.4907,
"step": 10200
},
{
"epoch": 1.1899486940298507,
"grad_norm": 0.4392831600817961,
"learning_rate": 4.579039794088842e-05,
"loss": 0.5172,
"step": 10205
},
{
"epoch": 1.1905317164179103,
"grad_norm": 0.5130394151474353,
"learning_rate": 4.57853436779015e-05,
"loss": 0.5189,
"step": 10210
},
{
"epoch": 1.1911147388059702,
"grad_norm": 0.4736920577937518,
"learning_rate": 4.578028669605918e-05,
"loss": 0.4695,
"step": 10215
},
{
"epoch": 1.1916977611940298,
"grad_norm": 0.5675131307737508,
"learning_rate": 4.577522699611336e-05,
"loss": 0.4985,
"step": 10220
},
{
"epoch": 1.1922807835820897,
"grad_norm": 0.4390341606630438,
"learning_rate": 4.577016457881639e-05,
"loss": 0.5118,
"step": 10225
},
{
"epoch": 1.1928638059701493,
"grad_norm": 0.45624624457570856,
"learning_rate": 4.576509944492101e-05,
"loss": 0.487,
"step": 10230
},
{
"epoch": 1.193446828358209,
"grad_norm": 0.41894433880336407,
"learning_rate": 4.5760031595180325e-05,
"loss": 0.4869,
"step": 10235
},
{
"epoch": 1.1940298507462686,
"grad_norm": 0.436370405965647,
"learning_rate": 4.57549610303479e-05,
"loss": 0.4878,
"step": 10240
},
{
"epoch": 1.1946128731343284,
"grad_norm": 0.49373600917312327,
"learning_rate": 4.574988775117767e-05,
"loss": 0.5055,
"step": 10245
},
{
"epoch": 1.195195895522388,
"grad_norm": 0.5090184054460075,
"learning_rate": 4.5744811758424e-05,
"loss": 0.5114,
"step": 10250
},
{
"epoch": 1.1957789179104479,
"grad_norm": 0.4521890891822228,
"learning_rate": 4.573973305284162e-05,
"loss": 0.4756,
"step": 10255
},
{
"epoch": 1.1963619402985075,
"grad_norm": 0.48661728690265815,
"learning_rate": 4.573465163518569e-05,
"loss": 0.4901,
"step": 10260
},
{
"epoch": 1.1969449626865671,
"grad_norm": 0.4562515996289815,
"learning_rate": 4.572956750621178e-05,
"loss": 0.4752,
"step": 10265
},
{
"epoch": 1.1975279850746268,
"grad_norm": 0.48159859341372235,
"learning_rate": 4.572448066667584e-05,
"loss": 0.5142,
"step": 10270
},
{
"epoch": 1.1981110074626866,
"grad_norm": 0.4175615384612034,
"learning_rate": 4.571939111733423e-05,
"loss": 0.4917,
"step": 10275
},
{
"epoch": 1.1986940298507462,
"grad_norm": 0.48514238856571956,
"learning_rate": 4.571429885894373e-05,
"loss": 0.5033,
"step": 10280
},
{
"epoch": 1.1992770522388059,
"grad_norm": 0.4700533064137317,
"learning_rate": 4.5709203892261506e-05,
"loss": 0.5039,
"step": 10285
},
{
"epoch": 1.1998600746268657,
"grad_norm": 0.5155371027252977,
"learning_rate": 4.5704106218045124e-05,
"loss": 0.5267,
"step": 10290
},
{
"epoch": 1.2004430970149254,
"grad_norm": 0.4258916802960759,
"learning_rate": 4.569900583705257e-05,
"loss": 0.5128,
"step": 10295
},
{
"epoch": 1.201026119402985,
"grad_norm": 0.5218094530712158,
"learning_rate": 4.569390275004221e-05,
"loss": 0.4913,
"step": 10300
},
{
"epoch": 1.2016091417910448,
"grad_norm": 0.458383381872571,
"learning_rate": 4.568879695777283e-05,
"loss": 0.5071,
"step": 10305
},
{
"epoch": 1.2021921641791045,
"grad_norm": 0.447897704047899,
"learning_rate": 4.568368846100363e-05,
"loss": 0.4877,
"step": 10310
},
{
"epoch": 1.202775186567164,
"grad_norm": 0.4595544275808344,
"learning_rate": 4.567857726049415e-05,
"loss": 0.4911,
"step": 10315
},
{
"epoch": 1.203358208955224,
"grad_norm": 0.45837361012593975,
"learning_rate": 4.567346335700442e-05,
"loss": 0.5353,
"step": 10320
},
{
"epoch": 1.2039412313432836,
"grad_norm": 0.44439826613487565,
"learning_rate": 4.56683467512948e-05,
"loss": 0.4808,
"step": 10325
},
{
"epoch": 1.2045242537313432,
"grad_norm": 0.46198420918559796,
"learning_rate": 4.5663227444126114e-05,
"loss": 0.524,
"step": 10330
},
{
"epoch": 1.205107276119403,
"grad_norm": 0.45273321897247865,
"learning_rate": 4.565810543625952e-05,
"loss": 0.4827,
"step": 10335
},
{
"epoch": 1.2056902985074627,
"grad_norm": 0.4130813589196938,
"learning_rate": 4.565298072845662e-05,
"loss": 0.4525,
"step": 10340
},
{
"epoch": 1.2062733208955223,
"grad_norm": 0.4501895517862483,
"learning_rate": 4.5647853321479414e-05,
"loss": 0.4815,
"step": 10345
},
{
"epoch": 1.2068563432835822,
"grad_norm": 0.4850929960512088,
"learning_rate": 4.564272321609031e-05,
"loss": 0.4847,
"step": 10350
},
{
"epoch": 1.2074393656716418,
"grad_norm": 0.47342463711706323,
"learning_rate": 4.563759041305207e-05,
"loss": 0.5328,
"step": 10355
},
{
"epoch": 1.2080223880597014,
"grad_norm": 0.4591790988243215,
"learning_rate": 4.563245491312793e-05,
"loss": 0.5116,
"step": 10360
},
{
"epoch": 1.2086054104477613,
"grad_norm": 0.48391887993801336,
"learning_rate": 4.562731671708147e-05,
"loss": 0.475,
"step": 10365
},
{
"epoch": 1.209188432835821,
"grad_norm": 0.5940498487621698,
"learning_rate": 4.5622175825676695e-05,
"loss": 0.4691,
"step": 10370
},
{
"epoch": 1.2097714552238805,
"grad_norm": 0.45160586473877545,
"learning_rate": 4.5617032239678016e-05,
"loss": 0.4887,
"step": 10375
},
{
"epoch": 1.2103544776119404,
"grad_norm": 0.5472177361751651,
"learning_rate": 4.5611885959850216e-05,
"loss": 0.5053,
"step": 10380
},
{
"epoch": 1.2109375,
"grad_norm": 0.4361554855779335,
"learning_rate": 4.5606736986958504e-05,
"loss": 0.4706,
"step": 10385
},
{
"epoch": 1.2115205223880596,
"grad_norm": 0.48091608079580567,
"learning_rate": 4.560158532176849e-05,
"loss": 0.4846,
"step": 10390
},
{
"epoch": 1.2121035447761195,
"grad_norm": 0.5277938909221369,
"learning_rate": 4.559643096504618e-05,
"loss": 0.4529,
"step": 10395
},
{
"epoch": 1.212686567164179,
"grad_norm": 0.47110973401201595,
"learning_rate": 4.559127391755796e-05,
"loss": 0.4843,
"step": 10400
},
{
"epoch": 1.2132695895522387,
"grad_norm": 0.46646043048754804,
"learning_rate": 4.558611418007065e-05,
"loss": 0.5225,
"step": 10405
},
{
"epoch": 1.2138526119402986,
"grad_norm": 0.4117714748940162,
"learning_rate": 4.558095175335145e-05,
"loss": 0.4972,
"step": 10410
},
{
"epoch": 1.2144356343283582,
"grad_norm": 0.473548178220035,
"learning_rate": 4.557578663816796e-05,
"loss": 0.4888,
"step": 10415
},
{
"epoch": 1.2150186567164178,
"grad_norm": 0.44588850897264254,
"learning_rate": 4.557061883528818e-05,
"loss": 0.4889,
"step": 10420
},
{
"epoch": 1.2156016791044777,
"grad_norm": 0.4357468417538078,
"learning_rate": 4.556544834548052e-05,
"loss": 0.4935,
"step": 10425
},
{
"epoch": 1.2161847014925373,
"grad_norm": 0.47214744820744703,
"learning_rate": 4.5560275169513786e-05,
"loss": 0.4969,
"step": 10430
},
{
"epoch": 1.216767723880597,
"grad_norm": 0.5052146704131883,
"learning_rate": 4.5555099308157163e-05,
"loss": 0.4797,
"step": 10435
},
{
"epoch": 1.2173507462686568,
"grad_norm": 0.5143329850027023,
"learning_rate": 4.554992076218026e-05,
"loss": 0.4744,
"step": 10440
},
{
"epoch": 1.2179337686567164,
"grad_norm": 0.5905636689742819,
"learning_rate": 4.554473953235309e-05,
"loss": 0.4858,
"step": 10445
},
{
"epoch": 1.218516791044776,
"grad_norm": 0.4868778049016568,
"learning_rate": 4.553955561944603e-05,
"loss": 0.4674,
"step": 10450
},
{
"epoch": 1.219099813432836,
"grad_norm": 0.4735704731328373,
"learning_rate": 4.5534369024229903e-05,
"loss": 0.4945,
"step": 10455
},
{
"epoch": 1.2196828358208955,
"grad_norm": 0.4217849941090126,
"learning_rate": 4.552917974747588e-05,
"loss": 0.5251,
"step": 10460
},
{
"epoch": 1.2202658582089552,
"grad_norm": 0.4381496438873959,
"learning_rate": 4.552398778995558e-05,
"loss": 0.4855,
"step": 10465
},
{
"epoch": 1.220848880597015,
"grad_norm": 0.4682518660001828,
"learning_rate": 4.551879315244098e-05,
"loss": 0.4612,
"step": 10470
},
{
"epoch": 1.2214319029850746,
"grad_norm": 0.4711869099835559,
"learning_rate": 4.551359583570448e-05,
"loss": 0.4624,
"step": 10475
},
{
"epoch": 1.2220149253731343,
"grad_norm": 0.452571746667093,
"learning_rate": 4.5508395840518884e-05,
"loss": 0.4821,
"step": 10480
},
{
"epoch": 1.2225979477611941,
"grad_norm": 0.49725953336704015,
"learning_rate": 4.550319316765735e-05,
"loss": 0.4707,
"step": 10485
},
{
"epoch": 1.2231809701492538,
"grad_norm": 0.484195499245689,
"learning_rate": 4.549798781789349e-05,
"loss": 0.4914,
"step": 10490
},
{
"epoch": 1.2237639925373134,
"grad_norm": 0.47164140799331555,
"learning_rate": 4.5492779792001286e-05,
"loss": 0.5137,
"step": 10495
},
{
"epoch": 1.2243470149253732,
"grad_norm": 0.4764823439535703,
"learning_rate": 4.548756909075511e-05,
"loss": 0.4996,
"step": 10500
},
{
"epoch": 1.2249300373134329,
"grad_norm": 0.4481478128548299,
"learning_rate": 4.5482355714929764e-05,
"loss": 0.4917,
"step": 10505
},
{
"epoch": 1.2255130597014925,
"grad_norm": 0.45412757674717086,
"learning_rate": 4.5477139665300414e-05,
"loss": 0.4782,
"step": 10510
},
{
"epoch": 1.2260960820895521,
"grad_norm": 0.510120546472683,
"learning_rate": 4.5471920942642634e-05,
"loss": 0.4597,
"step": 10515
},
{
"epoch": 1.226679104477612,
"grad_norm": 0.472860169073306,
"learning_rate": 4.5466699547732405e-05,
"loss": 0.5071,
"step": 10520
},
{
"epoch": 1.2272621268656716,
"grad_norm": 0.48174398275339125,
"learning_rate": 4.5461475481346086e-05,
"loss": 0.5015,
"step": 10525
},
{
"epoch": 1.2278451492537314,
"grad_norm": 0.45223858027921016,
"learning_rate": 4.545624874426047e-05,
"loss": 0.4694,
"step": 10530
},
{
"epoch": 1.228428171641791,
"grad_norm": 0.4703622316483385,
"learning_rate": 4.54510193372527e-05,
"loss": 0.5083,
"step": 10535
},
{
"epoch": 1.2290111940298507,
"grad_norm": 0.4766899395190844,
"learning_rate": 4.544578726110035e-05,
"loss": 0.4635,
"step": 10540
},
{
"epoch": 1.2295942164179103,
"grad_norm": 0.4805029162802611,
"learning_rate": 4.544055251658137e-05,
"loss": 0.5375,
"step": 10545
},
{
"epoch": 1.2301772388059702,
"grad_norm": 0.4455488814122577,
"learning_rate": 4.5435315104474124e-05,
"loss": 0.4855,
"step": 10550
},
{
"epoch": 1.2307602611940298,
"grad_norm": 0.42671789157033496,
"learning_rate": 4.5430075025557365e-05,
"loss": 0.5019,
"step": 10555
},
{
"epoch": 1.2313432835820897,
"grad_norm": 0.46701158462531245,
"learning_rate": 4.5424832280610245e-05,
"loss": 0.5012,
"step": 10560
},
{
"epoch": 1.2319263059701493,
"grad_norm": 0.5297688795382418,
"learning_rate": 4.5419586870412306e-05,
"loss": 0.4908,
"step": 10565
},
{
"epoch": 1.232509328358209,
"grad_norm": 0.4204261464822169,
"learning_rate": 4.541433879574348e-05,
"loss": 0.4693,
"step": 10570
},
{
"epoch": 1.2330923507462686,
"grad_norm": 0.4673047492978749,
"learning_rate": 4.540908805738412e-05,
"loss": 0.4689,
"step": 10575
},
{
"epoch": 1.2336753731343284,
"grad_norm": 0.44842552442527106,
"learning_rate": 4.540383465611496e-05,
"loss": 0.4921,
"step": 10580
},
{
"epoch": 1.234258395522388,
"grad_norm": 0.46336131437386063,
"learning_rate": 4.5398578592717135e-05,
"loss": 0.5065,
"step": 10585
},
{
"epoch": 1.2348414179104479,
"grad_norm": 0.4759828002351157,
"learning_rate": 4.539331986797215e-05,
"loss": 0.5122,
"step": 10590
},
{
"epoch": 1.2354244402985075,
"grad_norm": 0.4717537952461409,
"learning_rate": 4.5388058482661944e-05,
"loss": 0.5031,
"step": 10595
},
{
"epoch": 1.2360074626865671,
"grad_norm": 0.4609477364032761,
"learning_rate": 4.5382794437568824e-05,
"loss": 0.487,
"step": 10600
},
{
"epoch": 1.2365904850746268,
"grad_norm": 0.5120862465684253,
"learning_rate": 4.537752773347552e-05,
"loss": 0.4904,
"step": 10605
},
{
"epoch": 1.2371735074626866,
"grad_norm": 0.45861223756305225,
"learning_rate": 4.537225837116512e-05,
"loss": 0.5004,
"step": 10610
},
{
"epoch": 1.2377565298507462,
"grad_norm": 0.47356352919508377,
"learning_rate": 4.5366986351421145e-05,
"loss": 0.4942,
"step": 10615
},
{
"epoch": 1.2383395522388059,
"grad_norm": 0.4919271378489869,
"learning_rate": 4.5361711675027484e-05,
"loss": 0.4831,
"step": 10620
},
{
"epoch": 1.2389225746268657,
"grad_norm": 0.4637408665064941,
"learning_rate": 4.5356434342768434e-05,
"loss": 0.4851,
"step": 10625
},
{
"epoch": 1.2395055970149254,
"grad_norm": 0.4879180836517928,
"learning_rate": 4.535115435542868e-05,
"loss": 0.491,
"step": 10630
},
{
"epoch": 1.240088619402985,
"grad_norm": 0.484147000323564,
"learning_rate": 4.5345871713793306e-05,
"loss": 0.532,
"step": 10635
},
{
"epoch": 1.2406716417910448,
"grad_norm": 0.48885676700578345,
"learning_rate": 4.53405864186478e-05,
"loss": 0.4839,
"step": 10640
},
{
"epoch": 1.2412546641791045,
"grad_norm": 0.500355902387549,
"learning_rate": 4.533529847077803e-05,
"loss": 0.4965,
"step": 10645
},
{
"epoch": 1.241837686567164,
"grad_norm": 0.45464876801725207,
"learning_rate": 4.5330007870970255e-05,
"loss": 0.4859,
"step": 10650
},
{
"epoch": 1.242420708955224,
"grad_norm": 0.41211951322449497,
"learning_rate": 4.532471462001114e-05,
"loss": 0.5091,
"step": 10655
},
{
"epoch": 1.2430037313432836,
"grad_norm": 0.4532085017482041,
"learning_rate": 4.531941871868775e-05,
"loss": 0.5016,
"step": 10660
},
{
"epoch": 1.2435867537313432,
"grad_norm": 0.5200222182217336,
"learning_rate": 4.531412016778752e-05,
"loss": 0.494,
"step": 10665
},
{
"epoch": 1.244169776119403,
"grad_norm": 0.44998083416964635,
"learning_rate": 4.530881896809831e-05,
"loss": 0.4777,
"step": 10670
},
{
"epoch": 1.2447527985074627,
"grad_norm": 0.46243724465256153,
"learning_rate": 4.530351512040834e-05,
"loss": 0.4682,
"step": 10675
},
{
"epoch": 1.2453358208955223,
"grad_norm": 0.4463317078925723,
"learning_rate": 4.5298208625506253e-05,
"loss": 0.5188,
"step": 10680
},
{
"epoch": 1.2459188432835822,
"grad_norm": 0.4957783469036879,
"learning_rate": 4.5292899484181074e-05,
"loss": 0.5024,
"step": 10685
},
{
"epoch": 1.2465018656716418,
"grad_norm": 0.47347227953314347,
"learning_rate": 4.5287587697222215e-05,
"loss": 0.4755,
"step": 10690
},
{
"epoch": 1.2470848880597014,
"grad_norm": 0.46933916348082055,
"learning_rate": 4.528227326541949e-05,
"loss": 0.5216,
"step": 10695
},
{
"epoch": 1.2476679104477613,
"grad_norm": 0.4274168227368018,
"learning_rate": 4.527695618956312e-05,
"loss": 0.4911,
"step": 10700
},
{
"epoch": 1.248250932835821,
"grad_norm": 0.47907368979204124,
"learning_rate": 4.527163647044367e-05,
"loss": 0.4884,
"step": 10705
},
{
"epoch": 1.2488339552238805,
"grad_norm": 0.46085266857825197,
"learning_rate": 4.5266314108852166e-05,
"loss": 0.5122,
"step": 10710
},
{
"epoch": 1.2494169776119404,
"grad_norm": 0.45342521267091646,
"learning_rate": 4.526098910557996e-05,
"loss": 0.5044,
"step": 10715
},
{
"epoch": 1.25,
"grad_norm": 0.5852842506751587,
"learning_rate": 4.5255661461418854e-05,
"loss": 0.5238,
"step": 10720
},
{
"epoch": 1.2505830223880596,
"grad_norm": 0.4673897355491722,
"learning_rate": 4.5250331177161014e-05,
"loss": 0.4876,
"step": 10725
},
{
"epoch": 1.2511660447761195,
"grad_norm": 0.45628166086918426,
"learning_rate": 4.5244998253598994e-05,
"loss": 0.4871,
"step": 10730
},
{
"epoch": 1.251749067164179,
"grad_norm": 0.47124101261746165,
"learning_rate": 4.5239662691525744e-05,
"loss": 0.5156,
"step": 10735
},
{
"epoch": 1.2523320895522387,
"grad_norm": 0.44725519298792205,
"learning_rate": 4.5234324491734624e-05,
"loss": 0.4716,
"step": 10740
},
{
"epoch": 1.2529151119402986,
"grad_norm": 0.42690219529381024,
"learning_rate": 4.522898365501938e-05,
"loss": 0.4991,
"step": 10745
},
{
"epoch": 1.2534981343283582,
"grad_norm": 0.4916189926473275,
"learning_rate": 4.5223640182174115e-05,
"loss": 0.5192,
"step": 10750
},
{
"epoch": 1.2540811567164178,
"grad_norm": 0.47347920108375546,
"learning_rate": 4.5218294073993374e-05,
"loss": 0.4843,
"step": 10755
},
{
"epoch": 1.2546641791044777,
"grad_norm": 0.5188258241202645,
"learning_rate": 4.521294533127206e-05,
"loss": 0.5152,
"step": 10760
},
{
"epoch": 1.2552472014925373,
"grad_norm": 0.4391126986556853,
"learning_rate": 4.5207593954805494e-05,
"loss": 0.4726,
"step": 10765
},
{
"epoch": 1.255830223880597,
"grad_norm": 0.4608047643671716,
"learning_rate": 4.520223994538937e-05,
"loss": 0.4968,
"step": 10770
},
{
"epoch": 1.2564132462686568,
"grad_norm": 0.5149266987509121,
"learning_rate": 4.519688330381976e-05,
"loss": 0.5188,
"step": 10775
},
{
"epoch": 1.2569962686567164,
"grad_norm": 0.4408037918106659,
"learning_rate": 4.519152403089317e-05,
"loss": 0.4712,
"step": 10780
},
{
"epoch": 1.257579291044776,
"grad_norm": 0.48078540476076054,
"learning_rate": 4.518616212740647e-05,
"loss": 0.5117,
"step": 10785
},
{
"epoch": 1.2581623134328357,
"grad_norm": 0.4937330498447094,
"learning_rate": 4.51807975941569e-05,
"loss": 0.4937,
"step": 10790
},
{
"epoch": 1.2587453358208955,
"grad_norm": 0.3956058070444886,
"learning_rate": 4.517543043194214e-05,
"loss": 0.448,
"step": 10795
},
{
"epoch": 1.2593283582089552,
"grad_norm": 0.46438661170530343,
"learning_rate": 4.517006064156023e-05,
"loss": 0.4839,
"step": 10800
},
{
"epoch": 1.259911380597015,
"grad_norm": 0.4576805182366751,
"learning_rate": 4.516468822380959e-05,
"loss": 0.5221,
"step": 10805
},
{
"epoch": 1.2604944029850746,
"grad_norm": 0.424073212385048,
"learning_rate": 4.515931317948907e-05,
"loss": 0.4614,
"step": 10810
},
{
"epoch": 1.2610774253731343,
"grad_norm": 0.4358863186762239,
"learning_rate": 4.515393550939787e-05,
"loss": 0.4648,
"step": 10815
},
{
"epoch": 1.261660447761194,
"grad_norm": 0.45400319344417495,
"learning_rate": 4.5148555214335616e-05,
"loss": 0.5128,
"step": 10820
},
{
"epoch": 1.2622434701492538,
"grad_norm": 0.4441332948804677,
"learning_rate": 4.514317229510228e-05,
"loss": 0.468,
"step": 10825
},
{
"epoch": 1.2628264925373134,
"grad_norm": 0.47556452899330476,
"learning_rate": 4.5137786752498285e-05,
"loss": 0.4899,
"step": 10830
},
{
"epoch": 1.2634095149253732,
"grad_norm": 0.4358375800095142,
"learning_rate": 4.513239858732438e-05,
"loss": 0.4863,
"step": 10835
},
{
"epoch": 1.2639925373134329,
"grad_norm": 0.4571425793619909,
"learning_rate": 4.512700780038174e-05,
"loss": 0.4896,
"step": 10840
},
{
"epoch": 1.2645755597014925,
"grad_norm": 0.5151550921023409,
"learning_rate": 4.5121614392471934e-05,
"loss": 0.4849,
"step": 10845
},
{
"epoch": 1.2651585820895521,
"grad_norm": 0.4603106071865496,
"learning_rate": 4.5116218364396904e-05,
"loss": 0.4666,
"step": 10850
},
{
"epoch": 1.265741604477612,
"grad_norm": 0.4211185066790496,
"learning_rate": 4.511081971695899e-05,
"loss": 0.4641,
"step": 10855
},
{
"epoch": 1.2663246268656716,
"grad_norm": 0.4678760624094688,
"learning_rate": 4.510541845096091e-05,
"loss": 0.4711,
"step": 10860
},
{
"epoch": 1.2669076492537314,
"grad_norm": 0.47620077865417704,
"learning_rate": 4.510001456720579e-05,
"loss": 0.4807,
"step": 10865
},
{
"epoch": 1.267490671641791,
"grad_norm": 0.46355439011575994,
"learning_rate": 4.509460806649714e-05,
"loss": 0.4881,
"step": 10870
},
{
"epoch": 1.2680736940298507,
"grad_norm": 0.4914208310417759,
"learning_rate": 4.508919894963884e-05,
"loss": 0.4937,
"step": 10875
},
{
"epoch": 1.2686567164179103,
"grad_norm": 0.4305195635584616,
"learning_rate": 4.5083787217435175e-05,
"loss": 0.4902,
"step": 10880
},
{
"epoch": 1.2692397388059702,
"grad_norm": 0.44759494114245063,
"learning_rate": 4.507837287069083e-05,
"loss": 0.4961,
"step": 10885
},
{
"epoch": 1.2698227611940298,
"grad_norm": 0.4440267264043612,
"learning_rate": 4.507295591021087e-05,
"loss": 0.509,
"step": 10890
},
{
"epoch": 1.2704057835820897,
"grad_norm": 0.43935234739243484,
"learning_rate": 4.5067536336800724e-05,
"loss": 0.4726,
"step": 10895
},
{
"epoch": 1.2709888059701493,
"grad_norm": 0.4573323836385464,
"learning_rate": 4.506211415126624e-05,
"loss": 0.5018,
"step": 10900
},
{
"epoch": 1.271571828358209,
"grad_norm": 0.46045982197691526,
"learning_rate": 4.5056689354413664e-05,
"loss": 0.4728,
"step": 10905
},
{
"epoch": 1.2721548507462686,
"grad_norm": 0.4630717093874329,
"learning_rate": 4.505126194704958e-05,
"loss": 0.4958,
"step": 10910
},
{
"epoch": 1.2727378731343284,
"grad_norm": 0.48470057016551354,
"learning_rate": 4.504583192998101e-05,
"loss": 0.5273,
"step": 10915
},
{
"epoch": 1.273320895522388,
"grad_norm": 0.45280566005549905,
"learning_rate": 4.504039930401535e-05,
"loss": 0.4964,
"step": 10920
},
{
"epoch": 1.2739039179104479,
"grad_norm": 0.5037372516068997,
"learning_rate": 4.503496406996037e-05,
"loss": 0.4999,
"step": 10925
},
{
"epoch": 1.2744869402985075,
"grad_norm": 0.47606837622249965,
"learning_rate": 4.5029526228624226e-05,
"loss": 0.5029,
"step": 10930
},
{
"epoch": 1.2750699626865671,
"grad_norm": 0.5703203195144877,
"learning_rate": 4.50240857808155e-05,
"loss": 0.538,
"step": 10935
},
{
"epoch": 1.2756529850746268,
"grad_norm": 0.45251604419229047,
"learning_rate": 4.501864272734311e-05,
"loss": 0.5015,
"step": 10940
},
{
"epoch": 1.2762360074626866,
"grad_norm": 0.49655362487006866,
"learning_rate": 4.50131970690164e-05,
"loss": 0.4847,
"step": 10945
},
{
"epoch": 1.2768190298507462,
"grad_norm": 0.4613013724647206,
"learning_rate": 4.500774880664508e-05,
"loss": 0.4787,
"step": 10950
},
{
"epoch": 1.277402052238806,
"grad_norm": 0.4302691188658478,
"learning_rate": 4.500229794103925e-05,
"loss": 0.5333,
"step": 10955
},
{
"epoch": 1.2779850746268657,
"grad_norm": 0.42502232464484513,
"learning_rate": 4.4996844473009425e-05,
"loss": 0.491,
"step": 10960
},
{
"epoch": 1.2785680970149254,
"grad_norm": 0.40601973604606284,
"learning_rate": 4.499138840336646e-05,
"loss": 0.4629,
"step": 10965
},
{
"epoch": 1.279151119402985,
"grad_norm": 0.45669189552203826,
"learning_rate": 4.498592973292162e-05,
"loss": 0.4843,
"step": 10970
},
{
"epoch": 1.2797341417910448,
"grad_norm": 0.4340655439220024,
"learning_rate": 4.498046846248658e-05,
"loss": 0.4678,
"step": 10975
},
{
"epoch": 1.2803171641791045,
"grad_norm": 0.4729483198135297,
"learning_rate": 4.497500459287335e-05,
"loss": 0.5303,
"step": 10980
},
{
"epoch": 1.2809001865671643,
"grad_norm": 0.4553839692139655,
"learning_rate": 4.496953812489438e-05,
"loss": 0.4713,
"step": 10985
},
{
"epoch": 1.281483208955224,
"grad_norm": 0.43086952416463076,
"learning_rate": 4.496406905936246e-05,
"loss": 0.4945,
"step": 10990
},
{
"epoch": 1.2820662313432836,
"grad_norm": 0.4957276784882462,
"learning_rate": 4.49585973970908e-05,
"loss": 0.4715,
"step": 10995
},
{
"epoch": 1.2826492537313432,
"grad_norm": 0.44036399286463346,
"learning_rate": 4.4953123138892984e-05,
"loss": 0.4556,
"step": 11000
},
{
"epoch": 1.283232276119403,
"grad_norm": 0.40191787688124264,
"learning_rate": 4.4947646285582974e-05,
"loss": 0.4984,
"step": 11005
},
{
"epoch": 1.2838152985074627,
"grad_norm": 0.5236556766781192,
"learning_rate": 4.4942166837975134e-05,
"loss": 0.5033,
"step": 11010
},
{
"epoch": 1.2843983208955223,
"grad_norm": 0.4324916399134003,
"learning_rate": 4.49366847968842e-05,
"loss": 0.5058,
"step": 11015
},
{
"epoch": 1.2849813432835822,
"grad_norm": 0.416311459433444,
"learning_rate": 4.4931200163125306e-05,
"loss": 0.4637,
"step": 11020
},
{
"epoch": 1.2855643656716418,
"grad_norm": 0.41026736261558155,
"learning_rate": 4.492571293751395e-05,
"loss": 0.4862,
"step": 11025
},
{
"epoch": 1.2861473880597014,
"grad_norm": 0.44497130836664395,
"learning_rate": 4.492022312086605e-05,
"loss": 0.4772,
"step": 11030
},
{
"epoch": 1.2867304104477613,
"grad_norm": 0.4469991233871229,
"learning_rate": 4.491473071399787e-05,
"loss": 0.5133,
"step": 11035
},
{
"epoch": 1.287313432835821,
"grad_norm": 0.5031693931998229,
"learning_rate": 4.4909235717726086e-05,
"loss": 0.4877,
"step": 11040
},
{
"epoch": 1.2878964552238805,
"grad_norm": 0.44744449115948626,
"learning_rate": 4.490373813286776e-05,
"loss": 0.4886,
"step": 11045
},
{
"epoch": 1.2884794776119404,
"grad_norm": 0.4715880128998123,
"learning_rate": 4.4898237960240315e-05,
"loss": 0.5093,
"step": 11050
},
{
"epoch": 1.2890625,
"grad_norm": 0.4623706251903199,
"learning_rate": 4.4892735200661584e-05,
"loss": 0.473,
"step": 11055
},
{
"epoch": 1.2896455223880596,
"grad_norm": 0.4076569329168991,
"learning_rate": 4.488722985494978e-05,
"loss": 0.5065,
"step": 11060
},
{
"epoch": 1.2902285447761195,
"grad_norm": 0.4563730078630026,
"learning_rate": 4.488172192392347e-05,
"loss": 0.4682,
"step": 11065
},
{
"epoch": 1.290811567164179,
"grad_norm": 0.4627222760790777,
"learning_rate": 4.487621140840165e-05,
"loss": 0.5048,
"step": 11070
},
{
"epoch": 1.2913945895522387,
"grad_norm": 0.47548018063252756,
"learning_rate": 4.487069830920369e-05,
"loss": 0.511,
"step": 11075
},
{
"epoch": 1.2919776119402986,
"grad_norm": 0.4303449084265729,
"learning_rate": 4.486518262714931e-05,
"loss": 0.4873,
"step": 11080
},
{
"epoch": 1.2925606343283582,
"grad_norm": 0.42835590214107866,
"learning_rate": 4.4859664363058665e-05,
"loss": 0.4907,
"step": 11085
},
{
"epoch": 1.2931436567164178,
"grad_norm": 0.4547377226856405,
"learning_rate": 4.485414351775224e-05,
"loss": 0.4938,
"step": 11090
},
{
"epoch": 1.2937266791044777,
"grad_norm": 0.4927578436352473,
"learning_rate": 4.484862009205096e-05,
"loss": 0.491,
"step": 11095
},
{
"epoch": 1.2943097014925373,
"grad_norm": 0.4525594158585282,
"learning_rate": 4.484309408677609e-05,
"loss": 0.4882,
"step": 11100
},
{
"epoch": 1.294892723880597,
"grad_norm": 0.4662919621023449,
"learning_rate": 4.48375655027493e-05,
"loss": 0.4851,
"step": 11105
},
{
"epoch": 1.2954757462686568,
"grad_norm": 0.4944056942567037,
"learning_rate": 4.483203434079263e-05,
"loss": 0.5345,
"step": 11110
},
{
"epoch": 1.2960587686567164,
"grad_norm": 0.4394858066693536,
"learning_rate": 4.4826500601728515e-05,
"loss": 0.4991,
"step": 11115
},
{
"epoch": 1.296641791044776,
"grad_norm": 0.48205648299926923,
"learning_rate": 4.4820964286379764e-05,
"loss": 0.5118,
"step": 11120
},
{
"epoch": 1.2972248134328357,
"grad_norm": 0.4460294687516735,
"learning_rate": 4.481542539556959e-05,
"loss": 0.4754,
"step": 11125
},
{
"epoch": 1.2978078358208955,
"grad_norm": 0.46649810464317104,
"learning_rate": 4.480988393012155e-05,
"loss": 0.4977,
"step": 11130
},
{
"epoch": 1.2983908582089552,
"grad_norm": 0.4543096879064287,
"learning_rate": 4.4804339890859625e-05,
"loss": 0.5205,
"step": 11135
},
{
"epoch": 1.298973880597015,
"grad_norm": 0.49816168005660294,
"learning_rate": 4.479879327860816e-05,
"loss": 0.5077,
"step": 11140
},
{
"epoch": 1.2995569029850746,
"grad_norm": 0.48361299010120495,
"learning_rate": 4.479324409419186e-05,
"loss": 0.513,
"step": 11145
},
{
"epoch": 1.3001399253731343,
"grad_norm": 0.48096288151197436,
"learning_rate": 4.478769233843587e-05,
"loss": 0.4814,
"step": 11150
},
{
"epoch": 1.300722947761194,
"grad_norm": 0.44871839828132487,
"learning_rate": 4.478213801216566e-05,
"loss": 0.4846,
"step": 11155
},
{
"epoch": 1.3013059701492538,
"grad_norm": 0.475520923780983,
"learning_rate": 4.477658111620711e-05,
"loss": 0.4792,
"step": 11160
},
{
"epoch": 1.3018889925373134,
"grad_norm": 0.45266920995262744,
"learning_rate": 4.477102165138648e-05,
"loss": 0.4876,
"step": 11165
},
{
"epoch": 1.3024720149253732,
"grad_norm": 0.5301680764440551,
"learning_rate": 4.4765459618530405e-05,
"loss": 0.4966,
"step": 11170
},
{
"epoch": 1.3030550373134329,
"grad_norm": 0.4280386589494294,
"learning_rate": 4.4759895018465906e-05,
"loss": 0.478,
"step": 11175
},
{
"epoch": 1.3036380597014925,
"grad_norm": 0.4016001646885821,
"learning_rate": 4.47543278520204e-05,
"loss": 0.469,
"step": 11180
},
{
"epoch": 1.3042210820895521,
"grad_norm": 0.4199568094935958,
"learning_rate": 4.474875812002165e-05,
"loss": 0.4642,
"step": 11185
},
{
"epoch": 1.304804104477612,
"grad_norm": 0.47030310198436404,
"learning_rate": 4.474318582329783e-05,
"loss": 0.4556,
"step": 11190
},
{
"epoch": 1.3053871268656716,
"grad_norm": 0.42443805232822085,
"learning_rate": 4.473761096267749e-05,
"loss": 0.4874,
"step": 11195
},
{
"epoch": 1.3059701492537314,
"grad_norm": 0.4709462751586026,
"learning_rate": 4.4732033538989556e-05,
"loss": 0.4638,
"step": 11200
},
{
"epoch": 1.306553171641791,
"grad_norm": 0.4222680732756702,
"learning_rate": 4.4726453553063343e-05,
"loss": 0.488,
"step": 11205
},
{
"epoch": 1.3071361940298507,
"grad_norm": 0.47741257493485134,
"learning_rate": 4.4720871005728526e-05,
"loss": 0.5127,
"step": 11210
},
{
"epoch": 1.3077192164179103,
"grad_norm": 0.5247863791189061,
"learning_rate": 4.4715285897815196e-05,
"loss": 0.4891,
"step": 11215
},
{
"epoch": 1.3083022388059702,
"grad_norm": 0.4515647058817243,
"learning_rate": 4.47096982301538e-05,
"loss": 0.487,
"step": 11220
},
{
"epoch": 1.3088852611940298,
"grad_norm": 0.403232965036066,
"learning_rate": 4.470410800357515e-05,
"loss": 0.5052,
"step": 11225
},
{
"epoch": 1.3094682835820897,
"grad_norm": 0.4426716841265954,
"learning_rate": 4.469851521891049e-05,
"loss": 0.4755,
"step": 11230
},
{
"epoch": 1.3100513059701493,
"grad_norm": 0.5160837707568388,
"learning_rate": 4.469291987699139e-05,
"loss": 0.5067,
"step": 11235
},
{
"epoch": 1.310634328358209,
"grad_norm": 0.4574976120420014,
"learning_rate": 4.468732197864984e-05,
"loss": 0.5015,
"step": 11240
},
{
"epoch": 1.3112173507462686,
"grad_norm": 0.5195077800060661,
"learning_rate": 4.468172152471818e-05,
"loss": 0.4991,
"step": 11245
},
{
"epoch": 1.3118003731343284,
"grad_norm": 0.46232057951489985,
"learning_rate": 4.467611851602916e-05,
"loss": 0.5187,
"step": 11250
},
{
"epoch": 1.312383395522388,
"grad_norm": 0.47008596948770354,
"learning_rate": 4.467051295341587e-05,
"loss": 0.5048,
"step": 11255
},
{
"epoch": 1.3129664179104479,
"grad_norm": 0.45102971581296836,
"learning_rate": 4.4664904837711835e-05,
"loss": 0.4824,
"step": 11260
},
{
"epoch": 1.3135494402985075,
"grad_norm": 0.4259458579905399,
"learning_rate": 4.4659294169750896e-05,
"loss": 0.4819,
"step": 11265
},
{
"epoch": 1.3141324626865671,
"grad_norm": 0.404768642006727,
"learning_rate": 4.465368095036733e-05,
"loss": 0.4568,
"step": 11270
},
{
"epoch": 1.3147154850746268,
"grad_norm": 0.4869900553207631,
"learning_rate": 4.464806518039575e-05,
"loss": 0.5059,
"step": 11275
},
{
"epoch": 1.3152985074626866,
"grad_norm": 0.46355528098325016,
"learning_rate": 4.4642446860671185e-05,
"loss": 0.4913,
"step": 11280
},
{
"epoch": 1.3158815298507462,
"grad_norm": 0.47138496613455305,
"learning_rate": 4.463682599202902e-05,
"loss": 0.4866,
"step": 11285
},
{
"epoch": 1.316464552238806,
"grad_norm": 0.47576375715614827,
"learning_rate": 4.463120257530501e-05,
"loss": 0.4941,
"step": 11290
},
{
"epoch": 1.3170475746268657,
"grad_norm": 0.4506773331630608,
"learning_rate": 4.462557661133532e-05,
"loss": 0.4827,
"step": 11295
},
{
"epoch": 1.3176305970149254,
"grad_norm": 0.5246787882695987,
"learning_rate": 4.461994810095647e-05,
"loss": 0.4724,
"step": 11300
},
{
"epoch": 1.318213619402985,
"grad_norm": 0.46038738940449725,
"learning_rate": 4.4614317045005365e-05,
"loss": 0.4945,
"step": 11305
},
{
"epoch": 1.3187966417910448,
"grad_norm": 0.46730360950405997,
"learning_rate": 4.46086834443193e-05,
"loss": 0.4796,
"step": 11310
},
{
"epoch": 1.3193796641791045,
"grad_norm": 0.5075247511570078,
"learning_rate": 4.460304729973592e-05,
"loss": 0.494,
"step": 11315
},
{
"epoch": 1.3199626865671643,
"grad_norm": 0.41706468419557796,
"learning_rate": 4.4597408612093265e-05,
"loss": 0.5149,
"step": 11320
},
{
"epoch": 1.320545708955224,
"grad_norm": 0.40811472809077004,
"learning_rate": 4.4591767382229776e-05,
"loss": 0.4671,
"step": 11325
},
{
"epoch": 1.3211287313432836,
"grad_norm": 0.4594248970927176,
"learning_rate": 4.458612361098423e-05,
"loss": 0.4783,
"step": 11330
},
{
"epoch": 1.3217117537313432,
"grad_norm": 1.024719586676958,
"learning_rate": 4.458047729919581e-05,
"loss": 0.5189,
"step": 11335
},
{
"epoch": 1.322294776119403,
"grad_norm": 0.4277748471869806,
"learning_rate": 4.457482844770408e-05,
"loss": 0.4929,
"step": 11340
},
{
"epoch": 1.3228777985074627,
"grad_norm": 0.5040113578496384,
"learning_rate": 4.456917705734894e-05,
"loss": 0.5262,
"step": 11345
},
{
"epoch": 1.3234608208955223,
"grad_norm": 0.43865739047204216,
"learning_rate": 4.456352312897072e-05,
"loss": 0.5148,
"step": 11350
},
{
"epoch": 1.3240438432835822,
"grad_norm": 0.44935657578945365,
"learning_rate": 4.45578666634101e-05,
"loss": 0.4761,
"step": 11355
},
{
"epoch": 1.3246268656716418,
"grad_norm": 0.5147493131186344,
"learning_rate": 4.455220766150814e-05,
"loss": 0.5198,
"step": 11360
},
{
"epoch": 1.3252098880597014,
"grad_norm": 0.4818581080990096,
"learning_rate": 4.454654612410628e-05,
"loss": 0.4745,
"step": 11365
},
{
"epoch": 1.3257929104477613,
"grad_norm": 0.4579540692943894,
"learning_rate": 4.454088205204634e-05,
"loss": 0.5275,
"step": 11370
},
{
"epoch": 1.326375932835821,
"grad_norm": 0.4177917530534575,
"learning_rate": 4.453521544617051e-05,
"loss": 0.472,
"step": 11375
},
{
"epoch": 1.3269589552238805,
"grad_norm": 0.4135399462613423,
"learning_rate": 4.452954630732136e-05,
"loss": 0.4785,
"step": 11380
},
{
"epoch": 1.3275419776119404,
"grad_norm": 0.43489658697969774,
"learning_rate": 4.452387463634185e-05,
"loss": 0.5032,
"step": 11385
},
{
"epoch": 1.328125,
"grad_norm": 0.49028362359902466,
"learning_rate": 4.451820043407527e-05,
"loss": 0.4864,
"step": 11390
},
{
"epoch": 1.3287080223880596,
"grad_norm": 0.4210675259494679,
"learning_rate": 4.451252370136536e-05,
"loss": 0.5168,
"step": 11395
},
{
"epoch": 1.3292910447761195,
"grad_norm": 0.4353151647796139,
"learning_rate": 4.450684443905615e-05,
"loss": 0.4912,
"step": 11400
},
{
"epoch": 1.329874067164179,
"grad_norm": 0.4876532002745093,
"learning_rate": 4.450116264799214e-05,
"loss": 0.4966,
"step": 11405
},
{
"epoch": 1.3304570895522387,
"grad_norm": 0.4373095960768425,
"learning_rate": 4.4495478329018125e-05,
"loss": 0.5092,
"step": 11410
},
{
"epoch": 1.3310401119402986,
"grad_norm": 0.4627316321181924,
"learning_rate": 4.448979148297932e-05,
"loss": 0.4972,
"step": 11415
},
{
"epoch": 1.3316231343283582,
"grad_norm": 1.252558091692985,
"learning_rate": 4.44841021107213e-05,
"loss": 0.5056,
"step": 11420
},
{
"epoch": 1.3322061567164178,
"grad_norm": 0.48649675930152275,
"learning_rate": 4.4478410213090035e-05,
"loss": 0.4868,
"step": 11425
},
{
"epoch": 1.3327891791044777,
"grad_norm": 0.4738155854412053,
"learning_rate": 4.447271579093185e-05,
"loss": 0.4811,
"step": 11430
},
{
"epoch": 1.3333722014925373,
"grad_norm": 0.42461686825587563,
"learning_rate": 4.446701884509343e-05,
"loss": 0.4794,
"step": 11435
},
{
"epoch": 1.333955223880597,
"grad_norm": 0.4596175102509013,
"learning_rate": 4.4461319376421875e-05,
"loss": 0.4838,
"step": 11440
},
{
"epoch": 1.3345382462686568,
"grad_norm": 0.4398843309190659,
"learning_rate": 4.445561738576464e-05,
"loss": 0.4937,
"step": 11445
},
{
"epoch": 1.3351212686567164,
"grad_norm": 0.4893163040733851,
"learning_rate": 4.444991287396955e-05,
"loss": 0.4924,
"step": 11450
},
{
"epoch": 1.335704291044776,
"grad_norm": 0.437191903940666,
"learning_rate": 4.444420584188482e-05,
"loss": 0.4925,
"step": 11455
},
{
"epoch": 1.3362873134328357,
"grad_norm": 0.417694375488598,
"learning_rate": 4.443849629035903e-05,
"loss": 0.4461,
"step": 11460
},
{
"epoch": 1.3368703358208955,
"grad_norm": 0.4417471806673306,
"learning_rate": 4.443278422024113e-05,
"loss": 0.4595,
"step": 11465
},
{
"epoch": 1.3374533582089552,
"grad_norm": 0.48774322715083146,
"learning_rate": 4.4427069632380455e-05,
"loss": 0.5065,
"step": 11470
},
{
"epoch": 1.338036380597015,
"grad_norm": 0.41939788274877826,
"learning_rate": 4.4421352527626706e-05,
"loss": 0.4693,
"step": 11475
},
{
"epoch": 1.3386194029850746,
"grad_norm": 0.4440878300703214,
"learning_rate": 4.441563290682996e-05,
"loss": 0.4914,
"step": 11480
},
{
"epoch": 1.3392024253731343,
"grad_norm": 0.45833074723308814,
"learning_rate": 4.440991077084067e-05,
"loss": 0.4647,
"step": 11485
},
{
"epoch": 1.339785447761194,
"grad_norm": 0.4256435153711335,
"learning_rate": 4.4404186120509674e-05,
"loss": 0.5241,
"step": 11490
},
{
"epoch": 1.3403684701492538,
"grad_norm": 0.4502983106443609,
"learning_rate": 4.4398458956688156e-05,
"loss": 0.4878,
"step": 11495
},
{
"epoch": 1.3409514925373134,
"grad_norm": 0.5696469219475851,
"learning_rate": 4.43927292802277e-05,
"loss": 0.5033,
"step": 11500
},
{
"epoch": 1.3415345149253732,
"grad_norm": 0.4461570791917921,
"learning_rate": 4.4386997091980255e-05,
"loss": 0.484,
"step": 11505
},
{
"epoch": 1.3421175373134329,
"grad_norm": 0.4340282222712749,
"learning_rate": 4.438126239279814e-05,
"loss": 0.5006,
"step": 11510
},
{
"epoch": 1.3427005597014925,
"grad_norm": 0.4837057378517508,
"learning_rate": 4.437552518353405e-05,
"loss": 0.4833,
"step": 11515
},
{
"epoch": 1.3432835820895521,
"grad_norm": 0.4750137544995223,
"learning_rate": 4.436978546504105e-05,
"loss": 0.4787,
"step": 11520
},
{
"epoch": 1.343866604477612,
"grad_norm": 0.4518116308598689,
"learning_rate": 4.436404323817258e-05,
"loss": 0.4907,
"step": 11525
},
{
"epoch": 1.3444496268656716,
"grad_norm": 0.44161839241271017,
"learning_rate": 4.435829850378247e-05,
"loss": 0.5339,
"step": 11530
},
{
"epoch": 1.3450326492537314,
"grad_norm": 0.46451523565596275,
"learning_rate": 4.435255126272489e-05,
"loss": 0.4786,
"step": 11535
},
{
"epoch": 1.345615671641791,
"grad_norm": 0.472507616974173,
"learning_rate": 4.43468015158544e-05,
"loss": 0.4932,
"step": 11540
},
{
"epoch": 1.3461986940298507,
"grad_norm": 0.44181411074076127,
"learning_rate": 4.434104926402594e-05,
"loss": 0.4824,
"step": 11545
},
{
"epoch": 1.3467817164179103,
"grad_norm": 0.5750209645654945,
"learning_rate": 4.433529450809481e-05,
"loss": 0.5099,
"step": 11550
},
{
"epoch": 1.3473647388059702,
"grad_norm": 0.43343667019679816,
"learning_rate": 4.432953724891669e-05,
"loss": 0.4618,
"step": 11555
},
{
"epoch": 1.3479477611940298,
"grad_norm": 0.6911166321822115,
"learning_rate": 4.432377748734763e-05,
"loss": 0.5044,
"step": 11560
},
{
"epoch": 1.3485307835820897,
"grad_norm": 0.4039695330997034,
"learning_rate": 4.4318015224244044e-05,
"loss": 0.464,
"step": 11565
},
{
"epoch": 1.3491138059701493,
"grad_norm": 0.45677998741177633,
"learning_rate": 4.431225046046274e-05,
"loss": 0.4892,
"step": 11570
},
{
"epoch": 1.349696828358209,
"grad_norm": 0.42028095255888276,
"learning_rate": 4.4306483196860866e-05,
"loss": 0.4714,
"step": 11575
},
{
"epoch": 1.3502798507462686,
"grad_norm": 0.43454719039788725,
"learning_rate": 4.430071343429597e-05,
"loss": 0.4839,
"step": 11580
},
{
"epoch": 1.3508628731343284,
"grad_norm": 0.4270067192100428,
"learning_rate": 4.429494117362595e-05,
"loss": 0.4919,
"step": 11585
},
{
"epoch": 1.351445895522388,
"grad_norm": 0.47601183858432783,
"learning_rate": 4.4289166415709096e-05,
"loss": 0.4867,
"step": 11590
},
{
"epoch": 1.3520289179104479,
"grad_norm": 0.39834803271753444,
"learning_rate": 4.428338916140406e-05,
"loss": 0.4529,
"step": 11595
},
{
"epoch": 1.3526119402985075,
"grad_norm": 0.41806798471822515,
"learning_rate": 4.427760941156986e-05,
"loss": 0.469,
"step": 11600
},
{
"epoch": 1.3531949626865671,
"grad_norm": 0.4704193815184823,
"learning_rate": 4.427182716706589e-05,
"loss": 0.4891,
"step": 11605
},
{
"epoch": 1.3537779850746268,
"grad_norm": 0.4476846974897753,
"learning_rate": 4.426604242875191e-05,
"loss": 0.4983,
"step": 11610
},
{
"epoch": 1.3543610074626866,
"grad_norm": 0.4472584935699144,
"learning_rate": 4.426025519748807e-05,
"loss": 0.4858,
"step": 11615
},
{
"epoch": 1.3549440298507462,
"grad_norm": 0.4288138419577691,
"learning_rate": 4.4254465474134856e-05,
"loss": 0.4909,
"step": 11620
},
{
"epoch": 1.355527052238806,
"grad_norm": 0.43688944499628235,
"learning_rate": 4.424867325955315e-05,
"loss": 0.4484,
"step": 11625
},
{
"epoch": 1.3561100746268657,
"grad_norm": 0.4467119070895311,
"learning_rate": 4.42428785546042e-05,
"loss": 0.4572,
"step": 11630
},
{
"epoch": 1.3566930970149254,
"grad_norm": 0.48775234180390487,
"learning_rate": 4.4237081360149646e-05,
"loss": 0.5326,
"step": 11635
},
{
"epoch": 1.357276119402985,
"grad_norm": 0.42440937988251515,
"learning_rate": 4.423128167705144e-05,
"loss": 0.5245,
"step": 11640
},
{
"epoch": 1.3578591417910448,
"grad_norm": 0.4265868048302539,
"learning_rate": 4.4225479506171956e-05,
"loss": 0.4683,
"step": 11645
},
{
"epoch": 1.3584421641791045,
"grad_norm": 0.4185160086710533,
"learning_rate": 4.4219674848373924e-05,
"loss": 0.4848,
"step": 11650
},
{
"epoch": 1.3590251865671643,
"grad_norm": 0.4282718371282175,
"learning_rate": 4.421386770452042e-05,
"loss": 0.4682,
"step": 11655
},
{
"epoch": 1.359608208955224,
"grad_norm": 0.3966715398630358,
"learning_rate": 4.4208058075474945e-05,
"loss": 0.4707,
"step": 11660
},
{
"epoch": 1.3601912313432836,
"grad_norm": 0.43928634287198337,
"learning_rate": 4.4202245962101314e-05,
"loss": 0.4829,
"step": 11665
},
{
"epoch": 1.3607742537313432,
"grad_norm": 0.4079043636556758,
"learning_rate": 4.419643136526373e-05,
"loss": 0.53,
"step": 11670
},
{
"epoch": 1.361357276119403,
"grad_norm": 0.47151409999679794,
"learning_rate": 4.419061428582678e-05,
"loss": 0.488,
"step": 11675
},
{
"epoch": 1.3619402985074627,
"grad_norm": 0.4928984564012108,
"learning_rate": 4.418479472465539e-05,
"loss": 0.4987,
"step": 11680
},
{
"epoch": 1.3625233208955223,
"grad_norm": 0.4344552366085095,
"learning_rate": 4.41789726826149e-05,
"loss": 0.4862,
"step": 11685
},
{
"epoch": 1.3631063432835822,
"grad_norm": 0.4668029110763993,
"learning_rate": 4.417314816057096e-05,
"loss": 0.5062,
"step": 11690
},
{
"epoch": 1.3636893656716418,
"grad_norm": 0.49558056969821035,
"learning_rate": 4.416732115938965e-05,
"loss": 0.5118,
"step": 11695
},
{
"epoch": 1.3642723880597014,
"grad_norm": 0.5214969389536915,
"learning_rate": 4.416149167993737e-05,
"loss": 0.4765,
"step": 11700
},
{
"epoch": 1.3648554104477613,
"grad_norm": 0.5157968315431264,
"learning_rate": 4.415565972308092e-05,
"loss": 0.5147,
"step": 11705
},
{
"epoch": 1.365438432835821,
"grad_norm": 0.42535300596999137,
"learning_rate": 4.4149825289687454e-05,
"loss": 0.4939,
"step": 11710
},
{
"epoch": 1.3660214552238805,
"grad_norm": 0.44669677244769357,
"learning_rate": 4.414398838062448e-05,
"loss": 0.4892,
"step": 11715
},
{
"epoch": 1.3666044776119404,
"grad_norm": 0.4703220090528917,
"learning_rate": 4.413814899675991e-05,
"loss": 0.514,
"step": 11720
},
{
"epoch": 1.3671875,
"grad_norm": 0.427440575861135,
"learning_rate": 4.413230713896199e-05,
"loss": 0.4975,
"step": 11725
},
{
"epoch": 1.3677705223880596,
"grad_norm": 0.4328688500353565,
"learning_rate": 4.4126462808099364e-05,
"loss": 0.5014,
"step": 11730
},
{
"epoch": 1.3683535447761195,
"grad_norm": 0.4368082006934825,
"learning_rate": 4.4120616005041014e-05,
"loss": 0.4863,
"step": 11735
},
{
"epoch": 1.368936567164179,
"grad_norm": 0.42516872447155635,
"learning_rate": 4.411476673065631e-05,
"loss": 0.4882,
"step": 11740
},
{
"epoch": 1.3695195895522387,
"grad_norm": 0.40472021682783205,
"learning_rate": 4.4108914985814985e-05,
"loss": 0.4384,
"step": 11745
},
{
"epoch": 1.3701026119402986,
"grad_norm": 0.4600007872153849,
"learning_rate": 4.410306077138713e-05,
"loss": 0.5051,
"step": 11750
},
{
"epoch": 1.3706856343283582,
"grad_norm": 0.5136663989613519,
"learning_rate": 4.409720408824323e-05,
"loss": 0.4675,
"step": 11755
},
{
"epoch": 1.3712686567164178,
"grad_norm": 0.4400619155182538,
"learning_rate": 4.409134493725409e-05,
"loss": 0.4662,
"step": 11760
},
{
"epoch": 1.3718516791044777,
"grad_norm": 0.449065541332509,
"learning_rate": 4.408548331929092e-05,
"loss": 0.5552,
"step": 11765
},
{
"epoch": 1.3724347014925373,
"grad_norm": 0.47887706466721547,
"learning_rate": 4.407961923522529e-05,
"loss": 0.5058,
"step": 11770
},
{
"epoch": 1.373017723880597,
"grad_norm": 0.38601814241288335,
"learning_rate": 4.407375268592914e-05,
"loss": 0.4809,
"step": 11775
},
{
"epoch": 1.3736007462686568,
"grad_norm": 0.45229299340709755,
"learning_rate": 4.406788367227475e-05,
"loss": 0.4887,
"step": 11780
},
{
"epoch": 1.3741837686567164,
"grad_norm": 0.49749738471497534,
"learning_rate": 4.4062012195134814e-05,
"loss": 0.5211,
"step": 11785
},
{
"epoch": 1.374766791044776,
"grad_norm": 0.4574798612280582,
"learning_rate": 4.4056138255382335e-05,
"loss": 0.4905,
"step": 11790
},
{
"epoch": 1.3753498134328357,
"grad_norm": 0.4533972511175499,
"learning_rate": 4.405026185389073e-05,
"loss": 0.536,
"step": 11795
},
{
"epoch": 1.3759328358208955,
"grad_norm": 0.4716256768316318,
"learning_rate": 4.404438299153376e-05,
"loss": 0.542,
"step": 11800
},
{
"epoch": 1.3765158582089552,
"grad_norm": 0.41848445317544863,
"learning_rate": 4.4038501669185544e-05,
"loss": 0.5097,
"step": 11805
},
{
"epoch": 1.377098880597015,
"grad_norm": 0.46718965756679864,
"learning_rate": 4.4032617887720604e-05,
"loss": 0.5251,
"step": 11810
},
{
"epoch": 1.3776819029850746,
"grad_norm": 0.40800098396507833,
"learning_rate": 4.402673164801377e-05,
"loss": 0.4919,
"step": 11815
},
{
"epoch": 1.3782649253731343,
"grad_norm": 0.45963121993501865,
"learning_rate": 4.4020842950940294e-05,
"loss": 0.527,
"step": 11820
},
{
"epoch": 1.378847947761194,
"grad_norm": 0.49038282501383,
"learning_rate": 4.401495179737576e-05,
"loss": 0.4581,
"step": 11825
},
{
"epoch": 1.3794309701492538,
"grad_norm": 0.44009117483164156,
"learning_rate": 4.400905818819613e-05,
"loss": 0.4752,
"step": 11830
},
{
"epoch": 1.3800139925373134,
"grad_norm": 0.4590450905749542,
"learning_rate": 4.400316212427772e-05,
"loss": 0.5268,
"step": 11835
},
{
"epoch": 1.3805970149253732,
"grad_norm": 0.48248426252968263,
"learning_rate": 4.3997263606497225e-05,
"loss": 0.4727,
"step": 11840
},
{
"epoch": 1.3811800373134329,
"grad_norm": 0.4814309262867952,
"learning_rate": 4.3991362635731684e-05,
"loss": 0.4886,
"step": 11845
},
{
"epoch": 1.3817630597014925,
"grad_norm": 0.44899737588705313,
"learning_rate": 4.3985459212858535e-05,
"loss": 0.5019,
"step": 11850
},
{
"epoch": 1.3823460820895521,
"grad_norm": 0.4414302378152841,
"learning_rate": 4.397955333875555e-05,
"loss": 0.5178,
"step": 11855
},
{
"epoch": 1.382929104477612,
"grad_norm": 0.6612564285628922,
"learning_rate": 4.397364501430088e-05,
"loss": 0.5117,
"step": 11860
},
{
"epoch": 1.3835121268656716,
"grad_norm": 0.4337120359984851,
"learning_rate": 4.3967734240373025e-05,
"loss": 0.5125,
"step": 11865
},
{
"epoch": 1.3840951492537314,
"grad_norm": 0.4440649198676861,
"learning_rate": 4.396182101785089e-05,
"loss": 0.4958,
"step": 11870
},
{
"epoch": 1.384678171641791,
"grad_norm": 9.663684236466022,
"learning_rate": 4.3955905347613666e-05,
"loss": 0.5006,
"step": 11875
},
{
"epoch": 1.3852611940298507,
"grad_norm": 0.44755912533370196,
"learning_rate": 4.3949987230541e-05,
"loss": 0.4754,
"step": 11880
},
{
"epoch": 1.3858442164179103,
"grad_norm": 0.5148467190767787,
"learning_rate": 4.394406666751284e-05,
"loss": 0.4958,
"step": 11885
},
{
"epoch": 1.3864272388059702,
"grad_norm": 0.43418661721534973,
"learning_rate": 4.3938143659409515e-05,
"loss": 0.4815,
"step": 11890
},
{
"epoch": 1.3870102611940298,
"grad_norm": 0.4691447765306588,
"learning_rate": 4.393221820711173e-05,
"loss": 0.5257,
"step": 11895
},
{
"epoch": 1.3875932835820897,
"grad_norm": 0.4291236683000934,
"learning_rate": 4.392629031150054e-05,
"loss": 0.4615,
"step": 11900
},
{
"epoch": 1.3881763059701493,
"grad_norm": 0.4901153343134047,
"learning_rate": 4.392035997345736e-05,
"loss": 0.4896,
"step": 11905
},
{
"epoch": 1.388759328358209,
"grad_norm": 0.3982941566431026,
"learning_rate": 4.391442719386398e-05,
"loss": 0.4629,
"step": 11910
},
{
"epoch": 1.3893423507462686,
"grad_norm": 0.4117537859151202,
"learning_rate": 4.390849197360254e-05,
"loss": 0.4508,
"step": 11915
},
{
"epoch": 1.3899253731343284,
"grad_norm": 0.43084463356368424,
"learning_rate": 4.390255431355557e-05,
"loss": 0.5008,
"step": 11920
},
{
"epoch": 1.390508395522388,
"grad_norm": 0.42556976076020264,
"learning_rate": 4.389661421460592e-05,
"loss": 0.4614,
"step": 11925
},
{
"epoch": 1.3910914179104479,
"grad_norm": 0.46479192102788114,
"learning_rate": 4.389067167763683e-05,
"loss": 0.4864,
"step": 11930
},
{
"epoch": 1.3916744402985075,
"grad_norm": 0.4816085181354391,
"learning_rate": 4.388472670353191e-05,
"loss": 0.5218,
"step": 11935
},
{
"epoch": 1.3922574626865671,
"grad_norm": 0.4610453279542012,
"learning_rate": 4.387877929317512e-05,
"loss": 0.5116,
"step": 11940
},
{
"epoch": 1.3928404850746268,
"grad_norm": 0.4504152664235486,
"learning_rate": 4.387282944745077e-05,
"loss": 0.4922,
"step": 11945
},
{
"epoch": 1.3934235074626866,
"grad_norm": 0.42146949445467374,
"learning_rate": 4.3866877167243554e-05,
"loss": 0.4677,
"step": 11950
},
{
"epoch": 1.3940065298507462,
"grad_norm": 0.44820586510689486,
"learning_rate": 4.3860922453438515e-05,
"loss": 0.5098,
"step": 11955
},
{
"epoch": 1.394589552238806,
"grad_norm": 0.4775795192769077,
"learning_rate": 4.3854965306921064e-05,
"loss": 0.5096,
"step": 11960
},
{
"epoch": 1.3951725746268657,
"grad_norm": 0.4322143493768893,
"learning_rate": 4.3849005728576975e-05,
"loss": 0.481,
"step": 11965
},
{
"epoch": 1.3957555970149254,
"grad_norm": 0.4594739362514963,
"learning_rate": 4.384304371929238e-05,
"loss": 0.4891,
"step": 11970
},
{
"epoch": 1.396338619402985,
"grad_norm": 0.38526098315876905,
"learning_rate": 4.383707927995377e-05,
"loss": 0.4581,
"step": 11975
},
{
"epoch": 1.3969216417910448,
"grad_norm": 0.4428604325260078,
"learning_rate": 4.383111241144798e-05,
"loss": 0.4766,
"step": 11980
},
{
"epoch": 1.3975046641791045,
"grad_norm": 0.6088745381734858,
"learning_rate": 4.3825143114662266e-05,
"loss": 0.472,
"step": 11985
},
{
"epoch": 1.3980876865671643,
"grad_norm": 0.4331264380996842,
"learning_rate": 4.3819171390484184e-05,
"loss": 0.4965,
"step": 11990
},
{
"epoch": 1.398670708955224,
"grad_norm": 0.45889896992386375,
"learning_rate": 4.381319723980167e-05,
"loss": 0.5156,
"step": 11995
},
{
"epoch": 1.3992537313432836,
"grad_norm": 0.4993820307760769,
"learning_rate": 4.380722066350303e-05,
"loss": 0.4816,
"step": 12000
},
{
"epoch": 1.3998367537313432,
"grad_norm": 0.44968773852326066,
"learning_rate": 4.380124166247691e-05,
"loss": 0.4763,
"step": 12005
},
{
"epoch": 1.400419776119403,
"grad_norm": 0.45066812781558496,
"learning_rate": 4.3795260237612353e-05,
"loss": 0.5029,
"step": 12010
},
{
"epoch": 1.4010027985074627,
"grad_norm": 0.44516074290492613,
"learning_rate": 4.378927638979871e-05,
"loss": 0.5189,
"step": 12015
},
{
"epoch": 1.4015858208955223,
"grad_norm": 0.4560257185162567,
"learning_rate": 4.378329011992575e-05,
"loss": 0.4886,
"step": 12020
},
{
"epoch": 1.4021688432835822,
"grad_norm": 0.451992209920357,
"learning_rate": 4.377730142888356e-05,
"loss": 0.484,
"step": 12025
},
{
"epoch": 1.4027518656716418,
"grad_norm": 0.42224508768129243,
"learning_rate": 4.37713103175626e-05,
"loss": 0.4594,
"step": 12030
},
{
"epoch": 1.4033348880597014,
"grad_norm": 0.4777349919451058,
"learning_rate": 4.376531678685369e-05,
"loss": 0.4843,
"step": 12035
},
{
"epoch": 1.4039179104477613,
"grad_norm": 0.45962348039814177,
"learning_rate": 4.375932083764803e-05,
"loss": 0.4696,
"step": 12040
},
{
"epoch": 1.404500932835821,
"grad_norm": 0.4355260970118706,
"learning_rate": 4.3753322470837135e-05,
"loss": 0.4979,
"step": 12045
},
{
"epoch": 1.4050839552238805,
"grad_norm": 0.4601965696034231,
"learning_rate": 4.3747321687312916e-05,
"loss": 0.4809,
"step": 12050
},
{
"epoch": 1.4056669776119404,
"grad_norm": 0.4828330309644055,
"learning_rate": 4.3741318487967634e-05,
"loss": 0.5025,
"step": 12055
},
{
"epoch": 1.40625,
"grad_norm": 0.4890520866160836,
"learning_rate": 4.37353128736939e-05,
"loss": 0.509,
"step": 12060
},
{
"epoch": 1.4068330223880596,
"grad_norm": 0.4064163487682252,
"learning_rate": 4.3729304845384695e-05,
"loss": 0.4689,
"step": 12065
},
{
"epoch": 1.4074160447761195,
"grad_norm": 0.4149499769771672,
"learning_rate": 4.3723294403933355e-05,
"loss": 0.5198,
"step": 12070
},
{
"epoch": 1.407999067164179,
"grad_norm": 0.43687821599607873,
"learning_rate": 4.371728155023358e-05,
"loss": 0.4837,
"step": 12075
},
{
"epoch": 1.4085820895522387,
"grad_norm": 0.45448131654117757,
"learning_rate": 4.3711266285179415e-05,
"loss": 0.4978,
"step": 12080
},
{
"epoch": 1.4091651119402986,
"grad_norm": 0.4723176347005134,
"learning_rate": 4.370524860966529e-05,
"loss": 0.5008,
"step": 12085
},
{
"epoch": 1.4097481343283582,
"grad_norm": 0.46347806983282575,
"learning_rate": 4.369922852458594e-05,
"loss": 0.515,
"step": 12090
},
{
"epoch": 1.4103311567164178,
"grad_norm": 0.4436342437266037,
"learning_rate": 4.369320603083653e-05,
"loss": 0.5111,
"step": 12095
},
{
"epoch": 1.4109141791044777,
"grad_norm": 0.418830107444406,
"learning_rate": 4.3687181129312534e-05,
"loss": 0.4707,
"step": 12100
},
{
"epoch": 1.4114972014925373,
"grad_norm": 0.4707556769100974,
"learning_rate": 4.368115382090979e-05,
"loss": 0.5223,
"step": 12105
},
{
"epoch": 1.412080223880597,
"grad_norm": 0.539125567160963,
"learning_rate": 4.3675124106524514e-05,
"loss": 0.4929,
"step": 12110
},
{
"epoch": 1.4126632462686568,
"grad_norm": 0.44387996002816776,
"learning_rate": 4.366909198705325e-05,
"loss": 0.4558,
"step": 12115
},
{
"epoch": 1.4132462686567164,
"grad_norm": 0.4469636312909113,
"learning_rate": 4.366305746339293e-05,
"loss": 0.5214,
"step": 12120
},
{
"epoch": 1.413829291044776,
"grad_norm": 0.4029416312443115,
"learning_rate": 4.365702053644083e-05,
"loss": 0.4463,
"step": 12125
},
{
"epoch": 1.4144123134328357,
"grad_norm": 0.4456710663609492,
"learning_rate": 4.365098120709458e-05,
"loss": 0.4572,
"step": 12130
},
{
"epoch": 1.4149953358208955,
"grad_norm": 0.44908954981495897,
"learning_rate": 4.364493947625217e-05,
"loss": 0.4933,
"step": 12135
},
{
"epoch": 1.4155783582089552,
"grad_norm": 0.4503655079561281,
"learning_rate": 4.363889534481195e-05,
"loss": 0.5046,
"step": 12140
},
{
"epoch": 1.416161380597015,
"grad_norm": 0.424261930398403,
"learning_rate": 4.3632848813672614e-05,
"loss": 0.4653,
"step": 12145
},
{
"epoch": 1.4167444029850746,
"grad_norm": 0.44105778628436115,
"learning_rate": 4.3626799883733236e-05,
"loss": 0.4921,
"step": 12150
},
{
"epoch": 1.4173274253731343,
"grad_norm": 0.42006012534571524,
"learning_rate": 4.362074855589322e-05,
"loss": 0.4848,
"step": 12155
},
{
"epoch": 1.417910447761194,
"grad_norm": 0.40687698386764304,
"learning_rate": 4.361469483105236e-05,
"loss": 0.4841,
"step": 12160
},
{
"epoch": 1.4184934701492538,
"grad_norm": 0.4102416331704393,
"learning_rate": 4.3608638710110775e-05,
"loss": 0.4639,
"step": 12165
},
{
"epoch": 1.4190764925373134,
"grad_norm": 0.42134394967904526,
"learning_rate": 4.360258019396895e-05,
"loss": 0.483,
"step": 12170
},
{
"epoch": 1.4196595149253732,
"grad_norm": 0.5353276112439099,
"learning_rate": 4.3596519283527745e-05,
"loss": 0.5029,
"step": 12175
},
{
"epoch": 1.4202425373134329,
"grad_norm": 0.47287256377441444,
"learning_rate": 4.3590455979688335e-05,
"loss": 0.5101,
"step": 12180
},
{
"epoch": 1.4208255597014925,
"grad_norm": 0.4626159700923588,
"learning_rate": 4.358439028335229e-05,
"loss": 0.5342,
"step": 12185
},
{
"epoch": 1.4214085820895521,
"grad_norm": 0.4720855733706812,
"learning_rate": 4.357832219542151e-05,
"loss": 0.5208,
"step": 12190
},
{
"epoch": 1.421991604477612,
"grad_norm": 0.4492415730868526,
"learning_rate": 4.357225171679828e-05,
"loss": 0.5087,
"step": 12195
},
{
"epoch": 1.4225746268656716,
"grad_norm": 0.4644605710513964,
"learning_rate": 4.3566178848385194e-05,
"loss": 0.4889,
"step": 12200
},
{
"epoch": 1.4231576492537314,
"grad_norm": 0.41588190425927396,
"learning_rate": 4.3560103591085264e-05,
"loss": 0.4761,
"step": 12205
},
{
"epoch": 1.423740671641791,
"grad_norm": 0.4786554490248903,
"learning_rate": 4.35540259458018e-05,
"loss": 0.4776,
"step": 12210
},
{
"epoch": 1.4243236940298507,
"grad_norm": 0.44410318260591897,
"learning_rate": 4.3547945913438494e-05,
"loss": 0.489,
"step": 12215
},
{
"epoch": 1.4249067164179103,
"grad_norm": 0.39959270238242506,
"learning_rate": 4.3541863494899385e-05,
"loss": 0.484,
"step": 12220
},
{
"epoch": 1.4254897388059702,
"grad_norm": 0.43498444964457617,
"learning_rate": 4.353577869108887e-05,
"loss": 0.4863,
"step": 12225
},
{
"epoch": 1.4260727611940298,
"grad_norm": 0.42608426819692075,
"learning_rate": 4.352969150291172e-05,
"loss": 0.4623,
"step": 12230
},
{
"epoch": 1.4266557835820897,
"grad_norm": 0.5302877096128493,
"learning_rate": 4.3523601931273024e-05,
"loss": 0.4656,
"step": 12235
},
{
"epoch": 1.4272388059701493,
"grad_norm": 0.43611984433348255,
"learning_rate": 4.351750997707824e-05,
"loss": 0.4614,
"step": 12240
},
{
"epoch": 1.427821828358209,
"grad_norm": 0.4261003310907538,
"learning_rate": 4.351141564123319e-05,
"loss": 0.4668,
"step": 12245
},
{
"epoch": 1.4284048507462686,
"grad_norm": 0.4483288490424969,
"learning_rate": 4.3505318924644036e-05,
"loss": 0.4852,
"step": 12250
},
{
"epoch": 1.4289878731343284,
"grad_norm": 0.4745429308517631,
"learning_rate": 4.349921982821732e-05,
"loss": 0.5018,
"step": 12255
},
{
"epoch": 1.429570895522388,
"grad_norm": 0.40722023696620646,
"learning_rate": 4.34931183528599e-05,
"loss": 0.4365,
"step": 12260
},
{
"epoch": 1.4301539179104479,
"grad_norm": 0.4028239085418656,
"learning_rate": 4.3487014499479016e-05,
"loss": 0.4778,
"step": 12265
},
{
"epoch": 1.4307369402985075,
"grad_norm": 0.4247854997712988,
"learning_rate": 4.348090826898225e-05,
"loss": 0.4988,
"step": 12270
},
{
"epoch": 1.4313199626865671,
"grad_norm": 0.4048901928518633,
"learning_rate": 4.3474799662277534e-05,
"loss": 0.4755,
"step": 12275
},
{
"epoch": 1.4319029850746268,
"grad_norm": 0.4437514833245814,
"learning_rate": 4.346868868027318e-05,
"loss": 0.495,
"step": 12280
},
{
"epoch": 1.4324860074626866,
"grad_norm": 0.4756215405863023,
"learning_rate": 4.3462575323877804e-05,
"loss": 0.4735,
"step": 12285
},
{
"epoch": 1.4330690298507462,
"grad_norm": 0.42528856885323113,
"learning_rate": 4.345645959400043e-05,
"loss": 0.4771,
"step": 12290
},
{
"epoch": 1.433652052238806,
"grad_norm": 0.4391604171730095,
"learning_rate": 4.345034149155039e-05,
"loss": 0.4647,
"step": 12295
},
{
"epoch": 1.4342350746268657,
"grad_norm": 0.40144749141967323,
"learning_rate": 4.344422101743739e-05,
"loss": 0.4597,
"step": 12300
},
{
"epoch": 1.4348180970149254,
"grad_norm": 0.7966750253828946,
"learning_rate": 4.343809817257149e-05,
"loss": 0.4805,
"step": 12305
},
{
"epoch": 1.435401119402985,
"grad_norm": 0.47601640019814095,
"learning_rate": 4.3431972957863106e-05,
"loss": 0.4944,
"step": 12310
},
{
"epoch": 1.4359841417910448,
"grad_norm": 0.4536554294663507,
"learning_rate": 4.342584537422298e-05,
"loss": 0.5054,
"step": 12315
},
{
"epoch": 1.4365671641791045,
"grad_norm": 0.4887767983955777,
"learning_rate": 4.341971542256225e-05,
"loss": 0.492,
"step": 12320
},
{
"epoch": 1.4371501865671643,
"grad_norm": 0.3975790565593417,
"learning_rate": 4.341358310379235e-05,
"loss": 0.4937,
"step": 12325
},
{
"epoch": 1.437733208955224,
"grad_norm": 0.44357373763971175,
"learning_rate": 4.340744841882512e-05,
"loss": 0.4881,
"step": 12330
},
{
"epoch": 1.4383162313432836,
"grad_norm": 0.3987880605562992,
"learning_rate": 4.3401311368572723e-05,
"loss": 0.4645,
"step": 12335
},
{
"epoch": 1.4388992537313432,
"grad_norm": 0.40039316439532135,
"learning_rate": 4.339517195394768e-05,
"loss": 0.4984,
"step": 12340
},
{
"epoch": 1.439482276119403,
"grad_norm": 0.43680532542532763,
"learning_rate": 4.3389030175862854e-05,
"loss": 0.4876,
"step": 12345
},
{
"epoch": 1.4400652985074627,
"grad_norm": 0.4777486287265028,
"learning_rate": 4.3382886035231484e-05,
"loss": 0.5109,
"step": 12350
},
{
"epoch": 1.4406483208955223,
"grad_norm": 0.39871466038545517,
"learning_rate": 4.337673953296714e-05,
"loss": 0.4875,
"step": 12355
},
{
"epoch": 1.4412313432835822,
"grad_norm": 0.41247498145818423,
"learning_rate": 4.3370590669983736e-05,
"loss": 0.4728,
"step": 12360
},
{
"epoch": 1.4418143656716418,
"grad_norm": 0.4447465949127571,
"learning_rate": 4.3364439447195565e-05,
"loss": 0.4983,
"step": 12365
},
{
"epoch": 1.4423973880597014,
"grad_norm": 0.4382953649379091,
"learning_rate": 4.335828586551725e-05,
"loss": 0.484,
"step": 12370
},
{
"epoch": 1.4429804104477613,
"grad_norm": 0.43182985591318496,
"learning_rate": 4.335212992586376e-05,
"loss": 0.5007,
"step": 12375
},
{
"epoch": 1.443563432835821,
"grad_norm": 0.41190256419156196,
"learning_rate": 4.334597162915045e-05,
"loss": 0.4622,
"step": 12380
},
{
"epoch": 1.4441464552238805,
"grad_norm": 0.5890419819477933,
"learning_rate": 4.333981097629296e-05,
"loss": 0.4612,
"step": 12385
},
{
"epoch": 1.4447294776119404,
"grad_norm": 0.4122620827750522,
"learning_rate": 4.333364796820735e-05,
"loss": 0.5001,
"step": 12390
},
{
"epoch": 1.4453125,
"grad_norm": 0.4424009618506142,
"learning_rate": 4.332748260580999e-05,
"loss": 0.5064,
"step": 12395
},
{
"epoch": 1.4458955223880596,
"grad_norm": 0.5333277773391598,
"learning_rate": 4.332131489001762e-05,
"loss": 0.5058,
"step": 12400
},
{
"epoch": 1.4464785447761195,
"grad_norm": 0.4252961144391402,
"learning_rate": 4.331514482174731e-05,
"loss": 0.4903,
"step": 12405
},
{
"epoch": 1.447061567164179,
"grad_norm": 0.5145602598846947,
"learning_rate": 4.3308972401916495e-05,
"loss": 0.4512,
"step": 12410
},
{
"epoch": 1.4476445895522387,
"grad_norm": 0.44184089921348135,
"learning_rate": 4.330279763144296e-05,
"loss": 0.4768,
"step": 12415
},
{
"epoch": 1.4482276119402986,
"grad_norm": 0.42191608325707347,
"learning_rate": 4.3296620511244804e-05,
"loss": 0.5121,
"step": 12420
},
{
"epoch": 1.4488106343283582,
"grad_norm": 0.48904339278157805,
"learning_rate": 4.3290441042240544e-05,
"loss": 0.4953,
"step": 12425
},
{
"epoch": 1.4493936567164178,
"grad_norm": 0.45283412501132253,
"learning_rate": 4.3284259225348985e-05,
"loss": 0.4775,
"step": 12430
},
{
"epoch": 1.4499766791044777,
"grad_norm": 0.44979751951002644,
"learning_rate": 4.327807506148931e-05,
"loss": 0.4713,
"step": 12435
},
{
"epoch": 1.4505597014925373,
"grad_norm": 0.4444340926415052,
"learning_rate": 4.327188855158106e-05,
"loss": 0.5068,
"step": 12440
},
{
"epoch": 1.451142723880597,
"grad_norm": 0.4318952777015103,
"learning_rate": 4.3265699696544085e-05,
"loss": 0.5089,
"step": 12445
},
{
"epoch": 1.4517257462686568,
"grad_norm": 0.4653500509979724,
"learning_rate": 4.325950849729862e-05,
"loss": 0.4885,
"step": 12450
},
{
"epoch": 1.4523087686567164,
"grad_norm": 0.4878397446682021,
"learning_rate": 4.325331495476523e-05,
"loss": 0.4948,
"step": 12455
},
{
"epoch": 1.452891791044776,
"grad_norm": 0.4724080194199092,
"learning_rate": 4.3247119069864856e-05,
"loss": 0.4807,
"step": 12460
},
{
"epoch": 1.4534748134328357,
"grad_norm": 0.42491183132207805,
"learning_rate": 4.3240920843518746e-05,
"loss": 0.482,
"step": 12465
},
{
"epoch": 1.4540578358208955,
"grad_norm": 0.40164134969279697,
"learning_rate": 4.323472027664852e-05,
"loss": 0.4506,
"step": 12470
},
{
"epoch": 1.4546408582089552,
"grad_norm": 0.4186169308135839,
"learning_rate": 4.322851737017615e-05,
"loss": 0.4801,
"step": 12475
},
{
"epoch": 1.455223880597015,
"grad_norm": 0.41053533762928546,
"learning_rate": 4.322231212502394e-05,
"loss": 0.475,
"step": 12480
},
{
"epoch": 1.4558069029850746,
"grad_norm": 0.43523829404928854,
"learning_rate": 4.321610454211456e-05,
"loss": 0.4734,
"step": 12485
},
{
"epoch": 1.4563899253731343,
"grad_norm": 0.4235532721449954,
"learning_rate": 4.320989462237101e-05,
"loss": 0.4718,
"step": 12490
},
{
"epoch": 1.456972947761194,
"grad_norm": 0.5057469098686633,
"learning_rate": 4.3203682366716645e-05,
"loss": 0.4987,
"step": 12495
},
{
"epoch": 1.4575559701492538,
"grad_norm": 0.4421119102163701,
"learning_rate": 4.3197467776075185e-05,
"loss": 0.4994,
"step": 12500
},
{
"epoch": 1.4581389925373134,
"grad_norm": 0.40877775171569397,
"learning_rate": 4.3191250851370655e-05,
"loss": 0.492,
"step": 12505
},
{
"epoch": 1.4587220149253732,
"grad_norm": 0.47115934720890046,
"learning_rate": 4.318503159352748e-05,
"loss": 0.4853,
"step": 12510
},
{
"epoch": 1.4593050373134329,
"grad_norm": 0.40289925144193833,
"learning_rate": 4.317881000347037e-05,
"loss": 0.475,
"step": 12515
},
{
"epoch": 1.4598880597014925,
"grad_norm": 0.4168163898786155,
"learning_rate": 4.317258608212444e-05,
"loss": 0.4831,
"step": 12520
},
{
"epoch": 1.4604710820895521,
"grad_norm": 0.40185861798071193,
"learning_rate": 4.316635983041512e-05,
"loss": 0.4604,
"step": 12525
},
{
"epoch": 1.461054104477612,
"grad_norm": 0.39226299307634493,
"learning_rate": 4.31601312492682e-05,
"loss": 0.4625,
"step": 12530
},
{
"epoch": 1.4616371268656716,
"grad_norm": 0.45982557786300016,
"learning_rate": 4.3153900339609804e-05,
"loss": 0.5358,
"step": 12535
},
{
"epoch": 1.4622201492537314,
"grad_norm": 0.4649164262902816,
"learning_rate": 4.3147667102366415e-05,
"loss": 0.4701,
"step": 12540
},
{
"epoch": 1.462803171641791,
"grad_norm": 0.5759458752065588,
"learning_rate": 4.3141431538464846e-05,
"loss": 0.5232,
"step": 12545
},
{
"epoch": 1.4633861940298507,
"grad_norm": 0.4097720584944929,
"learning_rate": 4.313519364883227e-05,
"loss": 0.4726,
"step": 12550
},
{
"epoch": 1.4639692164179103,
"grad_norm": 0.428403317278586,
"learning_rate": 4.31289534343962e-05,
"loss": 0.4703,
"step": 12555
},
{
"epoch": 1.4645522388059702,
"grad_norm": 0.4817780524968433,
"learning_rate": 4.3122710896084504e-05,
"loss": 0.4996,
"step": 12560
},
{
"epoch": 1.4651352611940298,
"grad_norm": 0.44708410299744195,
"learning_rate": 4.311646603482538e-05,
"loss": 0.4941,
"step": 12565
},
{
"epoch": 1.4657182835820897,
"grad_norm": 0.4474854917989366,
"learning_rate": 4.3110218851547384e-05,
"loss": 0.4667,
"step": 12570
},
{
"epoch": 1.4663013059701493,
"grad_norm": 0.4232582351074315,
"learning_rate": 4.31039693471794e-05,
"loss": 0.4855,
"step": 12575
},
{
"epoch": 1.466884328358209,
"grad_norm": 0.41313500688774263,
"learning_rate": 4.309771752265069e-05,
"loss": 0.4631,
"step": 12580
},
{
"epoch": 1.4674673507462686,
"grad_norm": 0.43845478336810845,
"learning_rate": 4.309146337889082e-05,
"loss": 0.4786,
"step": 12585
},
{
"epoch": 1.4680503731343284,
"grad_norm": 0.40820439801952435,
"learning_rate": 4.308520691682974e-05,
"loss": 0.4524,
"step": 12590
},
{
"epoch": 1.468633395522388,
"grad_norm": 0.4259621029588171,
"learning_rate": 4.30789481373977e-05,
"loss": 0.4863,
"step": 12595
},
{
"epoch": 1.4692164179104479,
"grad_norm": 0.4133570436825587,
"learning_rate": 4.307268704152535e-05,
"loss": 0.4689,
"step": 12600
},
{
"epoch": 1.4697994402985075,
"grad_norm": 0.4457935140957306,
"learning_rate": 4.306642363014363e-05,
"loss": 0.5014,
"step": 12605
},
{
"epoch": 1.4703824626865671,
"grad_norm": 0.46010583854052445,
"learning_rate": 4.3060157904183873e-05,
"loss": 0.4938,
"step": 12610
},
{
"epoch": 1.4709654850746268,
"grad_norm": 0.4006398533843468,
"learning_rate": 4.305388986457772e-05,
"loss": 0.4763,
"step": 12615
},
{
"epoch": 1.4715485074626866,
"grad_norm": 0.4234382305757754,
"learning_rate": 4.3047619512257164e-05,
"loss": 0.4519,
"step": 12620
},
{
"epoch": 1.4721315298507462,
"grad_norm": 0.4093525102315727,
"learning_rate": 4.304134684815455e-05,
"loss": 0.4689,
"step": 12625
},
{
"epoch": 1.472714552238806,
"grad_norm": 0.4196275704466523,
"learning_rate": 4.3035071873202563e-05,
"loss": 0.4616,
"step": 12630
},
{
"epoch": 1.4732975746268657,
"grad_norm": 0.4259015394284289,
"learning_rate": 4.3028794588334246e-05,
"loss": 0.5045,
"step": 12635
},
{
"epoch": 1.4738805970149254,
"grad_norm": 0.3939914414962169,
"learning_rate": 4.302251499448294e-05,
"loss": 0.4961,
"step": 12640
},
{
"epoch": 1.474463619402985,
"grad_norm": 0.44185520364444403,
"learning_rate": 4.301623309258239e-05,
"loss": 0.4972,
"step": 12645
},
{
"epoch": 1.4750466417910448,
"grad_norm": 0.46608646485523214,
"learning_rate": 4.3009948883566645e-05,
"loss": 0.5024,
"step": 12650
},
{
"epoch": 1.4756296641791045,
"grad_norm": 0.39388424659760235,
"learning_rate": 4.3003662368370094e-05,
"loss": 0.5013,
"step": 12655
},
{
"epoch": 1.4762126865671643,
"grad_norm": 0.40116437547289896,
"learning_rate": 4.29973735479275e-05,
"loss": 0.4845,
"step": 12660
},
{
"epoch": 1.476795708955224,
"grad_norm": 0.43411032490509016,
"learning_rate": 4.299108242317393e-05,
"loss": 0.4605,
"step": 12665
},
{
"epoch": 1.4773787313432836,
"grad_norm": 0.39980492762307934,
"learning_rate": 4.298478899504485e-05,
"loss": 0.4931,
"step": 12670
},
{
"epoch": 1.4779617537313432,
"grad_norm": 0.4688618156846162,
"learning_rate": 4.297849326447599e-05,
"loss": 0.4978,
"step": 12675
},
{
"epoch": 1.478544776119403,
"grad_norm": 0.40564826416738364,
"learning_rate": 4.297219523240349e-05,
"loss": 0.4496,
"step": 12680
},
{
"epoch": 1.4791277985074627,
"grad_norm": 0.4345714909631159,
"learning_rate": 4.2965894899763796e-05,
"loss": 0.4752,
"step": 12685
},
{
"epoch": 1.4797108208955223,
"grad_norm": 0.4334778318451127,
"learning_rate": 4.2959592267493715e-05,
"loss": 0.49,
"step": 12690
},
{
"epoch": 1.4802938432835822,
"grad_norm": 0.4166685929591067,
"learning_rate": 4.29532873365304e-05,
"loss": 0.4756,
"step": 12695
},
{
"epoch": 1.4808768656716418,
"grad_norm": 0.5209174926132473,
"learning_rate": 4.2946980107811295e-05,
"loss": 0.5279,
"step": 12700
},
{
"epoch": 1.4814598880597014,
"grad_norm": 0.4827579151766335,
"learning_rate": 4.2940670582274265e-05,
"loss": 0.4841,
"step": 12705
},
{
"epoch": 1.4820429104477613,
"grad_norm": 0.48644754238159954,
"learning_rate": 4.2934358760857454e-05,
"loss": 0.5016,
"step": 12710
},
{
"epoch": 1.482625932835821,
"grad_norm": 0.4335735653052962,
"learning_rate": 4.2928044644499374e-05,
"loss": 0.4904,
"step": 12715
},
{
"epoch": 1.4832089552238805,
"grad_norm": 0.4125137397259088,
"learning_rate": 4.292172823413887e-05,
"loss": 0.4414,
"step": 12720
},
{
"epoch": 1.4837919776119404,
"grad_norm": 0.46539183622297464,
"learning_rate": 4.2915409530715144e-05,
"loss": 0.5214,
"step": 12725
},
{
"epoch": 1.484375,
"grad_norm": 0.4300560406379161,
"learning_rate": 4.2909088535167714e-05,
"loss": 0.445,
"step": 12730
},
{
"epoch": 1.4849580223880596,
"grad_norm": 0.4360665003036945,
"learning_rate": 4.2902765248436454e-05,
"loss": 0.5046,
"step": 12735
},
{
"epoch": 1.4855410447761195,
"grad_norm": 0.4335094594302782,
"learning_rate": 4.289643967146158e-05,
"loss": 0.4974,
"step": 12740
},
{
"epoch": 1.486124067164179,
"grad_norm": 0.48488705242221036,
"learning_rate": 4.2890111805183646e-05,
"loss": 0.5096,
"step": 12745
},
{
"epoch": 1.4867070895522387,
"grad_norm": 0.40918438376476834,
"learning_rate": 4.288378165054354e-05,
"loss": 0.4499,
"step": 12750
},
{
"epoch": 1.4872901119402986,
"grad_norm": 0.39800136322363067,
"learning_rate": 4.28774492084825e-05,
"loss": 0.4899,
"step": 12755
},
{
"epoch": 1.4878731343283582,
"grad_norm": 0.41219574591314934,
"learning_rate": 4.28711144799421e-05,
"loss": 0.4824,
"step": 12760
},
{
"epoch": 1.4884561567164178,
"grad_norm": 0.4484850868217721,
"learning_rate": 4.286477746586424e-05,
"loss": 0.4791,
"step": 12765
},
{
"epoch": 1.4890391791044777,
"grad_norm": 0.49051255211363426,
"learning_rate": 4.2858438167191185e-05,
"loss": 0.4777,
"step": 12770
},
{
"epoch": 1.4896222014925373,
"grad_norm": 0.425993495221689,
"learning_rate": 4.285209658486553e-05,
"loss": 0.4787,
"step": 12775
},
{
"epoch": 1.490205223880597,
"grad_norm": 0.4847765436901161,
"learning_rate": 4.2845752719830206e-05,
"loss": 0.4705,
"step": 12780
},
{
"epoch": 1.4907882462686568,
"grad_norm": 0.4175921333004011,
"learning_rate": 4.283940657302848e-05,
"loss": 0.5082,
"step": 12785
},
{
"epoch": 1.4913712686567164,
"grad_norm": 0.4735360584268163,
"learning_rate": 4.283305814540397e-05,
"loss": 0.4641,
"step": 12790
},
{
"epoch": 1.491954291044776,
"grad_norm": 0.4278079405722102,
"learning_rate": 4.282670743790062e-05,
"loss": 0.4927,
"step": 12795
},
{
"epoch": 1.4925373134328357,
"grad_norm": 0.43228098072331395,
"learning_rate": 4.282035445146272e-05,
"loss": 0.4979,
"step": 12800
},
{
"epoch": 1.4931203358208955,
"grad_norm": 0.4924860189029467,
"learning_rate": 4.28139991870349e-05,
"loss": 0.4795,
"step": 12805
},
{
"epoch": 1.4937033582089552,
"grad_norm": 0.3976744876424136,
"learning_rate": 4.2807641645562134e-05,
"loss": 0.5012,
"step": 12810
},
{
"epoch": 1.494286380597015,
"grad_norm": 0.4800407795785969,
"learning_rate": 4.280128182798972e-05,
"loss": 0.4818,
"step": 12815
},
{
"epoch": 1.4948694029850746,
"grad_norm": 0.4455319688757278,
"learning_rate": 4.2794919735263295e-05,
"loss": 0.4624,
"step": 12820
},
{
"epoch": 1.4954524253731343,
"grad_norm": 0.41882899861591605,
"learning_rate": 4.278855536832885e-05,
"loss": 0.5013,
"step": 12825
},
{
"epoch": 1.496035447761194,
"grad_norm": 0.4421966254620205,
"learning_rate": 4.278218872813271e-05,
"loss": 0.4942,
"step": 12830
},
{
"epoch": 1.4966184701492538,
"grad_norm": 0.45412038119801124,
"learning_rate": 4.277581981562152e-05,
"loss": 0.4706,
"step": 12835
},
{
"epoch": 1.4972014925373134,
"grad_norm": 0.4367038149200649,
"learning_rate": 4.276944863174229e-05,
"loss": 0.4855,
"step": 12840
},
{
"epoch": 1.4977845149253732,
"grad_norm": 0.4494163430328603,
"learning_rate": 4.2763075177442346e-05,
"loss": 0.4746,
"step": 12845
},
{
"epoch": 1.4983675373134329,
"grad_norm": 0.427417625421576,
"learning_rate": 4.275669945366936e-05,
"loss": 0.4898,
"step": 12850
},
{
"epoch": 1.4989505597014925,
"grad_norm": 0.4887656627468298,
"learning_rate": 4.275032146137135e-05,
"loss": 0.4808,
"step": 12855
},
{
"epoch": 1.4995335820895521,
"grad_norm": 0.42689802776408003,
"learning_rate": 4.2743941201496644e-05,
"loss": 0.4847,
"step": 12860
},
{
"epoch": 1.500116604477612,
"grad_norm": 0.43598094745387306,
"learning_rate": 4.2737558674993936e-05,
"loss": 0.477,
"step": 12865
},
{
"epoch": 1.5006996268656716,
"grad_norm": 0.4238323232177558,
"learning_rate": 4.2731173882812264e-05,
"loss": 0.4926,
"step": 12870
},
{
"epoch": 1.5012826492537314,
"grad_norm": 0.46203244450943115,
"learning_rate": 4.2724786825900955e-05,
"loss": 0.4728,
"step": 12875
},
{
"epoch": 1.501865671641791,
"grad_norm": 0.4587859587588373,
"learning_rate": 4.271839750520972e-05,
"loss": 0.4925,
"step": 12880
},
{
"epoch": 1.5024486940298507,
"grad_norm": 0.42984369024104474,
"learning_rate": 4.271200592168858e-05,
"loss": 0.4799,
"step": 12885
},
{
"epoch": 1.5030317164179103,
"grad_norm": 0.4231394348503417,
"learning_rate": 4.2705612076287907e-05,
"loss": 0.4559,
"step": 12890
},
{
"epoch": 1.5036147388059702,
"grad_norm": 0.3989365282022972,
"learning_rate": 4.269921596995842e-05,
"loss": 0.4657,
"step": 12895
},
{
"epoch": 1.5041977611940298,
"grad_norm": 0.4384376186803543,
"learning_rate": 4.2692817603651134e-05,
"loss": 0.4655,
"step": 12900
},
{
"epoch": 1.5047807835820897,
"grad_norm": 0.38533933255167363,
"learning_rate": 4.268641697831744e-05,
"loss": 0.4671,
"step": 12905
},
{
"epoch": 1.5053638059701493,
"grad_norm": 0.46050400043600404,
"learning_rate": 4.2680014094909035e-05,
"loss": 0.4811,
"step": 12910
},
{
"epoch": 1.505946828358209,
"grad_norm": 0.43695900398680027,
"learning_rate": 4.267360895437799e-05,
"loss": 0.484,
"step": 12915
},
{
"epoch": 1.5065298507462686,
"grad_norm": 0.42712376354348724,
"learning_rate": 4.2667201557676673e-05,
"loss": 0.4784,
"step": 12920
},
{
"epoch": 1.5071128731343284,
"grad_norm": 0.4445094978595546,
"learning_rate": 4.2660791905757794e-05,
"loss": 0.4795,
"step": 12925
},
{
"epoch": 1.507695895522388,
"grad_norm": 0.417555805193044,
"learning_rate": 4.2654379999574425e-05,
"loss": 0.4923,
"step": 12930
},
{
"epoch": 1.5082789179104479,
"grad_norm": 0.4406546422247791,
"learning_rate": 4.2647965840079945e-05,
"loss": 0.5034,
"step": 12935
},
{
"epoch": 1.5088619402985075,
"grad_norm": 0.44448449882347235,
"learning_rate": 4.2641549428228087e-05,
"loss": 0.4666,
"step": 12940
},
{
"epoch": 1.5094449626865671,
"grad_norm": 0.39874278334141017,
"learning_rate": 4.263513076497289e-05,
"loss": 0.4714,
"step": 12945
},
{
"epoch": 1.5100279850746268,
"grad_norm": 0.4479067494744725,
"learning_rate": 4.2628709851268775e-05,
"loss": 0.4956,
"step": 12950
},
{
"epoch": 1.5106110074626866,
"grad_norm": 0.44966984070826277,
"learning_rate": 4.262228668807044e-05,
"loss": 0.4794,
"step": 12955
},
{
"epoch": 1.5111940298507462,
"grad_norm": 0.4119622476774015,
"learning_rate": 4.261586127633297e-05,
"loss": 0.4726,
"step": 12960
},
{
"epoch": 1.511777052238806,
"grad_norm": 0.40137599613034686,
"learning_rate": 4.260943361701176e-05,
"loss": 0.4591,
"step": 12965
},
{
"epoch": 1.5123600746268657,
"grad_norm": 0.42023299547437754,
"learning_rate": 4.2603003711062536e-05,
"loss": 0.4836,
"step": 12970
},
{
"epoch": 1.5129430970149254,
"grad_norm": 0.40037349675158795,
"learning_rate": 4.259657155944136e-05,
"loss": 0.4896,
"step": 12975
},
{
"epoch": 1.513526119402985,
"grad_norm": 0.4145572583666219,
"learning_rate": 4.259013716310465e-05,
"loss": 0.4834,
"step": 12980
},
{
"epoch": 1.5141091417910446,
"grad_norm": 0.4198272987992974,
"learning_rate": 4.258370052300911e-05,
"loss": 0.4792,
"step": 12985
},
{
"epoch": 1.5146921641791045,
"grad_norm": 0.44070396389547273,
"learning_rate": 4.2577261640111834e-05,
"loss": 0.4668,
"step": 12990
},
{
"epoch": 1.5152751865671643,
"grad_norm": 0.4232891194001117,
"learning_rate": 4.25708205153702e-05,
"loss": 0.4957,
"step": 12995
},
{
"epoch": 1.515858208955224,
"grad_norm": 0.39904857993881904,
"learning_rate": 4.256437714974196e-05,
"loss": 0.5017,
"step": 13000
},
{
"epoch": 1.5164412313432836,
"grad_norm": 0.4353750148819155,
"learning_rate": 4.2557931544185166e-05,
"loss": 0.5133,
"step": 13005
},
{
"epoch": 1.5170242537313432,
"grad_norm": 0.41500382537486163,
"learning_rate": 4.255148369965822e-05,
"loss": 0.4685,
"step": 13010
},
{
"epoch": 1.5176072761194028,
"grad_norm": 0.43078388455839467,
"learning_rate": 4.254503361711987e-05,
"loss": 0.4516,
"step": 13015
},
{
"epoch": 1.5181902985074627,
"grad_norm": 0.5704149589769468,
"learning_rate": 4.253858129752916e-05,
"loss": 0.5069,
"step": 13020
},
{
"epoch": 1.5187733208955225,
"grad_norm": 0.43024437900631485,
"learning_rate": 4.2532126741845506e-05,
"loss": 0.491,
"step": 13025
},
{
"epoch": 1.5193563432835822,
"grad_norm": 0.4484607017917007,
"learning_rate": 4.252566995102864e-05,
"loss": 0.4886,
"step": 13030
},
{
"epoch": 1.5199393656716418,
"grad_norm": 0.44070631539484884,
"learning_rate": 4.25192109260386e-05,
"loss": 0.4941,
"step": 13035
},
{
"epoch": 1.5205223880597014,
"grad_norm": 0.4316579967713402,
"learning_rate": 4.251274966783579e-05,
"loss": 0.4646,
"step": 13040
},
{
"epoch": 1.521105410447761,
"grad_norm": 0.44767588050786555,
"learning_rate": 4.250628617738096e-05,
"loss": 0.4648,
"step": 13045
},
{
"epoch": 1.521688432835821,
"grad_norm": 0.40604247085739564,
"learning_rate": 4.2499820455635154e-05,
"loss": 0.4713,
"step": 13050
},
{
"epoch": 1.5222714552238807,
"grad_norm": 0.45662084505548567,
"learning_rate": 4.2493352503559756e-05,
"loss": 0.4988,
"step": 13055
},
{
"epoch": 1.5228544776119404,
"grad_norm": 0.4490900961893545,
"learning_rate": 4.24868823221165e-05,
"loss": 0.4959,
"step": 13060
},
{
"epoch": 1.5234375,
"grad_norm": 0.4289758911518763,
"learning_rate": 4.248040991226743e-05,
"loss": 0.4898,
"step": 13065
},
{
"epoch": 1.5240205223880596,
"grad_norm": 0.4416847350835621,
"learning_rate": 4.2473935274974944e-05,
"loss": 0.492,
"step": 13070
},
{
"epoch": 1.5246035447761193,
"grad_norm": 0.391925996428675,
"learning_rate": 4.246745841120174e-05,
"loss": 0.4779,
"step": 13075
},
{
"epoch": 1.525186567164179,
"grad_norm": 0.3826988879592668,
"learning_rate": 4.246097932191088e-05,
"loss": 0.4809,
"step": 13080
},
{
"epoch": 1.525769589552239,
"grad_norm": 0.39125499815155884,
"learning_rate": 4.245449800806574e-05,
"loss": 0.4665,
"step": 13085
},
{
"epoch": 1.5263526119402986,
"grad_norm": 0.42676458586294946,
"learning_rate": 4.2448014470630034e-05,
"loss": 0.479,
"step": 13090
},
{
"epoch": 1.5269356343283582,
"grad_norm": 0.4118712526151781,
"learning_rate": 4.244152871056779e-05,
"loss": 0.5024,
"step": 13095
},
{
"epoch": 1.5275186567164178,
"grad_norm": 0.39857692702972825,
"learning_rate": 4.2435040728843376e-05,
"loss": 0.4644,
"step": 13100
},
{
"epoch": 1.5281016791044775,
"grad_norm": 0.4219119952764566,
"learning_rate": 4.242855052642151e-05,
"loss": 0.4909,
"step": 13105
},
{
"epoch": 1.5286847014925373,
"grad_norm": 0.39937988060481167,
"learning_rate": 4.2422058104267215e-05,
"loss": 0.4833,
"step": 13110
},
{
"epoch": 1.5292677238805972,
"grad_norm": 0.4118277229315712,
"learning_rate": 4.241556346334584e-05,
"loss": 0.4853,
"step": 13115
},
{
"epoch": 1.5298507462686568,
"grad_norm": 0.4493140604958407,
"learning_rate": 4.2409066604623096e-05,
"loss": 0.5012,
"step": 13120
},
{
"epoch": 1.5304337686567164,
"grad_norm": 0.44284647507189534,
"learning_rate": 4.2402567529065e-05,
"loss": 0.4952,
"step": 13125
},
{
"epoch": 1.531016791044776,
"grad_norm": 0.40025342570562444,
"learning_rate": 4.239606623763789e-05,
"loss": 0.4714,
"step": 13130
},
{
"epoch": 1.5315998134328357,
"grad_norm": 0.46613399151775736,
"learning_rate": 4.2389562731308454e-05,
"loss": 0.4748,
"step": 13135
},
{
"epoch": 1.5321828358208955,
"grad_norm": 0.47135795761210825,
"learning_rate": 4.23830570110437e-05,
"loss": 0.4826,
"step": 13140
},
{
"epoch": 1.5327658582089554,
"grad_norm": 0.42414701916509373,
"learning_rate": 4.237654907781096e-05,
"loss": 0.5079,
"step": 13145
},
{
"epoch": 1.533348880597015,
"grad_norm": 0.421407479474492,
"learning_rate": 4.237003893257791e-05,
"loss": 0.4905,
"step": 13150
},
{
"epoch": 1.5339319029850746,
"grad_norm": 0.4728117585754357,
"learning_rate": 4.236352657631254e-05,
"loss": 0.5496,
"step": 13155
},
{
"epoch": 1.5345149253731343,
"grad_norm": 0.40357298044420986,
"learning_rate": 4.2357012009983185e-05,
"loss": 0.4637,
"step": 13160
},
{
"epoch": 1.535097947761194,
"grad_norm": 0.4009201035522565,
"learning_rate": 4.2350495234558494e-05,
"loss": 0.4566,
"step": 13165
},
{
"epoch": 1.5356809701492538,
"grad_norm": 0.43495102739745845,
"learning_rate": 4.234397625100745e-05,
"loss": 0.4985,
"step": 13170
},
{
"epoch": 1.5362639925373134,
"grad_norm": 0.4207053021942057,
"learning_rate": 4.233745506029934e-05,
"loss": 0.4779,
"step": 13175
},
{
"epoch": 1.5368470149253732,
"grad_norm": 0.46483422993637863,
"learning_rate": 4.2330931663403844e-05,
"loss": 0.491,
"step": 13180
},
{
"epoch": 1.5374300373134329,
"grad_norm": 0.3954227522834553,
"learning_rate": 4.232440606129089e-05,
"loss": 0.472,
"step": 13185
},
{
"epoch": 1.5380130597014925,
"grad_norm": 0.41282541519258287,
"learning_rate": 4.231787825493081e-05,
"loss": 0.4617,
"step": 13190
},
{
"epoch": 1.5385960820895521,
"grad_norm": 0.4130982502495838,
"learning_rate": 4.231134824529419e-05,
"loss": 0.4601,
"step": 13195
},
{
"epoch": 1.539179104477612,
"grad_norm": 0.43435287131953826,
"learning_rate": 4.230481603335201e-05,
"loss": 0.5106,
"step": 13200
},
{
"epoch": 1.5397621268656716,
"grad_norm": 0.38619999090987356,
"learning_rate": 4.229828162007553e-05,
"loss": 0.474,
"step": 13205
},
{
"epoch": 1.5403451492537314,
"grad_norm": 0.426167049344755,
"learning_rate": 4.229174500643634e-05,
"loss": 0.4757,
"step": 13210
},
{
"epoch": 1.540928171641791,
"grad_norm": 0.40446093120877646,
"learning_rate": 4.228520619340641e-05,
"loss": 0.4907,
"step": 13215
},
{
"epoch": 1.5415111940298507,
"grad_norm": 0.3992756880668716,
"learning_rate": 4.227866518195797e-05,
"loss": 0.467,
"step": 13220
},
{
"epoch": 1.5420942164179103,
"grad_norm": 0.43829585582456626,
"learning_rate": 4.227212197306362e-05,
"loss": 0.5113,
"step": 13225
},
{
"epoch": 1.5426772388059702,
"grad_norm": 0.4573036781634611,
"learning_rate": 4.226557656769626e-05,
"loss": 0.4979,
"step": 13230
},
{
"epoch": 1.5432602611940298,
"grad_norm": 0.47132710634129954,
"learning_rate": 4.225902896682914e-05,
"loss": 0.4863,
"step": 13235
},
{
"epoch": 1.5438432835820897,
"grad_norm": 0.4285096899328013,
"learning_rate": 4.225247917143582e-05,
"loss": 0.5048,
"step": 13240
},
{
"epoch": 1.5444263059701493,
"grad_norm": 0.4601748805852775,
"learning_rate": 4.2245927182490194e-05,
"loss": 0.4989,
"step": 13245
},
{
"epoch": 1.545009328358209,
"grad_norm": 0.46120571320041653,
"learning_rate": 4.223937300096648e-05,
"loss": 0.492,
"step": 13250
},
{
"epoch": 1.5455923507462686,
"grad_norm": 0.40037712446225193,
"learning_rate": 4.223281662783922e-05,
"loss": 0.4699,
"step": 13255
},
{
"epoch": 1.5461753731343284,
"grad_norm": 0.43357360655332144,
"learning_rate": 4.22262580640833e-05,
"loss": 0.5139,
"step": 13260
},
{
"epoch": 1.546758395522388,
"grad_norm": 0.4578799403414573,
"learning_rate": 4.221969731067388e-05,
"loss": 0.4858,
"step": 13265
},
{
"epoch": 1.5473414179104479,
"grad_norm": 0.4294249725349592,
"learning_rate": 4.221313436858651e-05,
"loss": 0.4817,
"step": 13270
},
{
"epoch": 1.5479244402985075,
"grad_norm": 0.4127021933030226,
"learning_rate": 4.2206569238797025e-05,
"loss": 0.4616,
"step": 13275
},
{
"epoch": 1.5485074626865671,
"grad_norm": 0.40377848223242957,
"learning_rate": 4.220000192228161e-05,
"loss": 0.5366,
"step": 13280
},
{
"epoch": 1.5490904850746268,
"grad_norm": 0.4267006224788294,
"learning_rate": 4.2193432420016746e-05,
"loss": 0.5087,
"step": 13285
},
{
"epoch": 1.5496735074626866,
"grad_norm": 0.4346604585728533,
"learning_rate": 4.218686073297926e-05,
"loss": 0.511,
"step": 13290
},
{
"epoch": 1.5502565298507462,
"grad_norm": 0.41716535906131436,
"learning_rate": 4.218028686214631e-05,
"loss": 0.4603,
"step": 13295
},
{
"epoch": 1.550839552238806,
"grad_norm": 0.4126400492390039,
"learning_rate": 4.217371080849535e-05,
"loss": 0.4532,
"step": 13300
},
{
"epoch": 1.5514225746268657,
"grad_norm": 0.408431332457747,
"learning_rate": 4.216713257300418e-05,
"loss": 0.4623,
"step": 13305
},
{
"epoch": 1.5520055970149254,
"grad_norm": 0.4087969742335221,
"learning_rate": 4.216055215665093e-05,
"loss": 0.4751,
"step": 13310
},
{
"epoch": 1.552588619402985,
"grad_norm": 0.44399390016750945,
"learning_rate": 4.215396956041404e-05,
"loss": 0.4946,
"step": 13315
},
{
"epoch": 1.5531716417910446,
"grad_norm": 0.40275491386457934,
"learning_rate": 4.2147384785272284e-05,
"loss": 0.4548,
"step": 13320
},
{
"epoch": 1.5537546641791045,
"grad_norm": 0.4225983009878632,
"learning_rate": 4.214079783220474e-05,
"loss": 0.4889,
"step": 13325
},
{
"epoch": 1.5543376865671643,
"grad_norm": 0.4424375061579874,
"learning_rate": 4.213420870219084e-05,
"loss": 0.522,
"step": 13330
},
{
"epoch": 1.554920708955224,
"grad_norm": 0.41521237834904035,
"learning_rate": 4.212761739621032e-05,
"loss": 0.4647,
"step": 13335
},
{
"epoch": 1.5555037313432836,
"grad_norm": 0.41541356196497864,
"learning_rate": 4.212102391524324e-05,
"loss": 0.4896,
"step": 13340
},
{
"epoch": 1.5560867537313432,
"grad_norm": 0.43157699479722045,
"learning_rate": 4.211442826027e-05,
"loss": 0.4708,
"step": 13345
},
{
"epoch": 1.5566697761194028,
"grad_norm": 0.41054688958194885,
"learning_rate": 4.210783043227129e-05,
"loss": 0.4784,
"step": 13350
},
{
"epoch": 1.5572527985074627,
"grad_norm": 0.5577820332160767,
"learning_rate": 4.210123043222816e-05,
"loss": 0.5368,
"step": 13355
},
{
"epoch": 1.5578358208955225,
"grad_norm": 0.42009766884515093,
"learning_rate": 4.209462826112195e-05,
"loss": 0.4593,
"step": 13360
},
{
"epoch": 1.5584188432835822,
"grad_norm": 0.40169251834467307,
"learning_rate": 4.2088023919934366e-05,
"loss": 0.4704,
"step": 13365
},
{
"epoch": 1.5590018656716418,
"grad_norm": 0.44031000256184927,
"learning_rate": 4.2081417409647386e-05,
"loss": 0.5025,
"step": 13370
},
{
"epoch": 1.5595848880597014,
"grad_norm": 0.3983519697254828,
"learning_rate": 4.207480873124335e-05,
"loss": 0.5129,
"step": 13375
},
{
"epoch": 1.560167910447761,
"grad_norm": 0.4168826424645943,
"learning_rate": 4.2068197885704904e-05,
"loss": 0.4901,
"step": 13380
},
{
"epoch": 1.560750932835821,
"grad_norm": 0.4247715644740658,
"learning_rate": 4.2061584874015006e-05,
"loss": 0.4944,
"step": 13385
},
{
"epoch": 1.5613339552238807,
"grad_norm": 0.4488066024401581,
"learning_rate": 4.205496969715696e-05,
"loss": 0.5082,
"step": 13390
},
{
"epoch": 1.5619169776119404,
"grad_norm": 0.6085524137681178,
"learning_rate": 4.2048352356114366e-05,
"loss": 0.4868,
"step": 13395
},
{
"epoch": 1.5625,
"grad_norm": 0.44574728325793944,
"learning_rate": 4.204173285187117e-05,
"loss": 0.4842,
"step": 13400
},
{
"epoch": 1.5630830223880596,
"grad_norm": 0.4283934712785062,
"learning_rate": 4.203511118541163e-05,
"loss": 0.5008,
"step": 13405
},
{
"epoch": 1.5636660447761193,
"grad_norm": 0.46648021154345765,
"learning_rate": 4.202848735772031e-05,
"loss": 0.4999,
"step": 13410
},
{
"epoch": 1.564249067164179,
"grad_norm": 0.38962086511204524,
"learning_rate": 4.202186136978213e-05,
"loss": 0.4569,
"step": 13415
},
{
"epoch": 1.564832089552239,
"grad_norm": 0.4053152724284662,
"learning_rate": 4.201523322258231e-05,
"loss": 0.4452,
"step": 13420
},
{
"epoch": 1.5654151119402986,
"grad_norm": 0.45365460844311817,
"learning_rate": 4.2008602917106365e-05,
"loss": 0.5115,
"step": 13425
},
{
"epoch": 1.5659981343283582,
"grad_norm": 0.41269832929867595,
"learning_rate": 4.2001970454340185e-05,
"loss": 0.4741,
"step": 13430
},
{
"epoch": 1.5665811567164178,
"grad_norm": 0.39565240262036605,
"learning_rate": 4.199533583526994e-05,
"loss": 0.4491,
"step": 13435
},
{
"epoch": 1.5671641791044775,
"grad_norm": 0.4410938378323605,
"learning_rate": 4.1988699060882144e-05,
"loss": 0.4826,
"step": 13440
},
{
"epoch": 1.5677472014925373,
"grad_norm": 0.39842097053892017,
"learning_rate": 4.198206013216361e-05,
"loss": 0.507,
"step": 13445
},
{
"epoch": 1.5683302238805972,
"grad_norm": 0.45863932701511173,
"learning_rate": 4.197541905010149e-05,
"loss": 0.4471,
"step": 13450
},
{
"epoch": 1.5689132462686568,
"grad_norm": 0.4311043355544529,
"learning_rate": 4.196877581568326e-05,
"loss": 0.5115,
"step": 13455
},
{
"epoch": 1.5694962686567164,
"grad_norm": 0.43044704887228286,
"learning_rate": 4.196213042989668e-05,
"loss": 0.4931,
"step": 13460
},
{
"epoch": 1.570079291044776,
"grad_norm": 0.40671533407476856,
"learning_rate": 4.195548289372988e-05,
"loss": 0.5098,
"step": 13465
},
{
"epoch": 1.5706623134328357,
"grad_norm": 0.43825156815662814,
"learning_rate": 4.194883320817127e-05,
"loss": 0.4953,
"step": 13470
},
{
"epoch": 1.5712453358208955,
"grad_norm": 0.4444194837704474,
"learning_rate": 4.1942181374209596e-05,
"loss": 0.4999,
"step": 13475
},
{
"epoch": 1.5718283582089554,
"grad_norm": 0.4009905992752678,
"learning_rate": 4.193552739283393e-05,
"loss": 0.4662,
"step": 13480
},
{
"epoch": 1.572411380597015,
"grad_norm": 0.4316134200753667,
"learning_rate": 4.192887126503364e-05,
"loss": 0.5024,
"step": 13485
},
{
"epoch": 1.5729944029850746,
"grad_norm": 0.40927502514366515,
"learning_rate": 4.192221299179845e-05,
"loss": 0.4842,
"step": 13490
},
{
"epoch": 1.5735774253731343,
"grad_norm": 0.40540407523137834,
"learning_rate": 4.191555257411837e-05,
"loss": 0.4879,
"step": 13495
},
{
"epoch": 1.574160447761194,
"grad_norm": 0.403099969585178,
"learning_rate": 4.190889001298373e-05,
"loss": 0.4629,
"step": 13500
},
{
"epoch": 1.5747434701492538,
"grad_norm": 0.434966788230841,
"learning_rate": 4.190222530938521e-05,
"loss": 0.4871,
"step": 13505
},
{
"epoch": 1.5753264925373134,
"grad_norm": 0.4122870548217484,
"learning_rate": 4.189555846431377e-05,
"loss": 0.4661,
"step": 13510
},
{
"epoch": 1.5759095149253732,
"grad_norm": 0.5319251320420735,
"learning_rate": 4.188888947876071e-05,
"loss": 0.5356,
"step": 13515
},
{
"epoch": 1.5764925373134329,
"grad_norm": 0.42569822190937806,
"learning_rate": 4.188221835371766e-05,
"loss": 0.4437,
"step": 13520
},
{
"epoch": 1.5770755597014925,
"grad_norm": 0.45460568678761587,
"learning_rate": 4.187554509017653e-05,
"loss": 0.4794,
"step": 13525
},
{
"epoch": 1.5776585820895521,
"grad_norm": 0.4077073604912347,
"learning_rate": 4.1868869689129584e-05,
"loss": 0.4625,
"step": 13530
},
{
"epoch": 1.578241604477612,
"grad_norm": 0.4283745446098321,
"learning_rate": 4.186219215156938e-05,
"loss": 0.486,
"step": 13535
},
{
"epoch": 1.5788246268656716,
"grad_norm": 0.4748030475902487,
"learning_rate": 4.1855512478488816e-05,
"loss": 0.4668,
"step": 13540
},
{
"epoch": 1.5794076492537314,
"grad_norm": 0.4240629031299508,
"learning_rate": 4.184883067088108e-05,
"loss": 0.4829,
"step": 13545
},
{
"epoch": 1.579990671641791,
"grad_norm": 0.40511524150360195,
"learning_rate": 4.184214672973971e-05,
"loss": 0.5097,
"step": 13550
},
{
"epoch": 1.5805736940298507,
"grad_norm": 0.41235092373023413,
"learning_rate": 4.183546065605855e-05,
"loss": 0.5137,
"step": 13555
},
{
"epoch": 1.5811567164179103,
"grad_norm": 0.41714248354462624,
"learning_rate": 4.182877245083172e-05,
"loss": 0.4558,
"step": 13560
},
{
"epoch": 1.5817397388059702,
"grad_norm": 0.403392553276603,
"learning_rate": 4.1822082115053717e-05,
"loss": 0.4481,
"step": 13565
},
{
"epoch": 1.5823227611940298,
"grad_norm": 0.4508819632643197,
"learning_rate": 4.181538964971933e-05,
"loss": 0.4628,
"step": 13570
},
{
"epoch": 1.5829057835820897,
"grad_norm": 0.4525378549777149,
"learning_rate": 4.180869505582366e-05,
"loss": 0.4909,
"step": 13575
},
{
"epoch": 1.5834888059701493,
"grad_norm": 0.44597128262577185,
"learning_rate": 4.180199833436213e-05,
"loss": 0.4769,
"step": 13580
},
{
"epoch": 1.584071828358209,
"grad_norm": 0.4459012583943023,
"learning_rate": 4.179529948633047e-05,
"loss": 0.5133,
"step": 13585
},
{
"epoch": 1.5846548507462686,
"grad_norm": 0.4679210869513856,
"learning_rate": 4.178859851272475e-05,
"loss": 0.5075,
"step": 13590
},
{
"epoch": 1.5852378731343284,
"grad_norm": 0.4317612218037428,
"learning_rate": 4.1781895414541326e-05,
"loss": 0.5058,
"step": 13595
},
{
"epoch": 1.585820895522388,
"grad_norm": 0.455624049071983,
"learning_rate": 4.1775190192776905e-05,
"loss": 0.4975,
"step": 13600
},
{
"epoch": 1.5864039179104479,
"grad_norm": 0.431954202780353,
"learning_rate": 4.176848284842847e-05,
"loss": 0.5046,
"step": 13605
},
{
"epoch": 1.5869869402985075,
"grad_norm": 0.39093059851720136,
"learning_rate": 4.176177338249334e-05,
"loss": 0.4473,
"step": 13610
},
{
"epoch": 1.5875699626865671,
"grad_norm": 0.4572824921584941,
"learning_rate": 4.1755061795969155e-05,
"loss": 0.5043,
"step": 13615
},
{
"epoch": 1.5881529850746268,
"grad_norm": 0.39997096727612613,
"learning_rate": 4.1748348089853864e-05,
"loss": 0.473,
"step": 13620
},
{
"epoch": 1.5887360074626866,
"grad_norm": 0.4122896828831148,
"learning_rate": 4.1741632265145715e-05,
"loss": 0.4832,
"step": 13625
},
{
"epoch": 1.5893190298507462,
"grad_norm": 0.44332962502958334,
"learning_rate": 4.173491432284332e-05,
"loss": 0.5335,
"step": 13630
},
{
"epoch": 1.589902052238806,
"grad_norm": 0.4460466797414813,
"learning_rate": 4.172819426394554e-05,
"loss": 0.4858,
"step": 13635
},
{
"epoch": 1.5904850746268657,
"grad_norm": 0.43173460502304956,
"learning_rate": 4.172147208945159e-05,
"loss": 0.5037,
"step": 13640
},
{
"epoch": 1.5910680970149254,
"grad_norm": 0.4068645141608874,
"learning_rate": 4.171474780036101e-05,
"loss": 0.4539,
"step": 13645
},
{
"epoch": 1.591651119402985,
"grad_norm": 0.40161314212077937,
"learning_rate": 4.170802139767362e-05,
"loss": 0.5051,
"step": 13650
},
{
"epoch": 1.5922341417910446,
"grad_norm": 0.4212810978511474,
"learning_rate": 4.170129288238958e-05,
"loss": 0.4782,
"step": 13655
},
{
"epoch": 1.5928171641791045,
"grad_norm": 0.4672279762345317,
"learning_rate": 4.1694562255509354e-05,
"loss": 0.4761,
"step": 13660
},
{
"epoch": 1.5934001865671643,
"grad_norm": 0.4096412974311588,
"learning_rate": 4.1687829518033726e-05,
"loss": 0.4878,
"step": 13665
},
{
"epoch": 1.593983208955224,
"grad_norm": 0.43522674885978196,
"learning_rate": 4.168109467096378e-05,
"loss": 0.4661,
"step": 13670
},
{
"epoch": 1.5945662313432836,
"grad_norm": 0.4211645815058826,
"learning_rate": 4.1674357715300924e-05,
"loss": 0.4589,
"step": 13675
},
{
"epoch": 1.5951492537313432,
"grad_norm": 0.466936568795633,
"learning_rate": 4.1667618652046894e-05,
"loss": 0.4982,
"step": 13680
},
{
"epoch": 1.5957322761194028,
"grad_norm": 0.4488551762263432,
"learning_rate": 4.1660877482203704e-05,
"loss": 0.4972,
"step": 13685
},
{
"epoch": 1.5963152985074627,
"grad_norm": 0.3994647058487909,
"learning_rate": 4.165413420677372e-05,
"loss": 0.4738,
"step": 13690
},
{
"epoch": 1.5968983208955225,
"grad_norm": 0.4688525595679455,
"learning_rate": 4.164738882675958e-05,
"loss": 0.5398,
"step": 13695
},
{
"epoch": 1.5974813432835822,
"grad_norm": 0.45941567626509505,
"learning_rate": 4.164064134316428e-05,
"loss": 0.5057,
"step": 13700
},
{
"epoch": 1.5980643656716418,
"grad_norm": 0.47364171736560806,
"learning_rate": 4.163389175699109e-05,
"loss": 0.4975,
"step": 13705
},
{
"epoch": 1.5986473880597014,
"grad_norm": 0.439734025007231,
"learning_rate": 4.162714006924362e-05,
"loss": 0.5169,
"step": 13710
},
{
"epoch": 1.599230410447761,
"grad_norm": 0.4587953563310536,
"learning_rate": 4.1620386280925776e-05,
"loss": 0.4744,
"step": 13715
},
{
"epoch": 1.599813432835821,
"grad_norm": 0.5455847649812249,
"learning_rate": 4.161363039304177e-05,
"loss": 0.5009,
"step": 13720
},
{
"epoch": 1.6003964552238807,
"grad_norm": 0.4583565860258099,
"learning_rate": 4.160687240659616e-05,
"loss": 0.4638,
"step": 13725
},
{
"epoch": 1.6009794776119404,
"grad_norm": 0.41780067818965255,
"learning_rate": 4.160011232259378e-05,
"loss": 0.4963,
"step": 13730
},
{
"epoch": 1.6015625,
"grad_norm": 0.36360567577215513,
"learning_rate": 4.1593350142039806e-05,
"loss": 0.4521,
"step": 13735
},
{
"epoch": 1.6021455223880596,
"grad_norm": 0.5549813184778564,
"learning_rate": 4.158658586593969e-05,
"loss": 0.4994,
"step": 13740
},
{
"epoch": 1.6027285447761193,
"grad_norm": 0.3982818850740962,
"learning_rate": 4.157981949529922e-05,
"loss": 0.4716,
"step": 13745
},
{
"epoch": 1.603311567164179,
"grad_norm": 0.428213403831558,
"learning_rate": 4.1573051031124486e-05,
"loss": 0.4404,
"step": 13750
},
{
"epoch": 1.603894589552239,
"grad_norm": 0.43629496576454474,
"learning_rate": 4.156628047442191e-05,
"loss": 0.4643,
"step": 13755
},
{
"epoch": 1.6044776119402986,
"grad_norm": 0.4193637142109093,
"learning_rate": 4.155950782619819e-05,
"loss": 0.5045,
"step": 13760
},
{
"epoch": 1.6050606343283582,
"grad_norm": 0.41187735553233135,
"learning_rate": 4.155273308746037e-05,
"loss": 0.4731,
"step": 13765
},
{
"epoch": 1.6056436567164178,
"grad_norm": 0.4153061528694636,
"learning_rate": 4.1545956259215776e-05,
"loss": 0.4604,
"step": 13770
},
{
"epoch": 1.6062266791044775,
"grad_norm": 0.4265804750716642,
"learning_rate": 4.153917734247208e-05,
"loss": 0.4966,
"step": 13775
},
{
"epoch": 1.6068097014925373,
"grad_norm": 0.4247938848939225,
"learning_rate": 4.153239633823721e-05,
"loss": 0.4826,
"step": 13780
},
{
"epoch": 1.6073927238805972,
"grad_norm": 0.4080890456726183,
"learning_rate": 4.152561324751945e-05,
"loss": 0.4953,
"step": 13785
},
{
"epoch": 1.6079757462686568,
"grad_norm": 0.46540044999851854,
"learning_rate": 4.151882807132739e-05,
"loss": 0.5214,
"step": 13790
},
{
"epoch": 1.6085587686567164,
"grad_norm": 0.4374526811163078,
"learning_rate": 4.1512040810669905e-05,
"loss": 0.4926,
"step": 13795
},
{
"epoch": 1.609141791044776,
"grad_norm": 0.4312462430872498,
"learning_rate": 4.1505251466556206e-05,
"loss": 0.4902,
"step": 13800
},
{
"epoch": 1.6097248134328357,
"grad_norm": 0.45494365001895526,
"learning_rate": 4.1498460039995805e-05,
"loss": 0.4832,
"step": 13805
},
{
"epoch": 1.6103078358208955,
"grad_norm": 0.44416438041500533,
"learning_rate": 4.149166653199852e-05,
"loss": 0.4958,
"step": 13810
},
{
"epoch": 1.6108908582089554,
"grad_norm": 0.4476916723216493,
"learning_rate": 4.148487094357447e-05,
"loss": 0.4874,
"step": 13815
},
{
"epoch": 1.611473880597015,
"grad_norm": 0.44998270039657035,
"learning_rate": 4.1478073275734105e-05,
"loss": 0.472,
"step": 13820
},
{
"epoch": 1.6120569029850746,
"grad_norm": 0.414424612331312,
"learning_rate": 4.147127352948817e-05,
"loss": 0.4639,
"step": 13825
},
{
"epoch": 1.6126399253731343,
"grad_norm": 0.4235139396928849,
"learning_rate": 4.146447170584772e-05,
"loss": 0.4438,
"step": 13830
},
{
"epoch": 1.613222947761194,
"grad_norm": 0.4558452516562571,
"learning_rate": 4.145766780582413e-05,
"loss": 0.4894,
"step": 13835
},
{
"epoch": 1.6138059701492538,
"grad_norm": 0.45802437895486725,
"learning_rate": 4.145086183042907e-05,
"loss": 0.4825,
"step": 13840
},
{
"epoch": 1.6143889925373134,
"grad_norm": 0.4051697239513389,
"learning_rate": 4.1444053780674506e-05,
"loss": 0.4763,
"step": 13845
},
{
"epoch": 1.6149720149253732,
"grad_norm": 0.3891587238042614,
"learning_rate": 4.143724365757275e-05,
"loss": 0.4492,
"step": 13850
},
{
"epoch": 1.6155550373134329,
"grad_norm": 0.42980012673803525,
"learning_rate": 4.143043146213641e-05,
"loss": 0.4734,
"step": 13855
},
{
"epoch": 1.6161380597014925,
"grad_norm": 0.3918017871738858,
"learning_rate": 4.142361719537838e-05,
"loss": 0.478,
"step": 13860
},
{
"epoch": 1.6167210820895521,
"grad_norm": 0.38440071087680916,
"learning_rate": 4.1416800858311875e-05,
"loss": 0.4541,
"step": 13865
},
{
"epoch": 1.617304104477612,
"grad_norm": 0.4022474722165739,
"learning_rate": 4.140998245195042e-05,
"loss": 0.4635,
"step": 13870
},
{
"epoch": 1.6178871268656716,
"grad_norm": 0.394886036018443,
"learning_rate": 4.1403161977307845e-05,
"loss": 0.4969,
"step": 13875
},
{
"epoch": 1.6184701492537314,
"grad_norm": 0.4155495143145242,
"learning_rate": 4.13963394353983e-05,
"loss": 0.5149,
"step": 13880
},
{
"epoch": 1.619053171641791,
"grad_norm": 0.425780163466277,
"learning_rate": 4.1389514827236214e-05,
"loss": 0.5023,
"step": 13885
},
{
"epoch": 1.6196361940298507,
"grad_norm": 0.3970847797166461,
"learning_rate": 4.138268815383636e-05,
"loss": 0.4614,
"step": 13890
},
{
"epoch": 1.6202192164179103,
"grad_norm": 0.4841170861966004,
"learning_rate": 4.137585941621379e-05,
"loss": 0.5141,
"step": 13895
},
{
"epoch": 1.6208022388059702,
"grad_norm": 0.4321492633657125,
"learning_rate": 4.136902861538387e-05,
"loss": 0.4941,
"step": 13900
},
{
"epoch": 1.6213852611940298,
"grad_norm": 0.4193613731390034,
"learning_rate": 4.136219575236228e-05,
"loss": 0.4832,
"step": 13905
},
{
"epoch": 1.6219682835820897,
"grad_norm": 0.478542902077223,
"learning_rate": 4.135536082816499e-05,
"loss": 0.4919,
"step": 13910
},
{
"epoch": 1.6225513059701493,
"grad_norm": 0.3910423437514934,
"learning_rate": 4.13485238438083e-05,
"loss": 0.4782,
"step": 13915
},
{
"epoch": 1.623134328358209,
"grad_norm": 0.39792175646311384,
"learning_rate": 4.13416848003088e-05,
"loss": 0.4701,
"step": 13920
},
{
"epoch": 1.6237173507462686,
"grad_norm": 0.42639038291147685,
"learning_rate": 4.133484369868339e-05,
"loss": 0.4865,
"step": 13925
},
{
"epoch": 1.6243003731343284,
"grad_norm": 0.4834550137525021,
"learning_rate": 4.132800053994927e-05,
"loss": 0.5021,
"step": 13930
},
{
"epoch": 1.624883395522388,
"grad_norm": 0.40030111596840384,
"learning_rate": 4.132115532512397e-05,
"loss": 0.4933,
"step": 13935
},
{
"epoch": 1.6254664179104479,
"grad_norm": 0.40158229111374893,
"learning_rate": 4.1314308055225295e-05,
"loss": 0.5126,
"step": 13940
},
{
"epoch": 1.6260494402985075,
"grad_norm": 0.38529150774322646,
"learning_rate": 4.130745873127136e-05,
"loss": 0.4645,
"step": 13945
},
{
"epoch": 1.6266324626865671,
"grad_norm": 0.39397074973711016,
"learning_rate": 4.1300607354280605e-05,
"loss": 0.4933,
"step": 13950
},
{
"epoch": 1.6272154850746268,
"grad_norm": 0.4310925387629514,
"learning_rate": 4.129375392527177e-05,
"loss": 0.4634,
"step": 13955
},
{
"epoch": 1.6277985074626866,
"grad_norm": 0.4049957818820292,
"learning_rate": 4.128689844526388e-05,
"loss": 0.4614,
"step": 13960
},
{
"epoch": 1.6283815298507462,
"grad_norm": 0.4206548881995724,
"learning_rate": 4.128004091527629e-05,
"loss": 0.5058,
"step": 13965
},
{
"epoch": 1.628964552238806,
"grad_norm": 0.42818619835778277,
"learning_rate": 4.1273181336328646e-05,
"loss": 0.4768,
"step": 13970
},
{
"epoch": 1.6295475746268657,
"grad_norm": 0.49216574144431363,
"learning_rate": 4.1266319709440895e-05,
"loss": 0.4693,
"step": 13975
},
{
"epoch": 1.6301305970149254,
"grad_norm": 0.3772910316702474,
"learning_rate": 4.125945603563331e-05,
"loss": 0.4832,
"step": 13980
},
{
"epoch": 1.630713619402985,
"grad_norm": 0.3775254497117327,
"learning_rate": 4.1252590315926435e-05,
"loss": 0.4835,
"step": 13985
},
{
"epoch": 1.6312966417910446,
"grad_norm": 0.4604685839675388,
"learning_rate": 4.124572255134115e-05,
"loss": 0.4558,
"step": 13990
},
{
"epoch": 1.6318796641791045,
"grad_norm": 0.3961489769456557,
"learning_rate": 4.123885274289862e-05,
"loss": 0.4543,
"step": 13995
},
{
"epoch": 1.6324626865671643,
"grad_norm": 0.39566019294349775,
"learning_rate": 4.123198089162033e-05,
"loss": 0.4669,
"step": 14000
},
{
"epoch": 1.633045708955224,
"grad_norm": 0.4577549701799179,
"learning_rate": 4.122510699852803e-05,
"loss": 0.4815,
"step": 14005
},
{
"epoch": 1.6336287313432836,
"grad_norm": 0.4296076529412509,
"learning_rate": 4.121823106464384e-05,
"loss": 0.5045,
"step": 14010
},
{
"epoch": 1.6342117537313432,
"grad_norm": 0.412674549575472,
"learning_rate": 4.121135309099013e-05,
"loss": 0.4939,
"step": 14015
},
{
"epoch": 1.6347947761194028,
"grad_norm": 0.45241629533800665,
"learning_rate": 4.1204473078589575e-05,
"loss": 0.4878,
"step": 14020
},
{
"epoch": 1.6353777985074627,
"grad_norm": 0.5013913112036309,
"learning_rate": 4.119759102846518e-05,
"loss": 0.4996,
"step": 14025
},
{
"epoch": 1.6359608208955225,
"grad_norm": 0.3951691037062018,
"learning_rate": 4.119070694164024e-05,
"loss": 0.4782,
"step": 14030
},
{
"epoch": 1.6365438432835822,
"grad_norm": 0.4589785308608991,
"learning_rate": 4.1183820819138355e-05,
"loss": 0.4651,
"step": 14035
},
{
"epoch": 1.6371268656716418,
"grad_norm": 0.45276924257140067,
"learning_rate": 4.117693266198342e-05,
"loss": 0.4734,
"step": 14040
},
{
"epoch": 1.6377098880597014,
"grad_norm": 0.44295878617434087,
"learning_rate": 4.117004247119964e-05,
"loss": 0.4831,
"step": 14045
},
{
"epoch": 1.638292910447761,
"grad_norm": 0.410851387489969,
"learning_rate": 4.116315024781152e-05,
"loss": 0.4834,
"step": 14050
},
{
"epoch": 1.638875932835821,
"grad_norm": 0.4164900346469144,
"learning_rate": 4.115625599284386e-05,
"loss": 0.489,
"step": 14055
},
{
"epoch": 1.6394589552238807,
"grad_norm": 0.4459067435418286,
"learning_rate": 4.114935970732178e-05,
"loss": 0.5067,
"step": 14060
},
{
"epoch": 1.6400419776119404,
"grad_norm": 0.4291682494710737,
"learning_rate": 4.114246139227069e-05,
"loss": 0.5263,
"step": 14065
},
{
"epoch": 1.640625,
"grad_norm": 0.4216746640472223,
"learning_rate": 4.113556104871631e-05,
"loss": 0.4964,
"step": 14070
},
{
"epoch": 1.6412080223880596,
"grad_norm": 0.4211314766595737,
"learning_rate": 4.112865867768464e-05,
"loss": 0.4808,
"step": 14075
},
{
"epoch": 1.6417910447761193,
"grad_norm": 0.4053437292411652,
"learning_rate": 4.112175428020199e-05,
"loss": 0.4461,
"step": 14080
},
{
"epoch": 1.642374067164179,
"grad_norm": 0.396670578607788,
"learning_rate": 4.1114847857295006e-05,
"loss": 0.4605,
"step": 14085
},
{
"epoch": 1.642957089552239,
"grad_norm": 0.3995791356492982,
"learning_rate": 4.110793940999059e-05,
"loss": 0.4711,
"step": 14090
},
{
"epoch": 1.6435401119402986,
"grad_norm": 0.3954757796377986,
"learning_rate": 4.110102893931597e-05,
"loss": 0.4994,
"step": 14095
},
{
"epoch": 1.6441231343283582,
"grad_norm": 0.435294454301607,
"learning_rate": 4.1094116446298645e-05,
"loss": 0.4766,
"step": 14100
},
{
"epoch": 1.6447061567164178,
"grad_norm": 0.44310065079643096,
"learning_rate": 4.1087201931966463e-05,
"loss": 0.5113,
"step": 14105
},
{
"epoch": 1.6452891791044775,
"grad_norm": 0.3607871264105027,
"learning_rate": 4.108028539734753e-05,
"loss": 0.4855,
"step": 14110
},
{
"epoch": 1.6458722014925373,
"grad_norm": 0.42095392810100446,
"learning_rate": 4.1073366843470285e-05,
"loss": 0.4639,
"step": 14115
},
{
"epoch": 1.6464552238805972,
"grad_norm": 0.3766418494165451,
"learning_rate": 4.1066446271363426e-05,
"loss": 0.4926,
"step": 14120
},
{
"epoch": 1.6470382462686568,
"grad_norm": 0.4443662116425456,
"learning_rate": 4.1059523682056e-05,
"loss": 0.4919,
"step": 14125
},
{
"epoch": 1.6476212686567164,
"grad_norm": 0.44464467585543765,
"learning_rate": 4.1052599076577306e-05,
"loss": 0.5104,
"step": 14130
},
{
"epoch": 1.648204291044776,
"grad_norm": 0.4047224424351832,
"learning_rate": 4.104567245595699e-05,
"loss": 0.4564,
"step": 14135
},
{
"epoch": 1.6487873134328357,
"grad_norm": 0.42026335706897944,
"learning_rate": 4.103874382122496e-05,
"loss": 0.4832,
"step": 14140
},
{
"epoch": 1.6493703358208955,
"grad_norm": 0.4107997789237667,
"learning_rate": 4.103181317341144e-05,
"loss": 0.467,
"step": 14145
},
{
"epoch": 1.6499533582089554,
"grad_norm": 0.39889095294561816,
"learning_rate": 4.1024880513546955e-05,
"loss": 0.4714,
"step": 14150
},
{
"epoch": 1.650536380597015,
"grad_norm": 0.46962418152356855,
"learning_rate": 4.101794584266232e-05,
"loss": 0.4707,
"step": 14155
},
{
"epoch": 1.6511194029850746,
"grad_norm": 0.45018302237534263,
"learning_rate": 4.1011009161788655e-05,
"loss": 0.4952,
"step": 14160
},
{
"epoch": 1.6517024253731343,
"grad_norm": 0.3961606609630567,
"learning_rate": 4.100407047195738e-05,
"loss": 0.4712,
"step": 14165
},
{
"epoch": 1.652285447761194,
"grad_norm": 0.42032988863236986,
"learning_rate": 4.099712977420021e-05,
"loss": 0.4548,
"step": 14170
},
{
"epoch": 1.6528684701492538,
"grad_norm": 0.3703946722633511,
"learning_rate": 4.099018706954916e-05,
"loss": 0.4416,
"step": 14175
},
{
"epoch": 1.6534514925373134,
"grad_norm": 0.4098114340694798,
"learning_rate": 4.098324235903655e-05,
"loss": 0.4745,
"step": 14180
},
{
"epoch": 1.6540345149253732,
"grad_norm": 0.40705381450134037,
"learning_rate": 4.0976295643694986e-05,
"loss": 0.5113,
"step": 14185
},
{
"epoch": 1.6546175373134329,
"grad_norm": 0.43040331236645074,
"learning_rate": 4.0969346924557374e-05,
"loss": 0.5035,
"step": 14190
},
{
"epoch": 1.6552005597014925,
"grad_norm": 0.4405551738954265,
"learning_rate": 4.096239620265693e-05,
"loss": 0.4878,
"step": 14195
},
{
"epoch": 1.6557835820895521,
"grad_norm": 0.4967926559533798,
"learning_rate": 4.095544347902715e-05,
"loss": 0.5059,
"step": 14200
},
{
"epoch": 1.656366604477612,
"grad_norm": 0.4095083869310005,
"learning_rate": 4.0948488754701846e-05,
"loss": 0.4885,
"step": 14205
},
{
"epoch": 1.6569496268656716,
"grad_norm": 0.47061671668377314,
"learning_rate": 4.094153203071512e-05,
"loss": 0.5638,
"step": 14210
},
{
"epoch": 1.6575326492537314,
"grad_norm": 0.4081933179265426,
"learning_rate": 4.0934573308101376e-05,
"loss": 0.4959,
"step": 14215
},
{
"epoch": 1.658115671641791,
"grad_norm": 0.3996955520067459,
"learning_rate": 4.092761258789529e-05,
"loss": 0.4474,
"step": 14220
},
{
"epoch": 1.6586986940298507,
"grad_norm": 0.4314479874396754,
"learning_rate": 4.092064987113186e-05,
"loss": 0.4718,
"step": 14225
},
{
"epoch": 1.6592817164179103,
"grad_norm": 0.4438299110909517,
"learning_rate": 4.091368515884638e-05,
"loss": 0.5341,
"step": 14230
},
{
"epoch": 1.6598647388059702,
"grad_norm": 0.41651006798955637,
"learning_rate": 4.0906718452074435e-05,
"loss": 0.4727,
"step": 14235
},
{
"epoch": 1.6604477611940298,
"grad_norm": 0.4331091906462682,
"learning_rate": 4.089974975185192e-05,
"loss": 0.4836,
"step": 14240
},
{
"epoch": 1.6610307835820897,
"grad_norm": 0.4021683622488243,
"learning_rate": 4.0892779059214994e-05,
"loss": 0.497,
"step": 14245
},
{
"epoch": 1.6616138059701493,
"grad_norm": 0.4117795790481671,
"learning_rate": 4.088580637520015e-05,
"loss": 0.4425,
"step": 14250
},
{
"epoch": 1.662196828358209,
"grad_norm": 0.3895811651306167,
"learning_rate": 4.087883170084414e-05,
"loss": 0.4824,
"step": 14255
},
{
"epoch": 1.6627798507462686,
"grad_norm": 0.37307335490523996,
"learning_rate": 4.087185503718404e-05,
"loss": 0.4462,
"step": 14260
},
{
"epoch": 1.6633628731343284,
"grad_norm": 0.48478137335603105,
"learning_rate": 4.0864876385257225e-05,
"loss": 0.4726,
"step": 14265
},
{
"epoch": 1.663945895522388,
"grad_norm": 0.4395735769638866,
"learning_rate": 4.0857895746101335e-05,
"loss": 0.5045,
"step": 14270
},
{
"epoch": 1.6645289179104479,
"grad_norm": 0.4395663397458483,
"learning_rate": 4.085091312075434e-05,
"loss": 0.4661,
"step": 14275
},
{
"epoch": 1.6651119402985075,
"grad_norm": 0.3518701273075788,
"learning_rate": 4.084392851025447e-05,
"loss": 0.432,
"step": 14280
},
{
"epoch": 1.6656949626865671,
"grad_norm": 0.3887442406304548,
"learning_rate": 4.083694191564028e-05,
"loss": 0.4702,
"step": 14285
},
{
"epoch": 1.6662779850746268,
"grad_norm": 0.4184054471272332,
"learning_rate": 4.082995333795063e-05,
"loss": 0.4889,
"step": 14290
},
{
"epoch": 1.6668610074626866,
"grad_norm": 0.4173370886409284,
"learning_rate": 4.0822962778224613e-05,
"loss": 0.4917,
"step": 14295
},
{
"epoch": 1.6674440298507462,
"grad_norm": 0.4078412810184441,
"learning_rate": 4.081597023750169e-05,
"loss": 0.4649,
"step": 14300
},
{
"epoch": 1.668027052238806,
"grad_norm": 0.42994252550274004,
"learning_rate": 4.0808975716821574e-05,
"loss": 0.4807,
"step": 14305
},
{
"epoch": 1.6686100746268657,
"grad_norm": 0.39193144870866914,
"learning_rate": 4.0801979217224285e-05,
"loss": 0.4667,
"step": 14310
},
{
"epoch": 1.6691930970149254,
"grad_norm": 0.438265726796618,
"learning_rate": 4.079498073975013e-05,
"loss": 0.4994,
"step": 14315
},
{
"epoch": 1.669776119402985,
"grad_norm": 0.3858463575868289,
"learning_rate": 4.078798028543974e-05,
"loss": 0.4716,
"step": 14320
},
{
"epoch": 1.6703591417910446,
"grad_norm": 0.4501937346570606,
"learning_rate": 4.078097785533398e-05,
"loss": 0.4999,
"step": 14325
},
{
"epoch": 1.6709421641791045,
"grad_norm": 0.5008841356624519,
"learning_rate": 4.0773973450474055e-05,
"loss": 0.4859,
"step": 14330
},
{
"epoch": 1.6715251865671643,
"grad_norm": 0.42419460599067715,
"learning_rate": 4.076696707190147e-05,
"loss": 0.4636,
"step": 14335
},
{
"epoch": 1.672108208955224,
"grad_norm": 0.4280915136097193,
"learning_rate": 4.0759958720658e-05,
"loss": 0.4702,
"step": 14340
},
{
"epoch": 1.6726912313432836,
"grad_norm": 0.4195309591052,
"learning_rate": 4.07529483977857e-05,
"loss": 0.4785,
"step": 14345
},
{
"epoch": 1.6732742537313432,
"grad_norm": 0.4035026247735428,
"learning_rate": 4.074593610432695e-05,
"loss": 0.4677,
"step": 14350
},
{
"epoch": 1.6738572761194028,
"grad_norm": 0.37961678043272695,
"learning_rate": 4.073892184132442e-05,
"loss": 0.478,
"step": 14355
},
{
"epoch": 1.6744402985074627,
"grad_norm": 0.395602393674479,
"learning_rate": 4.073190560982106e-05,
"loss": 0.4684,
"step": 14360
},
{
"epoch": 1.6750233208955225,
"grad_norm": 0.42332378607544247,
"learning_rate": 4.072488741086011e-05,
"loss": 0.4963,
"step": 14365
},
{
"epoch": 1.6756063432835822,
"grad_norm": 0.469879001605267,
"learning_rate": 4.071786724548511e-05,
"loss": 0.4671,
"step": 14370
},
{
"epoch": 1.6761893656716418,
"grad_norm": 0.3733727950920635,
"learning_rate": 4.07108451147399e-05,
"loss": 0.4872,
"step": 14375
},
{
"epoch": 1.6767723880597014,
"grad_norm": 0.46457269040707083,
"learning_rate": 4.07038210196686e-05,
"loss": 0.4792,
"step": 14380
},
{
"epoch": 1.677355410447761,
"grad_norm": 0.429251035541327,
"learning_rate": 4.0696794961315605e-05,
"loss": 0.5056,
"step": 14385
},
{
"epoch": 1.677938432835821,
"grad_norm": 0.3983369316978038,
"learning_rate": 4.068976694072565e-05,
"loss": 0.4526,
"step": 14390
},
{
"epoch": 1.6785214552238807,
"grad_norm": 0.40331575606174813,
"learning_rate": 4.068273695894373e-05,
"loss": 0.471,
"step": 14395
},
{
"epoch": 1.6791044776119404,
"grad_norm": 0.40640282769794667,
"learning_rate": 4.067570501701513e-05,
"loss": 0.4954,
"step": 14400
},
{
"epoch": 1.6796875,
"grad_norm": 0.40774451617277263,
"learning_rate": 4.066867111598542e-05,
"loss": 0.4865,
"step": 14405
},
{
"epoch": 1.6802705223880596,
"grad_norm": 0.36969606212151795,
"learning_rate": 4.0661635256900505e-05,
"loss": 0.4753,
"step": 14410
},
{
"epoch": 1.6808535447761193,
"grad_norm": 0.42003634711138027,
"learning_rate": 4.065459744080652e-05,
"loss": 0.4599,
"step": 14415
},
{
"epoch": 1.681436567164179,
"grad_norm": 0.4042331899360192,
"learning_rate": 4.064755766874993e-05,
"loss": 0.462,
"step": 14420
},
{
"epoch": 1.682019589552239,
"grad_norm": 0.4189680958836512,
"learning_rate": 4.064051594177749e-05,
"loss": 0.486,
"step": 14425
},
{
"epoch": 1.6826026119402986,
"grad_norm": 0.43460367229952235,
"learning_rate": 4.0633472260936224e-05,
"loss": 0.4778,
"step": 14430
},
{
"epoch": 1.6831856343283582,
"grad_norm": 0.4118904647153084,
"learning_rate": 4.0626426627273474e-05,
"loss": 0.4786,
"step": 14435
},
{
"epoch": 1.6837686567164178,
"grad_norm": 0.37638816320378793,
"learning_rate": 4.061937904183685e-05,
"loss": 0.4494,
"step": 14440
},
{
"epoch": 1.6843516791044775,
"grad_norm": 0.3805988153836944,
"learning_rate": 4.061232950567427e-05,
"loss": 0.5114,
"step": 14445
},
{
"epoch": 1.6849347014925373,
"grad_norm": 0.4358490040848953,
"learning_rate": 4.060527801983391e-05,
"loss": 0.4923,
"step": 14450
},
{
"epoch": 1.6855177238805972,
"grad_norm": 0.42111081737665557,
"learning_rate": 4.0598224585364276e-05,
"loss": 0.4918,
"step": 14455
},
{
"epoch": 1.6861007462686568,
"grad_norm": 0.4466995410046454,
"learning_rate": 4.0591169203314145e-05,
"loss": 0.4887,
"step": 14460
},
{
"epoch": 1.6866837686567164,
"grad_norm": 0.427094980705651,
"learning_rate": 4.058411187473257e-05,
"loss": 0.4939,
"step": 14465
},
{
"epoch": 1.687266791044776,
"grad_norm": 0.44279907161841936,
"learning_rate": 4.057705260066894e-05,
"loss": 0.5105,
"step": 14470
},
{
"epoch": 1.6878498134328357,
"grad_norm": 0.5717252706474644,
"learning_rate": 4.056999138217287e-05,
"loss": 0.4705,
"step": 14475
},
{
"epoch": 1.6884328358208955,
"grad_norm": 0.4971844718005374,
"learning_rate": 4.056292822029432e-05,
"loss": 0.5335,
"step": 14480
},
{
"epoch": 1.6890158582089554,
"grad_norm": 0.4067642994557471,
"learning_rate": 4.055586311608349e-05,
"loss": 0.4599,
"step": 14485
},
{
"epoch": 1.689598880597015,
"grad_norm": 0.395307690446041,
"learning_rate": 4.05487960705909e-05,
"loss": 0.4821,
"step": 14490
},
{
"epoch": 1.6901819029850746,
"grad_norm": 0.42006160426752454,
"learning_rate": 4.054172708486737e-05,
"loss": 0.489,
"step": 14495
},
{
"epoch": 1.6907649253731343,
"grad_norm": 0.41681042217877773,
"learning_rate": 4.053465615996397e-05,
"loss": 0.4973,
"step": 14500
},
{
"epoch": 1.691347947761194,
"grad_norm": 0.4317642030407513,
"learning_rate": 4.05275832969321e-05,
"loss": 0.4785,
"step": 14505
},
{
"epoch": 1.6919309701492538,
"grad_norm": 0.4183629465011244,
"learning_rate": 4.0520508496823395e-05,
"loss": 0.5063,
"step": 14510
},
{
"epoch": 1.6925139925373134,
"grad_norm": 0.4227104803821206,
"learning_rate": 4.051343176068984e-05,
"loss": 0.512,
"step": 14515
},
{
"epoch": 1.6930970149253732,
"grad_norm": 0.39716646367954495,
"learning_rate": 4.050635308958366e-05,
"loss": 0.4791,
"step": 14520
},
{
"epoch": 1.6936800373134329,
"grad_norm": 0.4315919077606542,
"learning_rate": 4.04992724845574e-05,
"loss": 0.496,
"step": 14525
},
{
"epoch": 1.6942630597014925,
"grad_norm": 0.4190476408707576,
"learning_rate": 4.0492189946663864e-05,
"loss": 0.4841,
"step": 14530
},
{
"epoch": 1.6948460820895521,
"grad_norm": 0.39975994216386457,
"learning_rate": 4.048510547695616e-05,
"loss": 0.4635,
"step": 14535
},
{
"epoch": 1.695429104477612,
"grad_norm": 0.4021707350007361,
"learning_rate": 4.047801907648769e-05,
"loss": 0.4667,
"step": 14540
},
{
"epoch": 1.6960121268656716,
"grad_norm": 0.5246904053875769,
"learning_rate": 4.047093074631213e-05,
"loss": 0.471,
"step": 14545
},
{
"epoch": 1.6965951492537314,
"grad_norm": 0.41206111863478134,
"learning_rate": 4.046384048748344e-05,
"loss": 0.4911,
"step": 14550
},
{
"epoch": 1.697178171641791,
"grad_norm": 0.3873183615800148,
"learning_rate": 4.045674830105587e-05,
"loss": 0.4604,
"step": 14555
},
{
"epoch": 1.6977611940298507,
"grad_norm": 0.4486891743823356,
"learning_rate": 4.0449654188083985e-05,
"loss": 0.4897,
"step": 14560
},
{
"epoch": 1.6983442164179103,
"grad_norm": 0.44464525267632954,
"learning_rate": 4.0442558149622586e-05,
"loss": 0.4916,
"step": 14565
},
{
"epoch": 1.6989272388059702,
"grad_norm": 0.4310393568480325,
"learning_rate": 4.04354601867268e-05,
"loss": 0.4586,
"step": 14570
},
{
"epoch": 1.6995102611940298,
"grad_norm": 0.4134083766327794,
"learning_rate": 4.0428360300452024e-05,
"loss": 0.4872,
"step": 14575
},
{
"epoch": 1.7000932835820897,
"grad_norm": 0.44318935135672344,
"learning_rate": 4.042125849185394e-05,
"loss": 0.5347,
"step": 14580
},
{
"epoch": 1.7006763059701493,
"grad_norm": 0.44538180422620816,
"learning_rate": 4.0414154761988506e-05,
"loss": 0.4964,
"step": 14585
},
{
"epoch": 1.701259328358209,
"grad_norm": 0.4207916509197397,
"learning_rate": 4.040704911191201e-05,
"loss": 0.4929,
"step": 14590
},
{
"epoch": 1.7018423507462686,
"grad_norm": 0.42148426727773364,
"learning_rate": 4.0399941542680956e-05,
"loss": 0.4994,
"step": 14595
},
{
"epoch": 1.7024253731343284,
"grad_norm": 0.4197532755663317,
"learning_rate": 4.0392832055352205e-05,
"loss": 0.4699,
"step": 14600
},
{
"epoch": 1.703008395522388,
"grad_norm": 0.39751059891449686,
"learning_rate": 4.038572065098286e-05,
"loss": 0.5227,
"step": 14605
},
{
"epoch": 1.7035914179104479,
"grad_norm": 0.40816550933831536,
"learning_rate": 4.0378607330630304e-05,
"loss": 0.4642,
"step": 14610
},
{
"epoch": 1.7041744402985075,
"grad_norm": 0.3755709499880748,
"learning_rate": 4.037149209535222e-05,
"loss": 0.4619,
"step": 14615
},
{
"epoch": 1.7047574626865671,
"grad_norm": 0.41100106429253497,
"learning_rate": 4.036437494620661e-05,
"loss": 0.4661,
"step": 14620
},
{
"epoch": 1.7053404850746268,
"grad_norm": 0.44133548984910614,
"learning_rate": 4.0357255884251694e-05,
"loss": 0.4946,
"step": 14625
},
{
"epoch": 1.7059235074626866,
"grad_norm": 0.40752151523015684,
"learning_rate": 4.0350134910546e-05,
"loss": 0.4846,
"step": 14630
},
{
"epoch": 1.7065065298507462,
"grad_norm": 0.47322208410241956,
"learning_rate": 4.0343012026148384e-05,
"loss": 0.4784,
"step": 14635
},
{
"epoch": 1.707089552238806,
"grad_norm": 0.40403758744911183,
"learning_rate": 4.033588723211793e-05,
"loss": 0.4802,
"step": 14640
},
{
"epoch": 1.7076725746268657,
"grad_norm": 0.4412551393571504,
"learning_rate": 4.032876052951402e-05,
"loss": 0.5116,
"step": 14645
},
{
"epoch": 1.7082555970149254,
"grad_norm": 0.4082133615947316,
"learning_rate": 4.032163191939633e-05,
"loss": 0.4833,
"step": 14650
},
{
"epoch": 1.708838619402985,
"grad_norm": 0.44662383748583084,
"learning_rate": 4.0314501402824825e-05,
"loss": 0.4675,
"step": 14655
},
{
"epoch": 1.7094216417910446,
"grad_norm": 0.456074311014277,
"learning_rate": 4.030736898085974e-05,
"loss": 0.4761,
"step": 14660
},
{
"epoch": 1.7100046641791045,
"grad_norm": 0.42416393425691723,
"learning_rate": 4.03002346545616e-05,
"loss": 0.4865,
"step": 14665
},
{
"epoch": 1.7105876865671643,
"grad_norm": 0.40905104456516456,
"learning_rate": 4.02930984249912e-05,
"loss": 0.4791,
"step": 14670
},
{
"epoch": 1.711170708955224,
"grad_norm": 0.38741083727404313,
"learning_rate": 4.028596029320965e-05,
"loss": 0.4933,
"step": 14675
},
{
"epoch": 1.7117537313432836,
"grad_norm": 0.3984672434837992,
"learning_rate": 4.02788202602783e-05,
"loss": 0.4811,
"step": 14680
},
{
"epoch": 1.7123367537313432,
"grad_norm": 0.42142094914020906,
"learning_rate": 4.027167832725882e-05,
"loss": 0.4735,
"step": 14685
},
{
"epoch": 1.7129197761194028,
"grad_norm": 0.39812681935895405,
"learning_rate": 4.026453449521313e-05,
"loss": 0.5039,
"step": 14690
},
{
"epoch": 1.7135027985074627,
"grad_norm": 0.3855065197526314,
"learning_rate": 4.025738876520347e-05,
"loss": 0.4829,
"step": 14695
},
{
"epoch": 1.7140858208955225,
"grad_norm": 0.44644473762754755,
"learning_rate": 4.025024113829233e-05,
"loss": 0.4941,
"step": 14700
},
{
"epoch": 1.7146688432835822,
"grad_norm": 0.37300083226666375,
"learning_rate": 4.024309161554249e-05,
"loss": 0.4784,
"step": 14705
},
{
"epoch": 1.7152518656716418,
"grad_norm": 0.40617514898541857,
"learning_rate": 4.023594019801702e-05,
"loss": 0.4744,
"step": 14710
},
{
"epoch": 1.7158348880597014,
"grad_norm": 0.3840399029109539,
"learning_rate": 4.022878688677927e-05,
"loss": 0.4706,
"step": 14715
},
{
"epoch": 1.716417910447761,
"grad_norm": 0.36819401392874374,
"learning_rate": 4.022163168289287e-05,
"loss": 0.4714,
"step": 14720
},
{
"epoch": 1.717000932835821,
"grad_norm": 0.4237902145668965,
"learning_rate": 4.021447458742172e-05,
"loss": 0.4819,
"step": 14725
},
{
"epoch": 1.7175839552238807,
"grad_norm": 0.6219554874705321,
"learning_rate": 4.020731560143002e-05,
"loss": 0.4804,
"step": 14730
},
{
"epoch": 1.7181669776119404,
"grad_norm": 0.5096288424326689,
"learning_rate": 4.0200154725982245e-05,
"loss": 0.5171,
"step": 14735
},
{
"epoch": 1.71875,
"grad_norm": 0.46719677801906906,
"learning_rate": 4.019299196214315e-05,
"loss": 0.4781,
"step": 14740
},
{
"epoch": 1.7193330223880596,
"grad_norm": 0.41664870706254153,
"learning_rate": 4.0185827310977756e-05,
"loss": 0.4694,
"step": 14745
},
{
"epoch": 1.7199160447761193,
"grad_norm": 0.42107627767474415,
"learning_rate": 4.017866077355139e-05,
"loss": 0.4978,
"step": 14750
},
{
"epoch": 1.720499067164179,
"grad_norm": 0.4541677559740301,
"learning_rate": 4.0171492350929644e-05,
"loss": 0.4864,
"step": 14755
},
{
"epoch": 1.721082089552239,
"grad_norm": 0.4448482281379494,
"learning_rate": 4.016432204417839e-05,
"loss": 0.5001,
"step": 14760
},
{
"epoch": 1.7216651119402986,
"grad_norm": 0.4382211385559882,
"learning_rate": 4.015714985436379e-05,
"loss": 0.4768,
"step": 14765
},
{
"epoch": 1.7222481343283582,
"grad_norm": 0.43206927833480646,
"learning_rate": 4.014997578255227e-05,
"loss": 0.4862,
"step": 14770
},
{
"epoch": 1.7228311567164178,
"grad_norm": 0.36936715040033663,
"learning_rate": 4.014279982981057e-05,
"loss": 0.4378,
"step": 14775
},
{
"epoch": 1.7234141791044775,
"grad_norm": 0.4120603192454238,
"learning_rate": 4.0135621997205654e-05,
"loss": 0.4609,
"step": 14780
},
{
"epoch": 1.7239972014925373,
"grad_norm": 0.40226922820462935,
"learning_rate": 4.0128442285804815e-05,
"loss": 0.4681,
"step": 14785
},
{
"epoch": 1.7245802238805972,
"grad_norm": 0.4159642213364248,
"learning_rate": 4.01212606966756e-05,
"loss": 0.4813,
"step": 14790
},
{
"epoch": 1.7251632462686568,
"grad_norm": 0.4192546917299525,
"learning_rate": 4.0114077230885847e-05,
"loss": 0.5048,
"step": 14795
},
{
"epoch": 1.7257462686567164,
"grad_norm": 0.4259810736862134,
"learning_rate": 4.010689188950367e-05,
"loss": 0.5066,
"step": 14800
},
{
"epoch": 1.726329291044776,
"grad_norm": 0.38815377067775536,
"learning_rate": 4.009970467359746e-05,
"loss": 0.4957,
"step": 14805
},
{
"epoch": 1.7269123134328357,
"grad_norm": 0.4709225043046578,
"learning_rate": 4.009251558423588e-05,
"loss": 0.4877,
"step": 14810
},
{
"epoch": 1.7274953358208955,
"grad_norm": 0.4376423324231026,
"learning_rate": 4.008532462248789e-05,
"loss": 0.4778,
"step": 14815
},
{
"epoch": 1.7280783582089554,
"grad_norm": 0.4088402780943302,
"learning_rate": 4.00781317894227e-05,
"loss": 0.4791,
"step": 14820
},
{
"epoch": 1.728661380597015,
"grad_norm": 0.42516785610240804,
"learning_rate": 4.007093708610984e-05,
"loss": 0.4941,
"step": 14825
},
{
"epoch": 1.7292444029850746,
"grad_norm": 0.4350060525008231,
"learning_rate": 4.006374051361907e-05,
"loss": 0.4695,
"step": 14830
},
{
"epoch": 1.7298274253731343,
"grad_norm": 0.4314369115933184,
"learning_rate": 4.005654207302047e-05,
"loss": 0.4801,
"step": 14835
},
{
"epoch": 1.730410447761194,
"grad_norm": 0.3774773619724921,
"learning_rate": 4.004934176538436e-05,
"loss": 0.4677,
"step": 14840
},
{
"epoch": 1.7309934701492538,
"grad_norm": 0.4382139271394396,
"learning_rate": 4.004213959178137e-05,
"loss": 0.4693,
"step": 14845
},
{
"epoch": 1.7315764925373134,
"grad_norm": 0.44729161745585944,
"learning_rate": 4.0034935553282396e-05,
"loss": 0.5012,
"step": 14850
},
{
"epoch": 1.7321595149253732,
"grad_norm": 0.46312961873331987,
"learning_rate": 4.00277296509586e-05,
"loss": 0.4723,
"step": 14855
},
{
"epoch": 1.7327425373134329,
"grad_norm": 0.4268816266860399,
"learning_rate": 4.002052188588144e-05,
"loss": 0.5003,
"step": 14860
},
{
"epoch": 1.7333255597014925,
"grad_norm": 0.47299567394079345,
"learning_rate": 4.001331225912263e-05,
"loss": 0.4902,
"step": 14865
},
{
"epoch": 1.7339085820895521,
"grad_norm": 0.45090820016715927,
"learning_rate": 4.000610077175419e-05,
"loss": 0.5036,
"step": 14870
},
{
"epoch": 1.734491604477612,
"grad_norm": 0.4188897717540988,
"learning_rate": 3.999888742484838e-05,
"loss": 0.4765,
"step": 14875
},
{
"epoch": 1.7350746268656716,
"grad_norm": 0.38053316117089947,
"learning_rate": 3.999167221947777e-05,
"loss": 0.4722,
"step": 14880
},
{
"epoch": 1.7356576492537314,
"grad_norm": 0.4385997518493541,
"learning_rate": 3.9984455156715176e-05,
"loss": 0.4869,
"step": 14885
},
{
"epoch": 1.736240671641791,
"grad_norm": 0.44990818417600253,
"learning_rate": 3.997723623763372e-05,
"loss": 0.4819,
"step": 14890
},
{
"epoch": 1.7368236940298507,
"grad_norm": 0.4126902232835701,
"learning_rate": 3.997001546330679e-05,
"loss": 0.4975,
"step": 14895
},
{
"epoch": 1.7374067164179103,
"grad_norm": 0.4465958329751703,
"learning_rate": 3.9962792834808034e-05,
"loss": 0.4966,
"step": 14900
},
{
"epoch": 1.7379897388059702,
"grad_norm": 0.4092713940163117,
"learning_rate": 3.9955568353211384e-05,
"loss": 0.4841,
"step": 14905
},
{
"epoch": 1.7385727611940298,
"grad_norm": 0.39784336874622206,
"learning_rate": 3.9948342019591066e-05,
"loss": 0.4578,
"step": 14910
},
{
"epoch": 1.7391557835820897,
"grad_norm": 0.4304337157634813,
"learning_rate": 3.994111383502156e-05,
"loss": 0.508,
"step": 14915
},
{
"epoch": 1.7397388059701493,
"grad_norm": 0.7312527517732008,
"learning_rate": 3.993388380057763e-05,
"loss": 0.4789,
"step": 14920
},
{
"epoch": 1.740321828358209,
"grad_norm": 0.4550878792432819,
"learning_rate": 3.992665191733431e-05,
"loss": 0.4904,
"step": 14925
},
{
"epoch": 1.7409048507462686,
"grad_norm": 0.3994822719676129,
"learning_rate": 3.9919418186366905e-05,
"loss": 0.4608,
"step": 14930
},
{
"epoch": 1.7414878731343284,
"grad_norm": 0.4625317998459724,
"learning_rate": 3.991218260875101e-05,
"loss": 0.5032,
"step": 14935
},
{
"epoch": 1.742070895522388,
"grad_norm": 0.38077012965226736,
"learning_rate": 3.9904945185562484e-05,
"loss": 0.4565,
"step": 14940
},
{
"epoch": 1.7426539179104479,
"grad_norm": 0.4480502431336596,
"learning_rate": 3.989770591787747e-05,
"loss": 0.522,
"step": 14945
},
{
"epoch": 1.7432369402985075,
"grad_norm": 0.398867521435284,
"learning_rate": 3.989046480677236e-05,
"loss": 0.465,
"step": 14950
},
{
"epoch": 1.7438199626865671,
"grad_norm": 0.43754836115769846,
"learning_rate": 3.988322185332386e-05,
"loss": 0.484,
"step": 14955
},
{
"epoch": 1.7444029850746268,
"grad_norm": 0.41737523281789957,
"learning_rate": 3.987597705860891e-05,
"loss": 0.4685,
"step": 14960
},
{
"epoch": 1.7449860074626866,
"grad_norm": 0.43664087867694085,
"learning_rate": 3.9868730423704754e-05,
"loss": 0.4881,
"step": 14965
},
{
"epoch": 1.7455690298507462,
"grad_norm": 0.42139600304341013,
"learning_rate": 3.986148194968888e-05,
"loss": 0.4855,
"step": 14970
},
{
"epoch": 1.746152052238806,
"grad_norm": 0.5222225528877735,
"learning_rate": 3.985423163763909e-05,
"loss": 0.4786,
"step": 14975
},
{
"epoch": 1.7467350746268657,
"grad_norm": 0.4322503978118561,
"learning_rate": 3.9846979488633415e-05,
"loss": 0.4846,
"step": 14980
},
{
"epoch": 1.7473180970149254,
"grad_norm": 0.41162470239800886,
"learning_rate": 3.9839725503750185e-05,
"loss": 0.4843,
"step": 14985
},
{
"epoch": 1.747901119402985,
"grad_norm": 0.4231023984292742,
"learning_rate": 3.9832469684068007e-05,
"loss": 0.4647,
"step": 14990
},
{
"epoch": 1.7484841417910446,
"grad_norm": 0.4306464601193167,
"learning_rate": 3.982521203066575e-05,
"loss": 0.4401,
"step": 14995
},
{
"epoch": 1.7490671641791045,
"grad_norm": 0.40830830901483156,
"learning_rate": 3.9817952544622554e-05,
"loss": 0.4667,
"step": 15000
},
{
"epoch": 1.7496501865671643,
"grad_norm": 0.44644806445507973,
"learning_rate": 3.981069122701784e-05,
"loss": 0.5094,
"step": 15005
},
{
"epoch": 1.750233208955224,
"grad_norm": 0.4111577936580399,
"learning_rate": 3.9803428078931276e-05,
"loss": 0.5006,
"step": 15010
},
{
"epoch": 1.7508162313432836,
"grad_norm": 0.42627297333764413,
"learning_rate": 3.979616310144284e-05,
"loss": 0.4789,
"step": 15015
},
{
"epoch": 1.7513992537313432,
"grad_norm": 0.4058739814998778,
"learning_rate": 3.978889629563277e-05,
"loss": 0.4479,
"step": 15020
},
{
"epoch": 1.7519822761194028,
"grad_norm": 0.38798700870798036,
"learning_rate": 3.9781627662581575e-05,
"loss": 0.4926,
"step": 15025
},
{
"epoch": 1.7525652985074627,
"grad_norm": 0.4221107232729366,
"learning_rate": 3.977435720337e-05,
"loss": 0.483,
"step": 15030
},
{
"epoch": 1.7531483208955225,
"grad_norm": 0.45583926648182216,
"learning_rate": 3.976708491907912e-05,
"loss": 0.4824,
"step": 15035
},
{
"epoch": 1.7537313432835822,
"grad_norm": 0.4169946301163216,
"learning_rate": 3.9759810810790236e-05,
"loss": 0.4917,
"step": 15040
},
{
"epoch": 1.7543143656716418,
"grad_norm": 0.4210033692145867,
"learning_rate": 3.9752534879584954e-05,
"loss": 0.4882,
"step": 15045
},
{
"epoch": 1.7548973880597014,
"grad_norm": 0.466250165685237,
"learning_rate": 3.9745257126545146e-05,
"loss": 0.4934,
"step": 15050
},
{
"epoch": 1.755480410447761,
"grad_norm": 0.43642385025225544,
"learning_rate": 3.973797755275291e-05,
"loss": 0.4547,
"step": 15055
},
{
"epoch": 1.756063432835821,
"grad_norm": 0.4667562211514214,
"learning_rate": 3.9730696159290656e-05,
"loss": 0.4952,
"step": 15060
},
{
"epoch": 1.7566464552238807,
"grad_norm": 0.44193754300987165,
"learning_rate": 3.9723412947241085e-05,
"loss": 0.4613,
"step": 15065
},
{
"epoch": 1.7572294776119404,
"grad_norm": 0.41844044588590157,
"learning_rate": 3.971612791768712e-05,
"loss": 0.4676,
"step": 15070
},
{
"epoch": 1.7578125,
"grad_norm": 0.39445835046273797,
"learning_rate": 3.970884107171198e-05,
"loss": 0.4851,
"step": 15075
},
{
"epoch": 1.7583955223880596,
"grad_norm": 0.405745262922451,
"learning_rate": 3.970155241039914e-05,
"loss": 0.4777,
"step": 15080
},
{
"epoch": 1.7589785447761193,
"grad_norm": 0.40216231559136584,
"learning_rate": 3.969426193483237e-05,
"loss": 0.4627,
"step": 15085
},
{
"epoch": 1.759561567164179,
"grad_norm": 0.45390641354272226,
"learning_rate": 3.968696964609568e-05,
"loss": 0.4844,
"step": 15090
},
{
"epoch": 1.760144589552239,
"grad_norm": 0.43424950082620944,
"learning_rate": 3.967967554527338e-05,
"loss": 0.5153,
"step": 15095
},
{
"epoch": 1.7607276119402986,
"grad_norm": 0.4313331182458118,
"learning_rate": 3.967237963345001e-05,
"loss": 0.4671,
"step": 15100
},
{
"epoch": 1.7613106343283582,
"grad_norm": 0.4189378906681823,
"learning_rate": 3.966508191171041e-05,
"loss": 0.4915,
"step": 15105
},
{
"epoch": 1.7618936567164178,
"grad_norm": 0.4284044452739007,
"learning_rate": 3.9657782381139696e-05,
"loss": 0.4774,
"step": 15110
},
{
"epoch": 1.7624766791044775,
"grad_norm": 0.3976581151912186,
"learning_rate": 3.965048104282323e-05,
"loss": 0.5076,
"step": 15115
},
{
"epoch": 1.7630597014925373,
"grad_norm": 0.4121697800037609,
"learning_rate": 3.964317789784664e-05,
"loss": 0.4844,
"step": 15120
},
{
"epoch": 1.7636427238805972,
"grad_norm": 0.40692894613398783,
"learning_rate": 3.963587294729585e-05,
"loss": 0.4911,
"step": 15125
},
{
"epoch": 1.7642257462686568,
"grad_norm": 0.40246451021405094,
"learning_rate": 3.962856619225703e-05,
"loss": 0.488,
"step": 15130
},
{
"epoch": 1.7648087686567164,
"grad_norm": 0.393330174653249,
"learning_rate": 3.962125763381661e-05,
"loss": 0.4561,
"step": 15135
},
{
"epoch": 1.765391791044776,
"grad_norm": 0.3870639900750078,
"learning_rate": 3.961394727306133e-05,
"loss": 0.4619,
"step": 15140
},
{
"epoch": 1.7659748134328357,
"grad_norm": 0.4089934690121375,
"learning_rate": 3.9606635111078156e-05,
"loss": 0.4582,
"step": 15145
},
{
"epoch": 1.7665578358208955,
"grad_norm": 0.44166689072835347,
"learning_rate": 3.9599321148954325e-05,
"loss": 0.4706,
"step": 15150
},
{
"epoch": 1.7671408582089554,
"grad_norm": 0.4774148419314475,
"learning_rate": 3.959200538777738e-05,
"loss": 0.4628,
"step": 15155
},
{
"epoch": 1.767723880597015,
"grad_norm": 0.4079011504725099,
"learning_rate": 3.958468782863508e-05,
"loss": 0.4752,
"step": 15160
},
{
"epoch": 1.7683069029850746,
"grad_norm": 0.45201689672551193,
"learning_rate": 3.957736847261548e-05,
"loss": 0.5049,
"step": 15165
},
{
"epoch": 1.7688899253731343,
"grad_norm": 0.3802703247205141,
"learning_rate": 3.9570047320806916e-05,
"loss": 0.4596,
"step": 15170
},
{
"epoch": 1.769472947761194,
"grad_norm": 0.39404343060002167,
"learning_rate": 3.956272437429796e-05,
"loss": 0.4836,
"step": 15175
},
{
"epoch": 1.7700559701492538,
"grad_norm": 0.44542854788396297,
"learning_rate": 3.955539963417746e-05,
"loss": 0.4605,
"step": 15180
},
{
"epoch": 1.7706389925373134,
"grad_norm": 0.39794735367770373,
"learning_rate": 3.954807310153454e-05,
"loss": 0.4695,
"step": 15185
},
{
"epoch": 1.7712220149253732,
"grad_norm": 0.41241430426744424,
"learning_rate": 3.954074477745859e-05,
"loss": 0.4811,
"step": 15190
},
{
"epoch": 1.7718050373134329,
"grad_norm": 0.41950410364972074,
"learning_rate": 3.9533414663039246e-05,
"loss": 0.492,
"step": 15195
},
{
"epoch": 1.7723880597014925,
"grad_norm": 0.4198224492459816,
"learning_rate": 3.952608275936644e-05,
"loss": 0.5086,
"step": 15200
},
{
"epoch": 1.7729710820895521,
"grad_norm": 0.4022553273442121,
"learning_rate": 3.951874906753035e-05,
"loss": 0.4395,
"step": 15205
},
{
"epoch": 1.773554104477612,
"grad_norm": 0.4300154004601707,
"learning_rate": 3.9511413588621435e-05,
"loss": 0.4696,
"step": 15210
},
{
"epoch": 1.7741371268656716,
"grad_norm": 0.39241273429664947,
"learning_rate": 3.9504076323730396e-05,
"loss": 0.4657,
"step": 15215
},
{
"epoch": 1.7747201492537314,
"grad_norm": 0.3888676516483212,
"learning_rate": 3.949673727394823e-05,
"loss": 0.4709,
"step": 15220
},
{
"epoch": 1.775303171641791,
"grad_norm": 0.40229431096323487,
"learning_rate": 3.948939644036616e-05,
"loss": 0.4924,
"step": 15225
},
{
"epoch": 1.7758861940298507,
"grad_norm": 0.6464933825933378,
"learning_rate": 3.9482053824075716e-05,
"loss": 0.4674,
"step": 15230
},
{
"epoch": 1.7764692164179103,
"grad_norm": 0.4043491830389959,
"learning_rate": 3.947470942616868e-05,
"loss": 0.4623,
"step": 15235
},
{
"epoch": 1.7770522388059702,
"grad_norm": 0.38156501215349453,
"learning_rate": 3.946736324773707e-05,
"loss": 0.5024,
"step": 15240
},
{
"epoch": 1.7776352611940298,
"grad_norm": 0.39245586227753537,
"learning_rate": 3.946001528987322e-05,
"loss": 0.4861,
"step": 15245
},
{
"epoch": 1.7782182835820897,
"grad_norm": 0.44125935346496437,
"learning_rate": 3.945266555366968e-05,
"loss": 0.4624,
"step": 15250
},
{
"epoch": 1.7788013059701493,
"grad_norm": 0.37553696918559865,
"learning_rate": 3.94453140402193e-05,
"loss": 0.4735,
"step": 15255
},
{
"epoch": 1.779384328358209,
"grad_norm": 0.425390219319215,
"learning_rate": 3.943796075061517e-05,
"loss": 0.4895,
"step": 15260
},
{
"epoch": 1.7799673507462686,
"grad_norm": 0.416099722707022,
"learning_rate": 3.943060568595065e-05,
"loss": 0.482,
"step": 15265
},
{
"epoch": 1.7805503731343284,
"grad_norm": 0.4170100272677797,
"learning_rate": 3.942324884731938e-05,
"loss": 0.4697,
"step": 15270
},
{
"epoch": 1.781133395522388,
"grad_norm": 0.36539660037054905,
"learning_rate": 3.941589023581524e-05,
"loss": 0.4403,
"step": 15275
},
{
"epoch": 1.7817164179104479,
"grad_norm": 0.3867176150081152,
"learning_rate": 3.940852985253239e-05,
"loss": 0.4376,
"step": 15280
},
{
"epoch": 1.7822994402985075,
"grad_norm": 0.45738766430041133,
"learning_rate": 3.940116769856526e-05,
"loss": 0.493,
"step": 15285
},
{
"epoch": 1.7828824626865671,
"grad_norm": 0.4251208871596025,
"learning_rate": 3.9393803775008506e-05,
"loss": 0.4974,
"step": 15290
},
{
"epoch": 1.7834654850746268,
"grad_norm": 0.40352873209568857,
"learning_rate": 3.9386438082957096e-05,
"loss": 0.4754,
"step": 15295
},
{
"epoch": 1.7840485074626866,
"grad_norm": 0.3986071012356413,
"learning_rate": 3.937907062350622e-05,
"loss": 0.487,
"step": 15300
},
{
"epoch": 1.7846315298507462,
"grad_norm": 0.41332502961987394,
"learning_rate": 3.937170139775137e-05,
"loss": 0.4915,
"step": 15305
},
{
"epoch": 1.785214552238806,
"grad_norm": 0.43103245611688024,
"learning_rate": 3.9364330406788265e-05,
"loss": 0.4801,
"step": 15310
},
{
"epoch": 1.7857975746268657,
"grad_norm": 0.41205735119186915,
"learning_rate": 3.9356957651712894e-05,
"loss": 0.4816,
"step": 15315
},
{
"epoch": 1.7863805970149254,
"grad_norm": 0.48840694762325626,
"learning_rate": 3.9349583133621535e-05,
"loss": 0.4694,
"step": 15320
},
{
"epoch": 1.786963619402985,
"grad_norm": 0.4131915852321686,
"learning_rate": 3.934220685361069e-05,
"loss": 0.4827,
"step": 15325
},
{
"epoch": 1.7875466417910446,
"grad_norm": 0.3894300207628937,
"learning_rate": 3.933482881277715e-05,
"loss": 0.4892,
"step": 15330
},
{
"epoch": 1.7881296641791045,
"grad_norm": 0.38210279764096605,
"learning_rate": 3.9327449012217955e-05,
"loss": 0.4549,
"step": 15335
},
{
"epoch": 1.7887126865671643,
"grad_norm": 0.40537470988711033,
"learning_rate": 3.9320067453030415e-05,
"loss": 0.485,
"step": 15340
},
{
"epoch": 1.789295708955224,
"grad_norm": 0.4089454829387791,
"learning_rate": 3.9312684136312094e-05,
"loss": 0.4737,
"step": 15345
},
{
"epoch": 1.7898787313432836,
"grad_norm": 0.4372054009396863,
"learning_rate": 3.930529906316083e-05,
"loss": 0.5003,
"step": 15350
},
{
"epoch": 1.7904617537313432,
"grad_norm": 0.40955980587827856,
"learning_rate": 3.9297912234674694e-05,
"loss": 0.4944,
"step": 15355
},
{
"epoch": 1.7910447761194028,
"grad_norm": 0.400700542691212,
"learning_rate": 3.9290523651952046e-05,
"loss": 0.4647,
"step": 15360
},
{
"epoch": 1.7916277985074627,
"grad_norm": 0.40549421369688765,
"learning_rate": 3.92831333160915e-05,
"loss": 0.4676,
"step": 15365
},
{
"epoch": 1.7922108208955225,
"grad_norm": 0.45638556477073006,
"learning_rate": 3.927574122819193e-05,
"loss": 0.4765,
"step": 15370
},
{
"epoch": 1.7927938432835822,
"grad_norm": 0.4246020398214672,
"learning_rate": 3.9268347389352464e-05,
"loss": 0.4908,
"step": 15375
},
{
"epoch": 1.7933768656716418,
"grad_norm": 0.3833487818350352,
"learning_rate": 3.926095180067249e-05,
"loss": 0.4488,
"step": 15380
},
{
"epoch": 1.7939598880597014,
"grad_norm": 0.4093574048152263,
"learning_rate": 3.925355446325167e-05,
"loss": 0.5032,
"step": 15385
},
{
"epoch": 1.794542910447761,
"grad_norm": 0.3986170325896095,
"learning_rate": 3.924615537818992e-05,
"loss": 0.4791,
"step": 15390
},
{
"epoch": 1.795125932835821,
"grad_norm": 0.420527866036737,
"learning_rate": 3.92387545465874e-05,
"loss": 0.4917,
"step": 15395
},
{
"epoch": 1.7957089552238807,
"grad_norm": 0.45521502374903666,
"learning_rate": 3.923135196954456e-05,
"loss": 0.5045,
"step": 15400
},
{
"epoch": 1.7962919776119404,
"grad_norm": 0.41107859823130743,
"learning_rate": 3.922394764816208e-05,
"loss": 0.4777,
"step": 15405
},
{
"epoch": 1.796875,
"grad_norm": 0.40650658242433396,
"learning_rate": 3.92165415835409e-05,
"loss": 0.4683,
"step": 15410
},
{
"epoch": 1.7974580223880596,
"grad_norm": 0.3849555799439404,
"learning_rate": 3.920913377678226e-05,
"loss": 0.4823,
"step": 15415
},
{
"epoch": 1.7980410447761193,
"grad_norm": 0.418578702384259,
"learning_rate": 3.92017242289876e-05,
"loss": 0.4686,
"step": 15420
},
{
"epoch": 1.798624067164179,
"grad_norm": 0.43812039694586474,
"learning_rate": 3.919431294125868e-05,
"loss": 0.4732,
"step": 15425
},
{
"epoch": 1.799207089552239,
"grad_norm": 0.41127898907608035,
"learning_rate": 3.918689991469746e-05,
"loss": 0.4592,
"step": 15430
},
{
"epoch": 1.7997901119402986,
"grad_norm": 0.44816753133483084,
"learning_rate": 3.91794851504062e-05,
"loss": 0.475,
"step": 15435
},
{
"epoch": 1.8003731343283582,
"grad_norm": 0.40643338863374023,
"learning_rate": 3.9172068649487405e-05,
"loss": 0.4557,
"step": 15440
},
{
"epoch": 1.8009561567164178,
"grad_norm": 0.37782727725264037,
"learning_rate": 3.916465041304383e-05,
"loss": 0.478,
"step": 15445
},
{
"epoch": 1.8015391791044775,
"grad_norm": 0.44244631303917853,
"learning_rate": 3.91572304421785e-05,
"loss": 0.487,
"step": 15450
},
{
"epoch": 1.8021222014925373,
"grad_norm": 0.4084939226282715,
"learning_rate": 3.9149808737994705e-05,
"loss": 0.4687,
"step": 15455
},
{
"epoch": 1.8027052238805972,
"grad_norm": 0.4341262271358096,
"learning_rate": 3.914238530159595e-05,
"loss": 0.5069,
"step": 15460
},
{
"epoch": 1.8032882462686568,
"grad_norm": 0.4667319157767231,
"learning_rate": 3.9134960134086055e-05,
"loss": 0.4913,
"step": 15465
},
{
"epoch": 1.8038712686567164,
"grad_norm": 0.38542046105302263,
"learning_rate": 3.9127533236569077e-05,
"loss": 0.479,
"step": 15470
},
{
"epoch": 1.804454291044776,
"grad_norm": 0.3983104149375585,
"learning_rate": 3.91201046101493e-05,
"loss": 0.4763,
"step": 15475
},
{
"epoch": 1.8050373134328357,
"grad_norm": 0.4557044991237112,
"learning_rate": 3.9112674255931294e-05,
"loss": 0.4729,
"step": 15480
},
{
"epoch": 1.8056203358208955,
"grad_norm": 0.398605111774296,
"learning_rate": 3.9105242175019905e-05,
"loss": 0.4576,
"step": 15485
},
{
"epoch": 1.8062033582089554,
"grad_norm": 0.4020143944218762,
"learning_rate": 3.909780836852019e-05,
"loss": 0.493,
"step": 15490
},
{
"epoch": 1.806786380597015,
"grad_norm": 0.4258847032896706,
"learning_rate": 3.9090372837537496e-05,
"loss": 0.4779,
"step": 15495
},
{
"epoch": 1.8073694029850746,
"grad_norm": 0.3788449876872275,
"learning_rate": 3.908293558317741e-05,
"loss": 0.462,
"step": 15500
},
{
"epoch": 1.8079524253731343,
"grad_norm": 0.41534740906913603,
"learning_rate": 3.907549660654577e-05,
"loss": 0.476,
"step": 15505
},
{
"epoch": 1.808535447761194,
"grad_norm": 0.4332899675796817,
"learning_rate": 3.9068055908748706e-05,
"loss": 0.5103,
"step": 15510
},
{
"epoch": 1.8091184701492538,
"grad_norm": 0.4030914140670705,
"learning_rate": 3.9060613490892556e-05,
"loss": 0.4584,
"step": 15515
},
{
"epoch": 1.8097014925373134,
"grad_norm": 0.39597160258735686,
"learning_rate": 3.9053169354083946e-05,
"loss": 0.4421,
"step": 15520
},
{
"epoch": 1.8102845149253732,
"grad_norm": 0.398649547369236,
"learning_rate": 3.904572349942974e-05,
"loss": 0.4864,
"step": 15525
},
{
"epoch": 1.8108675373134329,
"grad_norm": 0.43211937478635676,
"learning_rate": 3.903827592803708e-05,
"loss": 0.4928,
"step": 15530
},
{
"epoch": 1.8114505597014925,
"grad_norm": 0.39353992590112763,
"learning_rate": 3.903082664101334e-05,
"loss": 0.4405,
"step": 15535
},
{
"epoch": 1.8120335820895521,
"grad_norm": 0.41038084236870254,
"learning_rate": 3.9023375639466156e-05,
"loss": 0.4444,
"step": 15540
},
{
"epoch": 1.812616604477612,
"grad_norm": 0.40088923901412543,
"learning_rate": 3.901592292450342e-05,
"loss": 0.4708,
"step": 15545
},
{
"epoch": 1.8131996268656716,
"grad_norm": 0.3747100514690436,
"learning_rate": 3.900846849723328e-05,
"loss": 0.4713,
"step": 15550
},
{
"epoch": 1.8137826492537314,
"grad_norm": 0.4339908537179978,
"learning_rate": 3.9001012358764146e-05,
"loss": 0.4815,
"step": 15555
},
{
"epoch": 1.814365671641791,
"grad_norm": 0.38610085982358383,
"learning_rate": 3.8993554510204664e-05,
"loss": 0.4761,
"step": 15560
},
{
"epoch": 1.8149486940298507,
"grad_norm": 0.42773321572752476,
"learning_rate": 3.898609495266375e-05,
"loss": 0.4818,
"step": 15565
},
{
"epoch": 1.8155317164179103,
"grad_norm": 0.45083045474627925,
"learning_rate": 3.897863368725056e-05,
"loss": 0.5221,
"step": 15570
},
{
"epoch": 1.8161147388059702,
"grad_norm": 0.36013209274777985,
"learning_rate": 3.8971170715074526e-05,
"loss": 0.4544,
"step": 15575
},
{
"epoch": 1.8166977611940298,
"grad_norm": 0.41184520107378214,
"learning_rate": 3.896370603724531e-05,
"loss": 0.4854,
"step": 15580
},
{
"epoch": 1.8172807835820897,
"grad_norm": 0.3885048648718672,
"learning_rate": 3.895623965487284e-05,
"loss": 0.4589,
"step": 15585
},
{
"epoch": 1.8178638059701493,
"grad_norm": 0.3986924147173458,
"learning_rate": 3.8948771569067305e-05,
"loss": 0.4751,
"step": 15590
},
{
"epoch": 1.818446828358209,
"grad_norm": 0.40984748035131313,
"learning_rate": 3.8941301780939124e-05,
"loss": 0.474,
"step": 15595
},
{
"epoch": 1.8190298507462686,
"grad_norm": 0.366818795293633,
"learning_rate": 3.893383029159899e-05,
"loss": 0.4709,
"step": 15600
},
{
"epoch": 1.8196128731343284,
"grad_norm": 0.4036551530085086,
"learning_rate": 3.892635710215785e-05,
"loss": 0.5008,
"step": 15605
},
{
"epoch": 1.820195895522388,
"grad_norm": 0.42129360937557636,
"learning_rate": 3.891888221372688e-05,
"loss": 0.4685,
"step": 15610
},
{
"epoch": 1.8207789179104479,
"grad_norm": 0.4444262082361587,
"learning_rate": 3.891140562741753e-05,
"loss": 0.4679,
"step": 15615
},
{
"epoch": 1.8213619402985075,
"grad_norm": 0.413109811823458,
"learning_rate": 3.89039273443415e-05,
"loss": 0.4626,
"step": 15620
},
{
"epoch": 1.8219449626865671,
"grad_norm": 0.37574586024072143,
"learning_rate": 3.889644736561073e-05,
"loss": 0.4729,
"step": 15625
},
{
"epoch": 1.8225279850746268,
"grad_norm": 0.3900772734777929,
"learning_rate": 3.888896569233744e-05,
"loss": 0.4716,
"step": 15630
},
{
"epoch": 1.8231110074626866,
"grad_norm": 0.4268307621995299,
"learning_rate": 3.888148232563407e-05,
"loss": 0.4949,
"step": 15635
},
{
"epoch": 1.8236940298507462,
"grad_norm": 0.4358830677717346,
"learning_rate": 3.887399726661332e-05,
"loss": 0.4726,
"step": 15640
},
{
"epoch": 1.824277052238806,
"grad_norm": 0.39883007170076384,
"learning_rate": 3.886651051638815e-05,
"loss": 0.4685,
"step": 15645
},
{
"epoch": 1.8248600746268657,
"grad_norm": 0.40396056332788266,
"learning_rate": 3.885902207607178e-05,
"loss": 0.4735,
"step": 15650
},
{
"epoch": 1.8254430970149254,
"grad_norm": 0.4167157619548304,
"learning_rate": 3.885153194677766e-05,
"loss": 0.4938,
"step": 15655
},
{
"epoch": 1.826026119402985,
"grad_norm": 0.4203187960069392,
"learning_rate": 3.88440401296195e-05,
"loss": 0.4929,
"step": 15660
},
{
"epoch": 1.8266091417910446,
"grad_norm": 0.4341211867568005,
"learning_rate": 3.8836546625711254e-05,
"loss": 0.5087,
"step": 15665
},
{
"epoch": 1.8271921641791045,
"grad_norm": 0.39075385186204453,
"learning_rate": 3.8829051436167144e-05,
"loss": 0.4711,
"step": 15670
},
{
"epoch": 1.8277751865671643,
"grad_norm": 0.41422888614841374,
"learning_rate": 3.882155456210164e-05,
"loss": 0.4642,
"step": 15675
},
{
"epoch": 1.828358208955224,
"grad_norm": 0.37785499219074814,
"learning_rate": 3.881405600462943e-05,
"loss": 0.4893,
"step": 15680
},
{
"epoch": 1.8289412313432836,
"grad_norm": 0.3927210965136199,
"learning_rate": 3.880655576486551e-05,
"loss": 0.445,
"step": 15685
},
{
"epoch": 1.8295242537313432,
"grad_norm": 0.41703084100536003,
"learning_rate": 3.879905384392508e-05,
"loss": 0.5177,
"step": 15690
},
{
"epoch": 1.8301072761194028,
"grad_norm": 0.41137230623778376,
"learning_rate": 3.8791550242923594e-05,
"loss": 0.4823,
"step": 15695
},
{
"epoch": 1.8306902985074627,
"grad_norm": 0.38682073394946265,
"learning_rate": 3.8784044962976776e-05,
"loss": 0.4707,
"step": 15700
},
{
"epoch": 1.8312733208955225,
"grad_norm": 0.40629038056312605,
"learning_rate": 3.877653800520058e-05,
"loss": 0.4682,
"step": 15705
},
{
"epoch": 1.8318563432835822,
"grad_norm": 0.41598586100196094,
"learning_rate": 3.8769029370711234e-05,
"loss": 0.5037,
"step": 15710
},
{
"epoch": 1.8324393656716418,
"grad_norm": 0.46420430672556257,
"learning_rate": 3.876151906062519e-05,
"loss": 0.4625,
"step": 15715
},
{
"epoch": 1.8330223880597014,
"grad_norm": 0.46369679722699775,
"learning_rate": 3.8754007076059155e-05,
"loss": 0.5175,
"step": 15720
},
{
"epoch": 1.833605410447761,
"grad_norm": 0.4181383714240629,
"learning_rate": 3.8746493418130096e-05,
"loss": 0.4867,
"step": 15725
},
{
"epoch": 1.834188432835821,
"grad_norm": 0.40288133499025486,
"learning_rate": 3.873897808795522e-05,
"loss": 0.4839,
"step": 15730
},
{
"epoch": 1.8347714552238807,
"grad_norm": 0.4205814700827323,
"learning_rate": 3.8731461086652e-05,
"loss": 0.5,
"step": 15735
},
{
"epoch": 1.8353544776119404,
"grad_norm": 0.411766350596247,
"learning_rate": 3.8723942415338105e-05,
"loss": 0.4873,
"step": 15740
},
{
"epoch": 1.8359375,
"grad_norm": 0.38163342055425953,
"learning_rate": 3.871642207513152e-05,
"loss": 0.451,
"step": 15745
},
{
"epoch": 1.8365205223880596,
"grad_norm": 0.39414867687151395,
"learning_rate": 3.870890006715044e-05,
"loss": 0.4919,
"step": 15750
},
{
"epoch": 1.8371035447761193,
"grad_norm": 0.4551187782615539,
"learning_rate": 3.870137639251331e-05,
"loss": 0.4682,
"step": 15755
},
{
"epoch": 1.837686567164179,
"grad_norm": 0.4591681715825297,
"learning_rate": 3.869385105233884e-05,
"loss": 0.4877,
"step": 15760
},
{
"epoch": 1.838269589552239,
"grad_norm": 0.4118018694693055,
"learning_rate": 3.868632404774597e-05,
"loss": 0.4968,
"step": 15765
},
{
"epoch": 1.8388526119402986,
"grad_norm": 0.509013827239908,
"learning_rate": 3.867879537985388e-05,
"loss": 0.4833,
"step": 15770
},
{
"epoch": 1.8394356343283582,
"grad_norm": 0.48060406980698844,
"learning_rate": 3.867126504978204e-05,
"loss": 0.4879,
"step": 15775
},
{
"epoch": 1.8400186567164178,
"grad_norm": 0.3938739727043748,
"learning_rate": 3.8663733058650104e-05,
"loss": 0.5192,
"step": 15780
},
{
"epoch": 1.8406016791044775,
"grad_norm": 0.38171962022765954,
"learning_rate": 3.865619940757804e-05,
"loss": 0.4585,
"step": 15785
},
{
"epoch": 1.8411847014925373,
"grad_norm": 0.3789930067999332,
"learning_rate": 3.8648664097686e-05,
"loss": 0.4511,
"step": 15790
},
{
"epoch": 1.8417677238805972,
"grad_norm": 0.3702657242133424,
"learning_rate": 3.8641127130094434e-05,
"loss": 0.4756,
"step": 15795
},
{
"epoch": 1.8423507462686568,
"grad_norm": 0.4039677147752383,
"learning_rate": 3.8633588505924e-05,
"loss": 0.5078,
"step": 15800
},
{
"epoch": 1.8429337686567164,
"grad_norm": 0.44270006231799147,
"learning_rate": 3.862604822629564e-05,
"loss": 0.4749,
"step": 15805
},
{
"epoch": 1.843516791044776,
"grad_norm": 0.4071226582925008,
"learning_rate": 3.861850629233051e-05,
"loss": 0.4973,
"step": 15810
},
{
"epoch": 1.8440998134328357,
"grad_norm": 0.3844192653576694,
"learning_rate": 3.861096270515001e-05,
"loss": 0.4487,
"step": 15815
},
{
"epoch": 1.8446828358208955,
"grad_norm": 0.46835696367368285,
"learning_rate": 3.8603417465875816e-05,
"loss": 0.5028,
"step": 15820
},
{
"epoch": 1.8452658582089554,
"grad_norm": 0.417841442575684,
"learning_rate": 3.8595870575629834e-05,
"loss": 0.4793,
"step": 15825
},
{
"epoch": 1.845848880597015,
"grad_norm": 0.4381855887470024,
"learning_rate": 3.858832203553421e-05,
"loss": 0.5293,
"step": 15830
},
{
"epoch": 1.8464319029850746,
"grad_norm": 0.3992981993529362,
"learning_rate": 3.8580771846711335e-05,
"loss": 0.4555,
"step": 15835
},
{
"epoch": 1.8470149253731343,
"grad_norm": 0.4040824685345217,
"learning_rate": 3.857322001028385e-05,
"loss": 0.4593,
"step": 15840
},
{
"epoch": 1.847597947761194,
"grad_norm": 0.38293322273225205,
"learning_rate": 3.856566652737465e-05,
"loss": 0.5019,
"step": 15845
},
{
"epoch": 1.8481809701492538,
"grad_norm": 0.4585816035516511,
"learning_rate": 3.855811139910686e-05,
"loss": 0.4875,
"step": 15850
},
{
"epoch": 1.8487639925373134,
"grad_norm": 0.41162663830745716,
"learning_rate": 3.855055462660385e-05,
"loss": 0.4919,
"step": 15855
},
{
"epoch": 1.8493470149253732,
"grad_norm": 0.39170798846721,
"learning_rate": 3.854299621098925e-05,
"loss": 0.4544,
"step": 15860
},
{
"epoch": 1.8499300373134329,
"grad_norm": 0.37926089062207197,
"learning_rate": 3.853543615338691e-05,
"loss": 0.4406,
"step": 15865
},
{
"epoch": 1.8505130597014925,
"grad_norm": 0.37298974771470295,
"learning_rate": 3.8527874454920955e-05,
"loss": 0.4886,
"step": 15870
},
{
"epoch": 1.8510960820895521,
"grad_norm": 0.3870604348064698,
"learning_rate": 3.852031111671573e-05,
"loss": 0.459,
"step": 15875
},
{
"epoch": 1.851679104477612,
"grad_norm": 0.44136326104268875,
"learning_rate": 3.851274613989582e-05,
"loss": 0.5028,
"step": 15880
},
{
"epoch": 1.8522621268656716,
"grad_norm": 0.3667399614732067,
"learning_rate": 3.850517952558608e-05,
"loss": 0.4462,
"step": 15885
},
{
"epoch": 1.8528451492537314,
"grad_norm": 0.40808488522894626,
"learning_rate": 3.849761127491158e-05,
"loss": 0.4807,
"step": 15890
},
{
"epoch": 1.853428171641791,
"grad_norm": 0.38135841075395843,
"learning_rate": 3.849004138899766e-05,
"loss": 0.4928,
"step": 15895
},
{
"epoch": 1.8540111940298507,
"grad_norm": 0.38045258820885103,
"learning_rate": 3.848246986896989e-05,
"loss": 0.481,
"step": 15900
},
{
"epoch": 1.8545942164179103,
"grad_norm": 0.40655612248620926,
"learning_rate": 3.847489671595406e-05,
"loss": 0.4395,
"step": 15905
},
{
"epoch": 1.8551772388059702,
"grad_norm": 0.3829921354976652,
"learning_rate": 3.8467321931076255e-05,
"loss": 0.4578,
"step": 15910
},
{
"epoch": 1.8557602611940298,
"grad_norm": 0.39279724721647,
"learning_rate": 3.845974551546276e-05,
"loss": 0.5077,
"step": 15915
},
{
"epoch": 1.8563432835820897,
"grad_norm": 0.38717973434618574,
"learning_rate": 3.84521674702401e-05,
"loss": 0.5018,
"step": 15920
},
{
"epoch": 1.8569263059701493,
"grad_norm": 0.3753542167148014,
"learning_rate": 3.844458779653508e-05,
"loss": 0.4613,
"step": 15925
},
{
"epoch": 1.857509328358209,
"grad_norm": 0.36248882376059743,
"learning_rate": 3.8437006495474716e-05,
"loss": 0.4514,
"step": 15930
},
{
"epoch": 1.8580923507462686,
"grad_norm": 0.4130159463676241,
"learning_rate": 3.842942356818627e-05,
"loss": 0.5111,
"step": 15935
},
{
"epoch": 1.8586753731343284,
"grad_norm": 0.4022933677414021,
"learning_rate": 3.8421839015797265e-05,
"loss": 0.4664,
"step": 15940
},
{
"epoch": 1.859258395522388,
"grad_norm": 0.45940652562024537,
"learning_rate": 3.841425283943544e-05,
"loss": 0.5038,
"step": 15945
},
{
"epoch": 1.8598414179104479,
"grad_norm": 0.4338299573989669,
"learning_rate": 3.840666504022879e-05,
"loss": 0.469,
"step": 15950
},
{
"epoch": 1.8604244402985075,
"grad_norm": 0.38921153278825393,
"learning_rate": 3.839907561930554e-05,
"loss": 0.4724,
"step": 15955
},
{
"epoch": 1.8610074626865671,
"grad_norm": 0.4120809967947985,
"learning_rate": 3.839148457779418e-05,
"loss": 0.5102,
"step": 15960
},
{
"epoch": 1.8615904850746268,
"grad_norm": 0.43073826253995934,
"learning_rate": 3.838389191682341e-05,
"loss": 0.4868,
"step": 15965
},
{
"epoch": 1.8621735074626866,
"grad_norm": 0.37518589498092525,
"learning_rate": 3.837629763752219e-05,
"loss": 0.4519,
"step": 15970
},
{
"epoch": 1.8627565298507462,
"grad_norm": 0.42065130984639537,
"learning_rate": 3.8368701741019734e-05,
"loss": 0.5179,
"step": 15975
},
{
"epoch": 1.863339552238806,
"grad_norm": 0.44034209990802575,
"learning_rate": 3.8361104228445455e-05,
"loss": 0.5088,
"step": 15980
},
{
"epoch": 1.8639225746268657,
"grad_norm": 0.428726844646854,
"learning_rate": 3.835350510092904e-05,
"loss": 0.5116,
"step": 15985
},
{
"epoch": 1.8645055970149254,
"grad_norm": 0.41290111731726753,
"learning_rate": 3.834590435960041e-05,
"loss": 0.4702,
"step": 15990
},
{
"epoch": 1.865088619402985,
"grad_norm": 0.38778917040761307,
"learning_rate": 3.833830200558971e-05,
"loss": 0.4761,
"step": 15995
},
{
"epoch": 1.8656716417910446,
"grad_norm": 0.44238489509858847,
"learning_rate": 3.8330698040027345e-05,
"loss": 0.5411,
"step": 16000
},
{
"epoch": 1.8662546641791045,
"grad_norm": 0.42042102256172875,
"learning_rate": 3.832309246404396e-05,
"loss": 0.5205,
"step": 16005
},
{
"epoch": 1.8668376865671643,
"grad_norm": 0.42171279923682037,
"learning_rate": 3.8315485278770423e-05,
"loss": 0.4779,
"step": 16010
},
{
"epoch": 1.867420708955224,
"grad_norm": 0.4325489253113547,
"learning_rate": 3.830787648533785e-05,
"loss": 0.5003,
"step": 16015
},
{
"epoch": 1.8680037313432836,
"grad_norm": 0.44449707743834876,
"learning_rate": 3.83002660848776e-05,
"loss": 0.4702,
"step": 16020
},
{
"epoch": 1.8685867537313432,
"grad_norm": 0.4057244191845612,
"learning_rate": 3.829265407852125e-05,
"loss": 0.4906,
"step": 16025
},
{
"epoch": 1.8691697761194028,
"grad_norm": 0.4047231713285812,
"learning_rate": 3.828504046740065e-05,
"loss": 0.4838,
"step": 16030
},
{
"epoch": 1.8697527985074627,
"grad_norm": 0.4402545345321185,
"learning_rate": 3.827742525264787e-05,
"loss": 0.506,
"step": 16035
},
{
"epoch": 1.8703358208955225,
"grad_norm": 0.4318181827689146,
"learning_rate": 3.826980843539521e-05,
"loss": 0.4912,
"step": 16040
},
{
"epoch": 1.8709188432835822,
"grad_norm": 0.39281509103457857,
"learning_rate": 3.826219001677523e-05,
"loss": 0.4507,
"step": 16045
},
{
"epoch": 1.8715018656716418,
"grad_norm": 0.3790746086135691,
"learning_rate": 3.82545699979207e-05,
"loss": 0.4501,
"step": 16050
},
{
"epoch": 1.8720848880597014,
"grad_norm": 0.4044025067278778,
"learning_rate": 3.824694837996466e-05,
"loss": 0.4661,
"step": 16055
},
{
"epoch": 1.872667910447761,
"grad_norm": 0.42653108142887636,
"learning_rate": 3.823932516404036e-05,
"loss": 0.4689,
"step": 16060
},
{
"epoch": 1.873250932835821,
"grad_norm": 0.38397790509604734,
"learning_rate": 3.823170035128131e-05,
"loss": 0.4521,
"step": 16065
},
{
"epoch": 1.8738339552238807,
"grad_norm": 0.4002935434985125,
"learning_rate": 3.822407394282123e-05,
"loss": 0.4773,
"step": 16070
},
{
"epoch": 1.8744169776119404,
"grad_norm": 0.41580866271495587,
"learning_rate": 3.821644593979411e-05,
"loss": 0.5112,
"step": 16075
},
{
"epoch": 1.875,
"grad_norm": 0.38766810421634024,
"learning_rate": 3.8208816343334156e-05,
"loss": 0.4724,
"step": 16080
},
{
"epoch": 1.8755830223880596,
"grad_norm": 0.40069384581867035,
"learning_rate": 3.820118515457582e-05,
"loss": 0.4846,
"step": 16085
},
{
"epoch": 1.8761660447761193,
"grad_norm": 0.3903370227525892,
"learning_rate": 3.819355237465377e-05,
"loss": 0.4481,
"step": 16090
},
{
"epoch": 1.876749067164179,
"grad_norm": 0.4096276880014734,
"learning_rate": 3.818591800470295e-05,
"loss": 0.5202,
"step": 16095
},
{
"epoch": 1.877332089552239,
"grad_norm": 0.39952781068074394,
"learning_rate": 3.81782820458585e-05,
"loss": 0.4386,
"step": 16100
},
{
"epoch": 1.8779151119402986,
"grad_norm": 0.4743106559132605,
"learning_rate": 3.817064449925582e-05,
"loss": 0.4819,
"step": 16105
},
{
"epoch": 1.8784981343283582,
"grad_norm": 0.4121861252761789,
"learning_rate": 3.816300536603054e-05,
"loss": 0.4789,
"step": 16110
},
{
"epoch": 1.8790811567164178,
"grad_norm": 0.41347188150712333,
"learning_rate": 3.815536464731853e-05,
"loss": 0.4799,
"step": 16115
},
{
"epoch": 1.8796641791044775,
"grad_norm": 0.4027894743276831,
"learning_rate": 3.814772234425588e-05,
"loss": 0.4731,
"step": 16120
},
{
"epoch": 1.8802472014925373,
"grad_norm": 0.40856626217364383,
"learning_rate": 3.814007845797894e-05,
"loss": 0.4692,
"step": 16125
},
{
"epoch": 1.8808302238805972,
"grad_norm": 0.4391793128455858,
"learning_rate": 3.813243298962428e-05,
"loss": 0.4712,
"step": 16130
},
{
"epoch": 1.8814132462686568,
"grad_norm": 0.4023196419084873,
"learning_rate": 3.8124785940328704e-05,
"loss": 0.4837,
"step": 16135
},
{
"epoch": 1.8819962686567164,
"grad_norm": 0.4051129806594245,
"learning_rate": 3.8117137311229255e-05,
"loss": 0.4606,
"step": 16140
},
{
"epoch": 1.882579291044776,
"grad_norm": 0.384969224693189,
"learning_rate": 3.810948710346322e-05,
"loss": 0.4887,
"step": 16145
},
{
"epoch": 1.8831623134328357,
"grad_norm": 0.39416545187061863,
"learning_rate": 3.81018353181681e-05,
"loss": 0.4448,
"step": 16150
},
{
"epoch": 1.8837453358208955,
"grad_norm": 0.39071190107945314,
"learning_rate": 3.8094181956481645e-05,
"loss": 0.4937,
"step": 16155
},
{
"epoch": 1.8843283582089554,
"grad_norm": 0.40976718895408326,
"learning_rate": 3.808652701954183e-05,
"loss": 0.4899,
"step": 16160
},
{
"epoch": 1.884911380597015,
"grad_norm": 0.39986071039677495,
"learning_rate": 3.807887050848689e-05,
"loss": 0.4747,
"step": 16165
},
{
"epoch": 1.8854944029850746,
"grad_norm": 0.397199438043576,
"learning_rate": 3.807121242445526e-05,
"loss": 0.4768,
"step": 16170
},
{
"epoch": 1.8860774253731343,
"grad_norm": 0.37446958271136305,
"learning_rate": 3.806355276858562e-05,
"loss": 0.4702,
"step": 16175
},
{
"epoch": 1.886660447761194,
"grad_norm": 0.42441869032333956,
"learning_rate": 3.805589154201691e-05,
"loss": 0.462,
"step": 16180
},
{
"epoch": 1.8872434701492538,
"grad_norm": 0.41373456326126706,
"learning_rate": 3.804822874588825e-05,
"loss": 0.4603,
"step": 16185
},
{
"epoch": 1.8878264925373134,
"grad_norm": 0.4063084742024033,
"learning_rate": 3.804056438133905e-05,
"loss": 0.4649,
"step": 16190
},
{
"epoch": 1.8884095149253732,
"grad_norm": 0.42477797034907416,
"learning_rate": 3.803289844950893e-05,
"loss": 0.4836,
"step": 16195
},
{
"epoch": 1.8889925373134329,
"grad_norm": 0.3874744527316154,
"learning_rate": 3.80252309515377e-05,
"loss": 0.4607,
"step": 16200
},
{
"epoch": 1.8895755597014925,
"grad_norm": 0.3893900370916626,
"learning_rate": 3.801756188856549e-05,
"loss": 0.4657,
"step": 16205
},
{
"epoch": 1.8901585820895521,
"grad_norm": 0.3892493641847063,
"learning_rate": 3.800989126173259e-05,
"loss": 0.472,
"step": 16210
},
{
"epoch": 1.890741604477612,
"grad_norm": 0.3738549215638216,
"learning_rate": 3.800221907217956e-05,
"loss": 0.4704,
"step": 16215
},
{
"epoch": 1.8913246268656716,
"grad_norm": 0.408209912848128,
"learning_rate": 3.799454532104718e-05,
"loss": 0.473,
"step": 16220
},
{
"epoch": 1.8919076492537314,
"grad_norm": 0.42433965709059185,
"learning_rate": 3.7986870009476454e-05,
"loss": 0.4623,
"step": 16225
},
{
"epoch": 1.892490671641791,
"grad_norm": 0.3914089243364931,
"learning_rate": 3.7979193138608646e-05,
"loss": 0.4924,
"step": 16230
},
{
"epoch": 1.8930736940298507,
"grad_norm": 0.485375920959211,
"learning_rate": 3.797151470958521e-05,
"loss": 0.5154,
"step": 16235
},
{
"epoch": 1.8936567164179103,
"grad_norm": 0.38590075567283577,
"learning_rate": 3.7963834723547866e-05,
"loss": 0.4696,
"step": 16240
},
{
"epoch": 1.8942397388059702,
"grad_norm": 0.410772058988447,
"learning_rate": 3.795615318163856e-05,
"loss": 0.4443,
"step": 16245
},
{
"epoch": 1.8948227611940298,
"grad_norm": 0.38317535209122194,
"learning_rate": 3.794847008499946e-05,
"loss": 0.4993,
"step": 16250
},
{
"epoch": 1.8954057835820897,
"grad_norm": 0.4064916270947185,
"learning_rate": 3.7940785434772965e-05,
"loss": 0.4767,
"step": 16255
},
{
"epoch": 1.8959888059701493,
"grad_norm": 0.40411293880729643,
"learning_rate": 3.793309923210171e-05,
"loss": 0.4625,
"step": 16260
},
{
"epoch": 1.896571828358209,
"grad_norm": 0.37456650735591096,
"learning_rate": 3.792541147812856e-05,
"loss": 0.5228,
"step": 16265
},
{
"epoch": 1.8971548507462686,
"grad_norm": 0.4190371199656764,
"learning_rate": 3.791772217399661e-05,
"loss": 0.4706,
"step": 16270
},
{
"epoch": 1.8977378731343284,
"grad_norm": 0.3922319515428183,
"learning_rate": 3.791003132084919e-05,
"loss": 0.4767,
"step": 16275
},
{
"epoch": 1.898320895522388,
"grad_norm": 0.5470149003980315,
"learning_rate": 3.7902338919829854e-05,
"loss": 0.4731,
"step": 16280
},
{
"epoch": 1.8989039179104479,
"grad_norm": 0.40976174662007087,
"learning_rate": 3.789464497208238e-05,
"loss": 0.4642,
"step": 16285
},
{
"epoch": 1.8994869402985075,
"grad_norm": 0.43326296131089675,
"learning_rate": 3.788694947875079e-05,
"loss": 0.5053,
"step": 16290
},
{
"epoch": 1.9000699626865671,
"grad_norm": 0.3930508352423981,
"learning_rate": 3.787925244097935e-05,
"loss": 0.4555,
"step": 16295
},
{
"epoch": 1.9006529850746268,
"grad_norm": 0.3689438893532244,
"learning_rate": 3.78715538599125e-05,
"loss": 0.468,
"step": 16300
},
{
"epoch": 1.9012360074626866,
"grad_norm": 0.43438128865830056,
"learning_rate": 3.786385373669497e-05,
"loss": 0.4738,
"step": 16305
},
{
"epoch": 1.9018190298507462,
"grad_norm": 0.41853652061756247,
"learning_rate": 3.7856152072471686e-05,
"loss": 0.4671,
"step": 16310
},
{
"epoch": 1.902402052238806,
"grad_norm": 0.3945411702472311,
"learning_rate": 3.7848448868387806e-05,
"loss": 0.4479,
"step": 16315
},
{
"epoch": 1.9029850746268657,
"grad_norm": 0.38768017734875904,
"learning_rate": 3.784074412558875e-05,
"loss": 0.4574,
"step": 16320
},
{
"epoch": 1.9035680970149254,
"grad_norm": 0.39455269546284544,
"learning_rate": 3.7833037845220097e-05,
"loss": 0.4786,
"step": 16325
},
{
"epoch": 1.904151119402985,
"grad_norm": 0.4239773973420398,
"learning_rate": 3.782533002842773e-05,
"loss": 0.528,
"step": 16330
},
{
"epoch": 1.9047341417910446,
"grad_norm": 0.42244613712490187,
"learning_rate": 3.781762067635771e-05,
"loss": 0.4987,
"step": 16335
},
{
"epoch": 1.9053171641791045,
"grad_norm": 0.44852303417013406,
"learning_rate": 3.7809909790156355e-05,
"loss": 0.4746,
"step": 16340
},
{
"epoch": 1.9059001865671643,
"grad_norm": 0.4393086527481142,
"learning_rate": 3.78021973709702e-05,
"loss": 0.4542,
"step": 16345
},
{
"epoch": 1.906483208955224,
"grad_norm": 0.402224723983359,
"learning_rate": 3.7794483419946e-05,
"loss": 0.4717,
"step": 16350
},
{
"epoch": 1.9070662313432836,
"grad_norm": 0.4243706638845106,
"learning_rate": 3.778676793823075e-05,
"loss": 0.4572,
"step": 16355
},
{
"epoch": 1.9076492537313432,
"grad_norm": 0.41599027838302116,
"learning_rate": 3.777905092697166e-05,
"loss": 0.4829,
"step": 16360
},
{
"epoch": 1.9082322761194028,
"grad_norm": 0.4017264788022485,
"learning_rate": 3.7771332387316186e-05,
"loss": 0.4711,
"step": 16365
},
{
"epoch": 1.9088152985074627,
"grad_norm": 0.4550572468239916,
"learning_rate": 3.7763612320412e-05,
"loss": 0.4829,
"step": 16370
},
{
"epoch": 1.9093983208955225,
"grad_norm": 0.41334212997226194,
"learning_rate": 3.7755890727406994e-05,
"loss": 0.455,
"step": 16375
},
{
"epoch": 1.9099813432835822,
"grad_norm": 0.4022630693327699,
"learning_rate": 3.77481676094493e-05,
"loss": 0.4953,
"step": 16380
},
{
"epoch": 1.9105643656716418,
"grad_norm": 0.4484057334472698,
"learning_rate": 3.774044296768728e-05,
"loss": 0.4769,
"step": 16385
},
{
"epoch": 1.9111473880597014,
"grad_norm": 0.40114234694902606,
"learning_rate": 3.77327168032695e-05,
"loss": 0.4843,
"step": 16390
},
{
"epoch": 1.911730410447761,
"grad_norm": 0.42756820757515207,
"learning_rate": 3.772498911734478e-05,
"loss": 0.465,
"step": 16395
},
{
"epoch": 1.912313432835821,
"grad_norm": 0.4081033664707095,
"learning_rate": 3.771725991106214e-05,
"loss": 0.4989,
"step": 16400
},
{
"epoch": 1.9128964552238807,
"grad_norm": 0.39990963402946317,
"learning_rate": 3.7709529185570846e-05,
"loss": 0.4549,
"step": 16405
},
{
"epoch": 1.9134794776119404,
"grad_norm": 0.39415693532083024,
"learning_rate": 3.770179694202038e-05,
"loss": 0.5289,
"step": 16410
},
{
"epoch": 1.9140625,
"grad_norm": 0.36870474335573705,
"learning_rate": 3.7694063181560456e-05,
"loss": 0.4529,
"step": 16415
},
{
"epoch": 1.9146455223880596,
"grad_norm": 0.3811879050835146,
"learning_rate": 3.7686327905341014e-05,
"loss": 0.4886,
"step": 16420
},
{
"epoch": 1.9152285447761193,
"grad_norm": 0.42225081839415346,
"learning_rate": 3.76785911145122e-05,
"loss": 0.4569,
"step": 16425
},
{
"epoch": 1.915811567164179,
"grad_norm": 0.3893868986907824,
"learning_rate": 3.767085281022441e-05,
"loss": 0.5049,
"step": 16430
},
{
"epoch": 1.916394589552239,
"grad_norm": 0.3857552525470896,
"learning_rate": 3.766311299362826e-05,
"loss": 0.485,
"step": 16435
},
{
"epoch": 1.9169776119402986,
"grad_norm": 0.41590242594626253,
"learning_rate": 3.765537166587458e-05,
"loss": 0.4761,
"step": 16440
},
{
"epoch": 1.9175606343283582,
"grad_norm": 0.41966464258161296,
"learning_rate": 3.764762882811444e-05,
"loss": 0.4939,
"step": 16445
},
{
"epoch": 1.9181436567164178,
"grad_norm": 0.37378791935038896,
"learning_rate": 3.763988448149912e-05,
"loss": 0.4749,
"step": 16450
},
{
"epoch": 1.9187266791044775,
"grad_norm": 0.3623872216384651,
"learning_rate": 3.763213862718012e-05,
"loss": 0.4621,
"step": 16455
},
{
"epoch": 1.9193097014925373,
"grad_norm": 0.4088015820211004,
"learning_rate": 3.762439126630919e-05,
"loss": 0.4975,
"step": 16460
},
{
"epoch": 1.9198927238805972,
"grad_norm": 0.41167934288178915,
"learning_rate": 3.761664240003828e-05,
"loss": 0.483,
"step": 16465
},
{
"epoch": 1.9204757462686568,
"grad_norm": 0.3566403186749648,
"learning_rate": 3.7608892029519576e-05,
"loss": 0.4467,
"step": 16470
},
{
"epoch": 1.9210587686567164,
"grad_norm": 0.3960928211013188,
"learning_rate": 3.7601140155905485e-05,
"loss": 0.4905,
"step": 16475
},
{
"epoch": 1.921641791044776,
"grad_norm": 0.42244900696929966,
"learning_rate": 3.7593386780348625e-05,
"loss": 0.4826,
"step": 16480
},
{
"epoch": 1.9222248134328357,
"grad_norm": 0.41264798541535225,
"learning_rate": 3.758563190400187e-05,
"loss": 0.4481,
"step": 16485
},
{
"epoch": 1.9228078358208955,
"grad_norm": 0.4784051652523608,
"learning_rate": 3.757787552801827e-05,
"loss": 0.5125,
"step": 16490
},
{
"epoch": 1.9233908582089554,
"grad_norm": 0.40478268729486794,
"learning_rate": 3.757011765355115e-05,
"loss": 0.4877,
"step": 16495
},
{
"epoch": 1.923973880597015,
"grad_norm": 0.43157962225933183,
"learning_rate": 3.756235828175401e-05,
"loss": 0.4886,
"step": 16500
},
{
"epoch": 1.9245569029850746,
"grad_norm": 0.3870430698134395,
"learning_rate": 3.755459741378061e-05,
"loss": 0.4539,
"step": 16505
},
{
"epoch": 1.9251399253731343,
"grad_norm": 0.5381877988217303,
"learning_rate": 3.75468350507849e-05,
"loss": 0.5197,
"step": 16510
},
{
"epoch": 1.925722947761194,
"grad_norm": 0.42249259646697085,
"learning_rate": 3.753907119392108e-05,
"loss": 0.5015,
"step": 16515
},
{
"epoch": 1.9263059701492538,
"grad_norm": 0.4390311414973388,
"learning_rate": 3.753130584434357e-05,
"loss": 0.4683,
"step": 16520
},
{
"epoch": 1.9268889925373134,
"grad_norm": 0.4090149980450418,
"learning_rate": 3.752353900320698e-05,
"loss": 0.452,
"step": 16525
},
{
"epoch": 1.9274720149253732,
"grad_norm": 0.4061074236998217,
"learning_rate": 3.7515770671666175e-05,
"loss": 0.4854,
"step": 16530
},
{
"epoch": 1.9280550373134329,
"grad_norm": 0.4290232571924102,
"learning_rate": 3.750800085087625e-05,
"loss": 0.5024,
"step": 16535
},
{
"epoch": 1.9286380597014925,
"grad_norm": 0.5160988567004748,
"learning_rate": 3.750022954199248e-05,
"loss": 0.5046,
"step": 16540
},
{
"epoch": 1.9292210820895521,
"grad_norm": 0.38032340028583833,
"learning_rate": 3.7492456746170385e-05,
"loss": 0.4712,
"step": 16545
},
{
"epoch": 1.929804104477612,
"grad_norm": 0.41077435748539576,
"learning_rate": 3.748468246456572e-05,
"loss": 0.4761,
"step": 16550
},
{
"epoch": 1.9303871268656716,
"grad_norm": 0.4700124737492129,
"learning_rate": 3.747690669833443e-05,
"loss": 0.4658,
"step": 16555
},
{
"epoch": 1.9309701492537314,
"grad_norm": 0.46311118729692075,
"learning_rate": 3.7469129448632704e-05,
"loss": 0.4817,
"step": 16560
},
{
"epoch": 1.931553171641791,
"grad_norm": 0.42113606356414157,
"learning_rate": 3.7461350716616955e-05,
"loss": 0.4866,
"step": 16565
},
{
"epoch": 1.9321361940298507,
"grad_norm": 0.4025147398337016,
"learning_rate": 3.7453570503443785e-05,
"loss": 0.4595,
"step": 16570
},
{
"epoch": 1.9327192164179103,
"grad_norm": 0.4229540767535665,
"learning_rate": 3.744578881027006e-05,
"loss": 0.4806,
"step": 16575
},
{
"epoch": 1.9333022388059702,
"grad_norm": 0.45409889851507895,
"learning_rate": 3.743800563825283e-05,
"loss": 0.5039,
"step": 16580
},
{
"epoch": 1.9338852611940298,
"grad_norm": 0.38530571510428296,
"learning_rate": 3.743022098854937e-05,
"loss": 0.4766,
"step": 16585
},
{
"epoch": 1.9344682835820897,
"grad_norm": 0.46896570244232355,
"learning_rate": 3.742243486231719e-05,
"loss": 0.4815,
"step": 16590
},
{
"epoch": 1.9350513059701493,
"grad_norm": 0.43267921963119627,
"learning_rate": 3.7414647260714025e-05,
"loss": 0.4887,
"step": 16595
},
{
"epoch": 1.935634328358209,
"grad_norm": 0.7028618230877963,
"learning_rate": 3.74068581848978e-05,
"loss": 0.4806,
"step": 16600
},
{
"epoch": 1.9362173507462686,
"grad_norm": 0.3861540918854558,
"learning_rate": 3.739906763602669e-05,
"loss": 0.4705,
"step": 16605
},
{
"epoch": 1.9368003731343284,
"grad_norm": 2.3870802765535304,
"learning_rate": 3.7391275615259065e-05,
"loss": 0.4961,
"step": 16610
},
{
"epoch": 1.937383395522388,
"grad_norm": 0.3844269511493217,
"learning_rate": 3.738348212375352e-05,
"loss": 0.4621,
"step": 16615
},
{
"epoch": 1.9379664179104479,
"grad_norm": 0.4041414263644477,
"learning_rate": 3.737568716266888e-05,
"loss": 0.4719,
"step": 16620
},
{
"epoch": 1.9385494402985075,
"grad_norm": 0.37973447998386956,
"learning_rate": 3.7367890733164185e-05,
"loss": 0.4713,
"step": 16625
},
{
"epoch": 1.9391324626865671,
"grad_norm": 0.39106083913968837,
"learning_rate": 3.7360092836398686e-05,
"loss": 0.4714,
"step": 16630
},
{
"epoch": 1.9397154850746268,
"grad_norm": 0.39565639330982066,
"learning_rate": 3.7352293473531844e-05,
"loss": 0.4771,
"step": 16635
},
{
"epoch": 1.9402985074626866,
"grad_norm": 0.3578259873475878,
"learning_rate": 3.734449264572336e-05,
"loss": 0.459,
"step": 16640
},
{
"epoch": 1.9408815298507462,
"grad_norm": 0.3787654282979704,
"learning_rate": 3.733669035413315e-05,
"loss": 0.4745,
"step": 16645
},
{
"epoch": 1.941464552238806,
"grad_norm": 0.39718938432333034,
"learning_rate": 3.7328886599921327e-05,
"loss": 0.466,
"step": 16650
},
{
"epoch": 1.9420475746268657,
"grad_norm": 0.4102061499601817,
"learning_rate": 3.732108138424824e-05,
"loss": 0.5032,
"step": 16655
},
{
"epoch": 1.9426305970149254,
"grad_norm": 0.3774949062632342,
"learning_rate": 3.7313274708274445e-05,
"loss": 0.4674,
"step": 16660
},
{
"epoch": 1.943213619402985,
"grad_norm": 0.3984575933091318,
"learning_rate": 3.7305466573160725e-05,
"loss": 0.4679,
"step": 16665
},
{
"epoch": 1.9437966417910446,
"grad_norm": 0.40535943928527784,
"learning_rate": 3.729765698006808e-05,
"loss": 0.4664,
"step": 16670
},
{
"epoch": 1.9443796641791045,
"grad_norm": 0.5454599589638158,
"learning_rate": 3.7289845930157704e-05,
"loss": 0.4414,
"step": 16675
},
{
"epoch": 1.9449626865671643,
"grad_norm": 0.3859731888383419,
"learning_rate": 3.7282033424591043e-05,
"loss": 0.4554,
"step": 16680
},
{
"epoch": 1.945545708955224,
"grad_norm": 0.3685656495645585,
"learning_rate": 3.7274219464529736e-05,
"loss": 0.4528,
"step": 16685
},
{
"epoch": 1.9461287313432836,
"grad_norm": 0.4588171481071328,
"learning_rate": 3.726640405113564e-05,
"loss": 0.4844,
"step": 16690
},
{
"epoch": 1.9467117537313432,
"grad_norm": 0.36823038271837566,
"learning_rate": 3.725858718557084e-05,
"loss": 0.4454,
"step": 16695
},
{
"epoch": 1.9472947761194028,
"grad_norm": 0.40010987425600264,
"learning_rate": 3.725076886899763e-05,
"loss": 0.4504,
"step": 16700
},
{
"epoch": 1.9478777985074627,
"grad_norm": 0.38973950050762535,
"learning_rate": 3.72429491025785e-05,
"loss": 0.5048,
"step": 16705
},
{
"epoch": 1.9484608208955225,
"grad_norm": 0.4310358335759325,
"learning_rate": 3.723512788747619e-05,
"loss": 0.475,
"step": 16710
},
{
"epoch": 1.9490438432835822,
"grad_norm": 0.46712116479614596,
"learning_rate": 3.722730522485364e-05,
"loss": 0.541,
"step": 16715
},
{
"epoch": 1.9496268656716418,
"grad_norm": 0.4316970096467175,
"learning_rate": 3.721948111587399e-05,
"loss": 0.5075,
"step": 16720
},
{
"epoch": 1.9502098880597014,
"grad_norm": 0.4205955396297808,
"learning_rate": 3.721165556170065e-05,
"loss": 0.4548,
"step": 16725
},
{
"epoch": 1.950792910447761,
"grad_norm": 0.4203871211897958,
"learning_rate": 3.720382856349715e-05,
"loss": 0.4675,
"step": 16730
},
{
"epoch": 1.951375932835821,
"grad_norm": 0.41475197902523725,
"learning_rate": 3.719600012242733e-05,
"loss": 0.4433,
"step": 16735
},
{
"epoch": 1.9519589552238807,
"grad_norm": 0.4081830980890671,
"learning_rate": 3.718817023965519e-05,
"loss": 0.463,
"step": 16740
},
{
"epoch": 1.9525419776119404,
"grad_norm": 0.4211230101753503,
"learning_rate": 3.718033891634496e-05,
"loss": 0.4878,
"step": 16745
},
{
"epoch": 1.953125,
"grad_norm": 0.42193399504459156,
"learning_rate": 3.717250615366108e-05,
"loss": 0.5,
"step": 16750
},
{
"epoch": 1.9537080223880596,
"grad_norm": 0.4219503411049521,
"learning_rate": 3.7164671952768206e-05,
"loss": 0.4733,
"step": 16755
},
{
"epoch": 1.9542910447761193,
"grad_norm": 0.406463599638427,
"learning_rate": 3.715683631483121e-05,
"loss": 0.4603,
"step": 16760
},
{
"epoch": 1.954874067164179,
"grad_norm": 0.3855559528822891,
"learning_rate": 3.7148999241015185e-05,
"loss": 0.4534,
"step": 16765
},
{
"epoch": 1.955457089552239,
"grad_norm": 0.37082440935305794,
"learning_rate": 3.714116073248542e-05,
"loss": 0.4427,
"step": 16770
},
{
"epoch": 1.9560401119402986,
"grad_norm": 0.3607069535459937,
"learning_rate": 3.713332079040743e-05,
"loss": 0.4756,
"step": 16775
},
{
"epoch": 1.9566231343283582,
"grad_norm": 0.4334772889695182,
"learning_rate": 3.712547941594693e-05,
"loss": 0.4924,
"step": 16780
},
{
"epoch": 1.9572061567164178,
"grad_norm": 0.3856978224354025,
"learning_rate": 3.7117636610269875e-05,
"loss": 0.4632,
"step": 16785
},
{
"epoch": 1.9577891791044775,
"grad_norm": 0.4645421829101539,
"learning_rate": 3.71097923745424e-05,
"loss": 0.5006,
"step": 16790
},
{
"epoch": 1.9583722014925373,
"grad_norm": 0.3868172011914484,
"learning_rate": 3.710194670993087e-05,
"loss": 0.4852,
"step": 16795
},
{
"epoch": 1.9589552238805972,
"grad_norm": 0.42532981954584903,
"learning_rate": 3.709409961760186e-05,
"loss": 0.4869,
"step": 16800
},
{
"epoch": 1.9595382462686568,
"grad_norm": 0.39036454374697166,
"learning_rate": 3.708625109872217e-05,
"loss": 0.4682,
"step": 16805
},
{
"epoch": 1.9601212686567164,
"grad_norm": 0.9115506210889813,
"learning_rate": 3.707840115445877e-05,
"loss": 0.4796,
"step": 16810
},
{
"epoch": 1.960704291044776,
"grad_norm": 0.3994822012039142,
"learning_rate": 3.707054978597891e-05,
"loss": 0.4547,
"step": 16815
},
{
"epoch": 1.9612873134328357,
"grad_norm": 0.3920868940709815,
"learning_rate": 3.706269699444998e-05,
"loss": 0.4774,
"step": 16820
},
{
"epoch": 1.9618703358208955,
"grad_norm": 0.45976672383860007,
"learning_rate": 3.705484278103964e-05,
"loss": 0.482,
"step": 16825
},
{
"epoch": 1.9624533582089554,
"grad_norm": 0.40408221217538515,
"learning_rate": 3.704698714691572e-05,
"loss": 0.4559,
"step": 16830
},
{
"epoch": 1.963036380597015,
"grad_norm": 0.38867974780837394,
"learning_rate": 3.703913009324628e-05,
"loss": 0.4788,
"step": 16835
},
{
"epoch": 1.9636194029850746,
"grad_norm": 0.4074948137795385,
"learning_rate": 3.703127162119959e-05,
"loss": 0.4924,
"step": 16840
},
{
"epoch": 1.9642024253731343,
"grad_norm": 0.3532292695939483,
"learning_rate": 3.702341173194413e-05,
"loss": 0.4759,
"step": 16845
},
{
"epoch": 1.964785447761194,
"grad_norm": 0.3673161523930121,
"learning_rate": 3.701555042664861e-05,
"loss": 0.4246,
"step": 16850
},
{
"epoch": 1.9653684701492538,
"grad_norm": 0.3983167128787513,
"learning_rate": 3.7007687706481896e-05,
"loss": 0.4742,
"step": 16855
},
{
"epoch": 1.9659514925373134,
"grad_norm": 0.44069995874263695,
"learning_rate": 3.699982357261312e-05,
"loss": 0.4747,
"step": 16860
},
{
"epoch": 1.9665345149253732,
"grad_norm": 0.4179449484696286,
"learning_rate": 3.699195802621159e-05,
"loss": 0.4814,
"step": 16865
},
{
"epoch": 1.9671175373134329,
"grad_norm": 0.5313306303372765,
"learning_rate": 3.6984091068446855e-05,
"loss": 0.4724,
"step": 16870
},
{
"epoch": 1.9677005597014925,
"grad_norm": 0.4041824205186734,
"learning_rate": 3.697622270048864e-05,
"loss": 0.4871,
"step": 16875
},
{
"epoch": 1.9682835820895521,
"grad_norm": 0.38779126275640496,
"learning_rate": 3.69683529235069e-05,
"loss": 0.5014,
"step": 16880
},
{
"epoch": 1.968866604477612,
"grad_norm": 0.4087047082682217,
"learning_rate": 3.6960481738671806e-05,
"loss": 0.4835,
"step": 16885
},
{
"epoch": 1.9694496268656716,
"grad_norm": 0.4257340759685614,
"learning_rate": 3.695260914715372e-05,
"loss": 0.4789,
"step": 16890
},
{
"epoch": 1.9700326492537314,
"grad_norm": 0.3888475404211362,
"learning_rate": 3.6944735150123215e-05,
"loss": 0.4573,
"step": 16895
},
{
"epoch": 1.970615671641791,
"grad_norm": 0.43264952576993276,
"learning_rate": 3.693685974875109e-05,
"loss": 0.5198,
"step": 16900
},
{
"epoch": 1.9711986940298507,
"grad_norm": 0.375427913943805,
"learning_rate": 3.6928982944208344e-05,
"loss": 0.4761,
"step": 16905
},
{
"epoch": 1.9717817164179103,
"grad_norm": 0.39791450964579966,
"learning_rate": 3.692110473766616e-05,
"loss": 0.5142,
"step": 16910
},
{
"epoch": 1.9723647388059702,
"grad_norm": 0.38141927095940475,
"learning_rate": 3.6913225130295974e-05,
"loss": 0.4504,
"step": 16915
},
{
"epoch": 1.9729477611940298,
"grad_norm": 0.4014015389936595,
"learning_rate": 3.69053441232694e-05,
"loss": 0.4564,
"step": 16920
},
{
"epoch": 1.9735307835820897,
"grad_norm": 0.3993117903736875,
"learning_rate": 3.689746171775827e-05,
"loss": 0.4577,
"step": 16925
},
{
"epoch": 1.9741138059701493,
"grad_norm": 0.40597494605440465,
"learning_rate": 3.688957791493462e-05,
"loss": 0.4999,
"step": 16930
},
{
"epoch": 1.974696828358209,
"grad_norm": 0.40406283019005323,
"learning_rate": 3.68816927159707e-05,
"loss": 0.4712,
"step": 16935
},
{
"epoch": 1.9752798507462686,
"grad_norm": 0.368111443636303,
"learning_rate": 3.6873806122038964e-05,
"loss": 0.4703,
"step": 16940
},
{
"epoch": 1.9758628731343284,
"grad_norm": 0.406242724610839,
"learning_rate": 3.686591813431206e-05,
"loss": 0.4891,
"step": 16945
},
{
"epoch": 1.976445895522388,
"grad_norm": 0.39794611594957685,
"learning_rate": 3.685802875396287e-05,
"loss": 0.5092,
"step": 16950
},
{
"epoch": 1.9770289179104479,
"grad_norm": 0.47672970484842125,
"learning_rate": 3.6850137982164466e-05,
"loss": 0.5087,
"step": 16955
},
{
"epoch": 1.9776119402985075,
"grad_norm": 0.38184874607776165,
"learning_rate": 3.684224582009014e-05,
"loss": 0.4717,
"step": 16960
},
{
"epoch": 1.9781949626865671,
"grad_norm": 0.4241920379079499,
"learning_rate": 3.683435226891335e-05,
"loss": 0.4584,
"step": 16965
},
{
"epoch": 1.9787779850746268,
"grad_norm": 0.3959445216740194,
"learning_rate": 3.682645732980783e-05,
"loss": 0.4829,
"step": 16970
},
{
"epoch": 1.9793610074626866,
"grad_norm": 0.39963720749487497,
"learning_rate": 3.681856100394745e-05,
"loss": 0.4618,
"step": 16975
},
{
"epoch": 1.9799440298507462,
"grad_norm": 0.42456237621721227,
"learning_rate": 3.6810663292506344e-05,
"loss": 0.4766,
"step": 16980
},
{
"epoch": 1.980527052238806,
"grad_norm": 0.41505505807337784,
"learning_rate": 3.68027641966588e-05,
"loss": 0.4658,
"step": 16985
},
{
"epoch": 1.9811100746268657,
"grad_norm": 0.4117451536711639,
"learning_rate": 3.6794863717579365e-05,
"loss": 0.4696,
"step": 16990
},
{
"epoch": 1.9816930970149254,
"grad_norm": 0.37960314538128104,
"learning_rate": 3.6786961856442737e-05,
"loss": 0.4542,
"step": 16995
},
{
"epoch": 1.982276119402985,
"grad_norm": 0.3902955640541868,
"learning_rate": 3.677905861442387e-05,
"loss": 0.4879,
"step": 17000
},
{
"epoch": 1.9828591417910446,
"grad_norm": 0.41164180183844934,
"learning_rate": 3.677115399269789e-05,
"loss": 0.4881,
"step": 17005
},
{
"epoch": 1.9834421641791045,
"grad_norm": 0.3998624672180739,
"learning_rate": 3.676324799244014e-05,
"loss": 0.4764,
"step": 17010
},
{
"epoch": 1.9840251865671643,
"grad_norm": 0.4166789335922044,
"learning_rate": 3.675534061482617e-05,
"loss": 0.4781,
"step": 17015
},
{
"epoch": 1.984608208955224,
"grad_norm": 0.44753527894552464,
"learning_rate": 3.6747431861031716e-05,
"loss": 0.48,
"step": 17020
},
{
"epoch": 1.9851912313432836,
"grad_norm": 0.4377000277323754,
"learning_rate": 3.6739521732232756e-05,
"loss": 0.4719,
"step": 17025
},
{
"epoch": 1.9857742537313432,
"grad_norm": 0.4556311746195216,
"learning_rate": 3.673161022960544e-05,
"loss": 0.4919,
"step": 17030
},
{
"epoch": 1.9863572761194028,
"grad_norm": 0.39199825786880405,
"learning_rate": 3.6723697354326134e-05,
"loss": 0.4587,
"step": 17035
},
{
"epoch": 1.9869402985074627,
"grad_norm": 0.3974044923023687,
"learning_rate": 3.67157831075714e-05,
"loss": 0.4645,
"step": 17040
},
{
"epoch": 1.9875233208955225,
"grad_norm": 0.3778225135522206,
"learning_rate": 3.670786749051801e-05,
"loss": 0.466,
"step": 17045
},
{
"epoch": 1.9881063432835822,
"grad_norm": 0.40936053779607,
"learning_rate": 3.6699950504342954e-05,
"loss": 0.4591,
"step": 17050
},
{
"epoch": 1.9886893656716418,
"grad_norm": 0.44904800916291815,
"learning_rate": 3.669203215022341e-05,
"loss": 0.4915,
"step": 17055
},
{
"epoch": 1.9892723880597014,
"grad_norm": 0.39843809526649276,
"learning_rate": 3.6684112429336745e-05,
"loss": 0.4855,
"step": 17060
},
{
"epoch": 1.989855410447761,
"grad_norm": 0.3620310448140616,
"learning_rate": 3.667619134286057e-05,
"loss": 0.4501,
"step": 17065
},
{
"epoch": 1.990438432835821,
"grad_norm": 0.444308166017696,
"learning_rate": 3.666826889197265e-05,
"loss": 0.472,
"step": 17070
},
{
"epoch": 1.9910214552238807,
"grad_norm": 0.37952757162829676,
"learning_rate": 3.666034507785098e-05,
"loss": 0.4652,
"step": 17075
},
{
"epoch": 1.9916044776119404,
"grad_norm": 0.4185276423874473,
"learning_rate": 3.665241990167378e-05,
"loss": 0.4616,
"step": 17080
},
{
"epoch": 1.9921875,
"grad_norm": 0.38571830135742774,
"learning_rate": 3.664449336461943e-05,
"loss": 0.4805,
"step": 17085
},
{
"epoch": 1.9927705223880596,
"grad_norm": 0.42090409971706727,
"learning_rate": 3.663656546786653e-05,
"loss": 0.4862,
"step": 17090
},
{
"epoch": 1.9933535447761193,
"grad_norm": 0.40892314384825473,
"learning_rate": 3.6628636212593874e-05,
"loss": 0.4472,
"step": 17095
},
{
"epoch": 1.993936567164179,
"grad_norm": 0.4050926400176725,
"learning_rate": 3.6620705599980494e-05,
"loss": 0.4617,
"step": 17100
},
{
"epoch": 1.994519589552239,
"grad_norm": 0.37504799105020076,
"learning_rate": 3.6612773631205567e-05,
"loss": 0.432,
"step": 17105
},
{
"epoch": 1.9951026119402986,
"grad_norm": 0.3890783052111123,
"learning_rate": 3.660484030744852e-05,
"loss": 0.4245,
"step": 17110
},
{
"epoch": 1.9956856343283582,
"grad_norm": 0.5019978499285157,
"learning_rate": 3.659690562988894e-05,
"loss": 0.4986,
"step": 17115
},
{
"epoch": 1.9962686567164178,
"grad_norm": 0.38062845774729975,
"learning_rate": 3.6588969599706665e-05,
"loss": 0.4555,
"step": 17120
},
{
"epoch": 1.9968516791044775,
"grad_norm": 0.3986977104412468,
"learning_rate": 3.6581032218081685e-05,
"loss": 0.4601,
"step": 17125
},
{
"epoch": 1.9974347014925373,
"grad_norm": 0.41108104833354003,
"learning_rate": 3.6573093486194226e-05,
"loss": 0.4832,
"step": 17130
},
{
"epoch": 1.9980177238805972,
"grad_norm": 0.39946669843141763,
"learning_rate": 3.65651534052247e-05,
"loss": 0.4761,
"step": 17135
},
{
"epoch": 1.9986007462686568,
"grad_norm": 0.617041064444864,
"learning_rate": 3.655721197635371e-05,
"loss": 0.4633,
"step": 17140
},
{
"epoch": 1.9991837686567164,
"grad_norm": 0.3566803157427374,
"learning_rate": 3.654926920076208e-05,
"loss": 0.4548,
"step": 17145
},
{
"epoch": 1.999766791044776,
"grad_norm": 0.39221319928845394,
"learning_rate": 3.654132507963083e-05,
"loss": 0.4758,
"step": 17150
},
{
"epoch": 2.0003498134328357,
"grad_norm": 0.35974425418608424,
"learning_rate": 3.653337961414116e-05,
"loss": 0.4572,
"step": 17155
},
{
"epoch": 2.0009328358208953,
"grad_norm": 0.41393158661494905,
"learning_rate": 3.652543280547449e-05,
"loss": 0.4038,
"step": 17160
},
{
"epoch": 2.0015158582089554,
"grad_norm": 0.4318684000835193,
"learning_rate": 3.651748465481245e-05,
"loss": 0.4066,
"step": 17165
},
{
"epoch": 2.002098880597015,
"grad_norm": 0.42383911903824045,
"learning_rate": 3.650953516333682e-05,
"loss": 0.4109,
"step": 17170
},
{
"epoch": 2.0026819029850746,
"grad_norm": 0.4168315446980631,
"learning_rate": 3.6501584332229645e-05,
"loss": 0.3801,
"step": 17175
},
{
"epoch": 2.0032649253731343,
"grad_norm": 0.44231291127034256,
"learning_rate": 3.6493632162673125e-05,
"loss": 0.4155,
"step": 17180
},
{
"epoch": 2.003847947761194,
"grad_norm": 0.4186092913713343,
"learning_rate": 3.648567865584967e-05,
"loss": 0.4237,
"step": 17185
},
{
"epoch": 2.0044309701492535,
"grad_norm": 0.3977411578691464,
"learning_rate": 3.647772381294189e-05,
"loss": 0.4077,
"step": 17190
},
{
"epoch": 2.0050139925373136,
"grad_norm": 0.4615182032330395,
"learning_rate": 3.64697676351326e-05,
"loss": 0.3831,
"step": 17195
},
{
"epoch": 2.0055970149253732,
"grad_norm": 0.4155941705113631,
"learning_rate": 3.6461810123604805e-05,
"loss": 0.4059,
"step": 17200
},
{
"epoch": 2.006180037313433,
"grad_norm": 0.4474256705323867,
"learning_rate": 3.645385127954171e-05,
"loss": 0.4716,
"step": 17205
},
{
"epoch": 2.0067630597014925,
"grad_norm": 0.4054934065107215,
"learning_rate": 3.6445891104126714e-05,
"loss": 0.4249,
"step": 17210
},
{
"epoch": 2.007346082089552,
"grad_norm": 0.3976611767707526,
"learning_rate": 3.643792959854342e-05,
"loss": 0.4094,
"step": 17215
},
{
"epoch": 2.0079291044776117,
"grad_norm": 0.44449371418132566,
"learning_rate": 3.6429966763975636e-05,
"loss": 0.4026,
"step": 17220
},
{
"epoch": 2.008512126865672,
"grad_norm": 0.43500057395824304,
"learning_rate": 3.642200260160735e-05,
"loss": 0.4202,
"step": 17225
},
{
"epoch": 2.0090951492537314,
"grad_norm": 0.4155992350702752,
"learning_rate": 3.641403711262277e-05,
"loss": 0.3987,
"step": 17230
},
{
"epoch": 2.009678171641791,
"grad_norm": 0.41312574634007154,
"learning_rate": 3.6406070298206265e-05,
"loss": 0.418,
"step": 17235
},
{
"epoch": 2.0102611940298507,
"grad_norm": 0.446437880517514,
"learning_rate": 3.639810215954245e-05,
"loss": 0.4117,
"step": 17240
},
{
"epoch": 2.0108442164179103,
"grad_norm": 0.38507786194850685,
"learning_rate": 3.639013269781609e-05,
"loss": 0.4016,
"step": 17245
},
{
"epoch": 2.01142723880597,
"grad_norm": 0.4094945005900366,
"learning_rate": 3.638216191421218e-05,
"loss": 0.3944,
"step": 17250
},
{
"epoch": 2.01201026119403,
"grad_norm": 0.4471885484091221,
"learning_rate": 3.637418980991589e-05,
"loss": 0.4132,
"step": 17255
},
{
"epoch": 2.0125932835820897,
"grad_norm": 0.4416582578431907,
"learning_rate": 3.6366216386112605e-05,
"loss": 0.4041,
"step": 17260
},
{
"epoch": 2.0131763059701493,
"grad_norm": 0.44700347809049584,
"learning_rate": 3.635824164398789e-05,
"loss": 0.414,
"step": 17265
},
{
"epoch": 2.013759328358209,
"grad_norm": 0.4085205455503069,
"learning_rate": 3.635026558472752e-05,
"loss": 0.4136,
"step": 17270
},
{
"epoch": 2.0143423507462686,
"grad_norm": 0.3860550897181453,
"learning_rate": 3.634228820951744e-05,
"loss": 0.3729,
"step": 17275
},
{
"epoch": 2.014925373134328,
"grad_norm": 0.413879237651807,
"learning_rate": 3.633430951954383e-05,
"loss": 0.4095,
"step": 17280
},
{
"epoch": 2.0155083955223883,
"grad_norm": 0.4380213833210687,
"learning_rate": 3.6326329515993055e-05,
"loss": 0.4137,
"step": 17285
},
{
"epoch": 2.016091417910448,
"grad_norm": 0.4050950469907184,
"learning_rate": 3.631834820005163e-05,
"loss": 0.3836,
"step": 17290
},
{
"epoch": 2.0166744402985075,
"grad_norm": 0.4484025706465973,
"learning_rate": 3.6310365572906314e-05,
"loss": 0.3881,
"step": 17295
},
{
"epoch": 2.017257462686567,
"grad_norm": 0.3826212858166743,
"learning_rate": 3.6302381635744056e-05,
"loss": 0.3814,
"step": 17300
},
{
"epoch": 2.0178404850746268,
"grad_norm": 0.43005810923375426,
"learning_rate": 3.629439638975199e-05,
"loss": 0.4003,
"step": 17305
},
{
"epoch": 2.0184235074626864,
"grad_norm": 0.410666288794066,
"learning_rate": 3.628640983611744e-05,
"loss": 0.408,
"step": 17310
},
{
"epoch": 2.0190065298507465,
"grad_norm": 0.4918541323710734,
"learning_rate": 3.6278421976027926e-05,
"loss": 0.4103,
"step": 17315
},
{
"epoch": 2.019589552238806,
"grad_norm": 0.390538141531701,
"learning_rate": 3.6270432810671176e-05,
"loss": 0.4314,
"step": 17320
},
{
"epoch": 2.0201725746268657,
"grad_norm": 0.4457795006339435,
"learning_rate": 3.6262442341235105e-05,
"loss": 0.4374,
"step": 17325
},
{
"epoch": 2.0207555970149254,
"grad_norm": 0.380497746108089,
"learning_rate": 3.62544505689078e-05,
"loss": 0.3921,
"step": 17330
},
{
"epoch": 2.021338619402985,
"grad_norm": 5.372828015954768,
"learning_rate": 3.6246457494877585e-05,
"loss": 0.3873,
"step": 17335
},
{
"epoch": 2.0219216417910446,
"grad_norm": 0.45641737776863933,
"learning_rate": 3.623846312033294e-05,
"loss": 0.4123,
"step": 17340
},
{
"epoch": 2.0225046641791047,
"grad_norm": 0.43416818803809887,
"learning_rate": 3.623046744646254e-05,
"loss": 0.424,
"step": 17345
},
{
"epoch": 2.0230876865671643,
"grad_norm": 0.4204419218483119,
"learning_rate": 3.622247047445529e-05,
"loss": 0.3923,
"step": 17350
},
{
"epoch": 2.023670708955224,
"grad_norm": 0.49827384849949685,
"learning_rate": 3.6214472205500256e-05,
"loss": 0.4653,
"step": 17355
},
{
"epoch": 2.0242537313432836,
"grad_norm": 0.430083259667377,
"learning_rate": 3.6206472640786696e-05,
"loss": 0.4285,
"step": 17360
},
{
"epoch": 2.024836753731343,
"grad_norm": 0.4167994726809484,
"learning_rate": 3.6198471781504076e-05,
"loss": 0.399,
"step": 17365
},
{
"epoch": 2.025419776119403,
"grad_norm": 0.3992188423609367,
"learning_rate": 3.619046962884204e-05,
"loss": 0.4007,
"step": 17370
},
{
"epoch": 2.026002798507463,
"grad_norm": 0.40863350972882084,
"learning_rate": 3.618246618399043e-05,
"loss": 0.3853,
"step": 17375
},
{
"epoch": 2.0265858208955225,
"grad_norm": 0.4439383038448917,
"learning_rate": 3.617446144813929e-05,
"loss": 0.4105,
"step": 17380
},
{
"epoch": 2.027168843283582,
"grad_norm": 0.4482173126534646,
"learning_rate": 3.616645542247885e-05,
"loss": 0.389,
"step": 17385
},
{
"epoch": 2.027751865671642,
"grad_norm": 0.40762063689592337,
"learning_rate": 3.6158448108199515e-05,
"loss": 0.4125,
"step": 17390
},
{
"epoch": 2.0283348880597014,
"grad_norm": 0.41610468738070305,
"learning_rate": 3.6150439506491915e-05,
"loss": 0.3968,
"step": 17395
},
{
"epoch": 2.028917910447761,
"grad_norm": 0.4490666758996821,
"learning_rate": 3.614242961854683e-05,
"loss": 0.3972,
"step": 17400
},
{
"epoch": 2.029500932835821,
"grad_norm": 0.707190655829145,
"learning_rate": 3.6134418445555275e-05,
"loss": 0.4201,
"step": 17405
},
{
"epoch": 2.0300839552238807,
"grad_norm": 0.4761853979925262,
"learning_rate": 3.6126405988708424e-05,
"loss": 0.4067,
"step": 17410
},
{
"epoch": 2.0306669776119404,
"grad_norm": 0.4126466981845711,
"learning_rate": 3.611839224919765e-05,
"loss": 0.4175,
"step": 17415
},
{
"epoch": 2.03125,
"grad_norm": 0.4036158369871603,
"learning_rate": 3.611037722821452e-05,
"loss": 0.3865,
"step": 17420
},
{
"epoch": 2.0318330223880596,
"grad_norm": 0.4277080064913996,
"learning_rate": 3.6102360926950796e-05,
"loss": 0.3837,
"step": 17425
},
{
"epoch": 2.0324160447761193,
"grad_norm": 0.4345897946626314,
"learning_rate": 3.609434334659842e-05,
"loss": 0.4133,
"step": 17430
},
{
"epoch": 2.0329990671641793,
"grad_norm": 0.5322124187508095,
"learning_rate": 3.608632448834954e-05,
"loss": 0.4251,
"step": 17435
},
{
"epoch": 2.033582089552239,
"grad_norm": 0.4320842291475738,
"learning_rate": 3.607830435339648e-05,
"loss": 0.4183,
"step": 17440
},
{
"epoch": 2.0341651119402986,
"grad_norm": 0.37156559821800766,
"learning_rate": 3.6070282942931744e-05,
"loss": 0.4307,
"step": 17445
},
{
"epoch": 2.034748134328358,
"grad_norm": 0.5350659867774648,
"learning_rate": 3.606226025814805e-05,
"loss": 0.4295,
"step": 17450
},
{
"epoch": 2.035331156716418,
"grad_norm": 0.44133508577698394,
"learning_rate": 3.605423630023829e-05,
"loss": 0.4014,
"step": 17455
},
{
"epoch": 2.0359141791044775,
"grad_norm": 0.43885446890854674,
"learning_rate": 3.604621107039555e-05,
"loss": 0.4162,
"step": 17460
},
{
"epoch": 2.036497201492537,
"grad_norm": 0.48108449465160613,
"learning_rate": 3.603818456981312e-05,
"loss": 0.4243,
"step": 17465
},
{
"epoch": 2.037080223880597,
"grad_norm": 0.4169175488663904,
"learning_rate": 3.6030156799684435e-05,
"loss": 0.4115,
"step": 17470
},
{
"epoch": 2.037663246268657,
"grad_norm": 0.4833172600967252,
"learning_rate": 3.602212776120317e-05,
"loss": 0.4363,
"step": 17475
},
{
"epoch": 2.0382462686567164,
"grad_norm": 0.46381822008841517,
"learning_rate": 3.601409745556315e-05,
"loss": 0.4045,
"step": 17480
},
{
"epoch": 2.038829291044776,
"grad_norm": 0.43876403353906346,
"learning_rate": 3.6006065883958425e-05,
"loss": 0.3992,
"step": 17485
},
{
"epoch": 2.0394123134328357,
"grad_norm": 0.44697008478487776,
"learning_rate": 3.5998033047583194e-05,
"loss": 0.4048,
"step": 17490
},
{
"epoch": 2.0399953358208953,
"grad_norm": 0.37822668527686987,
"learning_rate": 3.598999894763187e-05,
"loss": 0.3751,
"step": 17495
},
{
"epoch": 2.0405783582089554,
"grad_norm": 0.4064353495153569,
"learning_rate": 3.598196358529906e-05,
"loss": 0.4138,
"step": 17500
},
{
"epoch": 2.041161380597015,
"grad_norm": 0.3799857869374489,
"learning_rate": 3.597392696177953e-05,
"loss": 0.3881,
"step": 17505
},
{
"epoch": 2.0417444029850746,
"grad_norm": 0.40655421341103043,
"learning_rate": 3.596588907826824e-05,
"loss": 0.4058,
"step": 17510
},
{
"epoch": 2.0423274253731343,
"grad_norm": 0.445390124206304,
"learning_rate": 3.595784993596036e-05,
"loss": 0.395,
"step": 17515
},
{
"epoch": 2.042910447761194,
"grad_norm": 0.41662013675369436,
"learning_rate": 3.5949809536051235e-05,
"loss": 0.4273,
"step": 17520
},
{
"epoch": 2.0434934701492535,
"grad_norm": 0.4259131767978619,
"learning_rate": 3.594176787973638e-05,
"loss": 0.4092,
"step": 17525
},
{
"epoch": 2.0440764925373136,
"grad_norm": 0.3834830846569681,
"learning_rate": 3.593372496821154e-05,
"loss": 0.404,
"step": 17530
},
{
"epoch": 2.0446595149253732,
"grad_norm": 0.3996890940345802,
"learning_rate": 3.5925680802672584e-05,
"loss": 0.4006,
"step": 17535
},
{
"epoch": 2.045242537313433,
"grad_norm": 0.413506389504434,
"learning_rate": 3.591763538431563e-05,
"loss": 0.3943,
"step": 17540
},
{
"epoch": 2.0458255597014925,
"grad_norm": 1.3631019268732503,
"learning_rate": 3.590958871433695e-05,
"loss": 0.3917,
"step": 17545
},
{
"epoch": 2.046408582089552,
"grad_norm": 0.4650214211421635,
"learning_rate": 3.5901540793933e-05,
"loss": 0.3964,
"step": 17550
},
{
"epoch": 2.0469916044776117,
"grad_norm": 0.4004480748536821,
"learning_rate": 3.5893491624300416e-05,
"loss": 0.4033,
"step": 17555
},
{
"epoch": 2.047574626865672,
"grad_norm": 0.427543704792774,
"learning_rate": 3.5885441206636065e-05,
"loss": 0.3948,
"step": 17560
},
{
"epoch": 2.0481576492537314,
"grad_norm": 0.44134543159574047,
"learning_rate": 3.587738954213694e-05,
"loss": 0.4285,
"step": 17565
},
{
"epoch": 2.048740671641791,
"grad_norm": 0.4966737984480068,
"learning_rate": 3.586933663200026e-05,
"loss": 0.4287,
"step": 17570
},
{
"epoch": 2.0493236940298507,
"grad_norm": 0.41118745931821765,
"learning_rate": 3.586128247742341e-05,
"loss": 0.4186,
"step": 17575
},
{
"epoch": 2.0499067164179103,
"grad_norm": 0.4024292533172103,
"learning_rate": 3.585322707960397e-05,
"loss": 0.3918,
"step": 17580
},
{
"epoch": 2.05048973880597,
"grad_norm": 0.4367564233329026,
"learning_rate": 3.584517043973969e-05,
"loss": 0.3929,
"step": 17585
},
{
"epoch": 2.05107276119403,
"grad_norm": 0.39290442314481433,
"learning_rate": 3.583711255902853e-05,
"loss": 0.3763,
"step": 17590
},
{
"epoch": 2.0516557835820897,
"grad_norm": 0.4096315820544621,
"learning_rate": 3.58290534386686e-05,
"loss": 0.4168,
"step": 17595
},
{
"epoch": 2.0522388059701493,
"grad_norm": 0.4237540152963559,
"learning_rate": 3.5820993079858235e-05,
"loss": 0.4077,
"step": 17600
},
{
"epoch": 2.052821828358209,
"grad_norm": 0.41721732565734926,
"learning_rate": 3.581293148379592e-05,
"loss": 0.4256,
"step": 17605
},
{
"epoch": 2.0534048507462686,
"grad_norm": 0.427197849133855,
"learning_rate": 3.580486865168034e-05,
"loss": 0.4099,
"step": 17610
},
{
"epoch": 2.053987873134328,
"grad_norm": 0.44396780922032564,
"learning_rate": 3.579680458471037e-05,
"loss": 0.4134,
"step": 17615
},
{
"epoch": 2.0545708955223883,
"grad_norm": 0.40085397386916805,
"learning_rate": 3.5788739284085044e-05,
"loss": 0.4077,
"step": 17620
},
{
"epoch": 2.055153917910448,
"grad_norm": 0.4183804859387029,
"learning_rate": 3.57806727510036e-05,
"loss": 0.3991,
"step": 17625
},
{
"epoch": 2.0557369402985075,
"grad_norm": 0.422113085026162,
"learning_rate": 3.577260498666546e-05,
"loss": 0.4177,
"step": 17630
},
{
"epoch": 2.056319962686567,
"grad_norm": 0.4010894845996832,
"learning_rate": 3.5764535992270226e-05,
"loss": 0.4245,
"step": 17635
},
{
"epoch": 2.0569029850746268,
"grad_norm": 0.4358002446247879,
"learning_rate": 3.575646576901767e-05,
"loss": 0.3888,
"step": 17640
},
{
"epoch": 2.0574860074626864,
"grad_norm": 0.417588300480475,
"learning_rate": 3.5748394318107765e-05,
"loss": 0.4181,
"step": 17645
},
{
"epoch": 2.0580690298507465,
"grad_norm": 0.3896131004903475,
"learning_rate": 3.5740321640740646e-05,
"loss": 0.3995,
"step": 17650
},
{
"epoch": 2.058652052238806,
"grad_norm": 0.44817209090164584,
"learning_rate": 3.573224773811665e-05,
"loss": 0.4132,
"step": 17655
},
{
"epoch": 2.0592350746268657,
"grad_norm": 0.4601241879579617,
"learning_rate": 3.57241726114363e-05,
"loss": 0.4123,
"step": 17660
},
{
"epoch": 2.0598180970149254,
"grad_norm": 0.4332211901795128,
"learning_rate": 3.5716096261900274e-05,
"loss": 0.4097,
"step": 17665
},
{
"epoch": 2.060401119402985,
"grad_norm": 0.4435205427580933,
"learning_rate": 3.570801869070945e-05,
"loss": 0.4233,
"step": 17670
},
{
"epoch": 2.0609841417910446,
"grad_norm": 0.4089469394108028,
"learning_rate": 3.5699939899064894e-05,
"loss": 0.3906,
"step": 17675
},
{
"epoch": 2.0615671641791047,
"grad_norm": 0.39929033875936315,
"learning_rate": 3.5691859888167846e-05,
"loss": 0.3697,
"step": 17680
},
{
"epoch": 2.0621501865671643,
"grad_norm": 0.4032698121477907,
"learning_rate": 3.5683778659219714e-05,
"loss": 0.3849,
"step": 17685
},
{
"epoch": 2.062733208955224,
"grad_norm": 0.39957157338321503,
"learning_rate": 3.5675696213422105e-05,
"loss": 0.4,
"step": 17690
},
{
"epoch": 2.0633162313432836,
"grad_norm": 0.47575181622872736,
"learning_rate": 3.56676125519768e-05,
"loss": 0.3997,
"step": 17695
},
{
"epoch": 2.063899253731343,
"grad_norm": 0.42660736711985703,
"learning_rate": 3.5659527676085774e-05,
"loss": 0.4026,
"step": 17700
},
{
"epoch": 2.064482276119403,
"grad_norm": 0.42536589886353837,
"learning_rate": 3.565144158695115e-05,
"loss": 0.4148,
"step": 17705
},
{
"epoch": 2.065065298507463,
"grad_norm": 0.420815781715264,
"learning_rate": 3.564335428577526e-05,
"loss": 0.4157,
"step": 17710
},
{
"epoch": 2.0656483208955225,
"grad_norm": 0.37727500584762214,
"learning_rate": 3.5635265773760625e-05,
"loss": 0.3936,
"step": 17715
},
{
"epoch": 2.066231343283582,
"grad_norm": 0.38246861442370994,
"learning_rate": 3.56271760521099e-05,
"loss": 0.3855,
"step": 17720
},
{
"epoch": 2.066814365671642,
"grad_norm": 0.4083430209998539,
"learning_rate": 3.561908512202597e-05,
"loss": 0.4223,
"step": 17725
},
{
"epoch": 2.0673973880597014,
"grad_norm": 0.40292647230277046,
"learning_rate": 3.561099298471187e-05,
"loss": 0.3884,
"step": 17730
},
{
"epoch": 2.067980410447761,
"grad_norm": 0.406003317049021,
"learning_rate": 3.560289964137083e-05,
"loss": 0.4112,
"step": 17735
},
{
"epoch": 2.0685634328358207,
"grad_norm": 0.4459596903475761,
"learning_rate": 3.559480509320625e-05,
"loss": 0.4178,
"step": 17740
},
{
"epoch": 2.0691464552238807,
"grad_norm": 0.42667037304533173,
"learning_rate": 3.558670934142171e-05,
"loss": 0.3871,
"step": 17745
},
{
"epoch": 2.0697294776119404,
"grad_norm": 0.41694123362390095,
"learning_rate": 3.557861238722097e-05,
"loss": 0.3812,
"step": 17750
},
{
"epoch": 2.0703125,
"grad_norm": 0.4639670327141468,
"learning_rate": 3.557051423180797e-05,
"loss": 0.406,
"step": 17755
},
{
"epoch": 2.0708955223880596,
"grad_norm": 0.4536747585597657,
"learning_rate": 3.556241487638682e-05,
"loss": 0.4184,
"step": 17760
},
{
"epoch": 2.0714785447761193,
"grad_norm": 0.463341281056315,
"learning_rate": 3.5554314322161844e-05,
"loss": 0.4251,
"step": 17765
},
{
"epoch": 2.0720615671641793,
"grad_norm": 0.4236756562769349,
"learning_rate": 3.554621257033749e-05,
"loss": 0.4342,
"step": 17770
},
{
"epoch": 2.072644589552239,
"grad_norm": 0.40098119433640333,
"learning_rate": 3.5538109622118414e-05,
"loss": 0.4252,
"step": 17775
},
{
"epoch": 2.0732276119402986,
"grad_norm": 0.5462436217767143,
"learning_rate": 3.5530005478709446e-05,
"loss": 0.4228,
"step": 17780
},
{
"epoch": 2.073810634328358,
"grad_norm": 0.5377323817355546,
"learning_rate": 3.552190014131562e-05,
"loss": 0.4102,
"step": 17785
},
{
"epoch": 2.074393656716418,
"grad_norm": 0.4414133041774776,
"learning_rate": 3.551379361114209e-05,
"loss": 0.3974,
"step": 17790
},
{
"epoch": 2.0749766791044775,
"grad_norm": 0.43093720318960205,
"learning_rate": 3.550568588939423e-05,
"loss": 0.3919,
"step": 17795
},
{
"epoch": 2.075559701492537,
"grad_norm": 0.3987041971189923,
"learning_rate": 3.549757697727759e-05,
"loss": 0.4058,
"step": 17800
},
{
"epoch": 2.076142723880597,
"grad_norm": 0.46671630573375056,
"learning_rate": 3.5489466875997876e-05,
"loss": 0.4105,
"step": 17805
},
{
"epoch": 2.076725746268657,
"grad_norm": 0.42786744581372393,
"learning_rate": 3.548135558676098e-05,
"loss": 0.3887,
"step": 17810
},
{
"epoch": 2.0773087686567164,
"grad_norm": 0.41607289536770725,
"learning_rate": 3.547324311077299e-05,
"loss": 0.4048,
"step": 17815
},
{
"epoch": 2.077891791044776,
"grad_norm": 0.45844722867997517,
"learning_rate": 3.546512944924014e-05,
"loss": 0.4312,
"step": 17820
},
{
"epoch": 2.0784748134328357,
"grad_norm": 0.6436067039745295,
"learning_rate": 3.5457014603368844e-05,
"loss": 0.3995,
"step": 17825
},
{
"epoch": 2.0790578358208953,
"grad_norm": 0.40376232253464883,
"learning_rate": 3.544889857436573e-05,
"loss": 0.4047,
"step": 17830
},
{
"epoch": 2.0796408582089554,
"grad_norm": 0.4424008134172488,
"learning_rate": 3.544078136343755e-05,
"loss": 0.3986,
"step": 17835
},
{
"epoch": 2.080223880597015,
"grad_norm": 0.45334469023841945,
"learning_rate": 3.5432662971791264e-05,
"loss": 0.4157,
"step": 17840
},
{
"epoch": 2.0808069029850746,
"grad_norm": 0.4197736763225393,
"learning_rate": 3.5424543400634e-05,
"loss": 0.4154,
"step": 17845
},
{
"epoch": 2.0813899253731343,
"grad_norm": 1.919560356302374,
"learning_rate": 3.541642265117306e-05,
"loss": 0.4086,
"step": 17850
},
{
"epoch": 2.081972947761194,
"grad_norm": 0.4376616144845669,
"learning_rate": 3.5408300724615915e-05,
"loss": 0.3762,
"step": 17855
},
{
"epoch": 2.0825559701492535,
"grad_norm": 0.394848239549364,
"learning_rate": 3.540017762217023e-05,
"loss": 0.405,
"step": 17860
},
{
"epoch": 2.0831389925373136,
"grad_norm": 0.39351720162653736,
"learning_rate": 3.539205334504382e-05,
"loss": 0.3845,
"step": 17865
},
{
"epoch": 2.0837220149253732,
"grad_norm": 0.4296218352693514,
"learning_rate": 3.5383927894444694e-05,
"loss": 0.4308,
"step": 17870
},
{
"epoch": 2.084305037313433,
"grad_norm": 0.44104359649301106,
"learning_rate": 3.537580127158103e-05,
"loss": 0.442,
"step": 17875
},
{
"epoch": 2.0848880597014925,
"grad_norm": 0.3921743818551184,
"learning_rate": 3.5367673477661174e-05,
"loss": 0.3775,
"step": 17880
},
{
"epoch": 2.085471082089552,
"grad_norm": 0.45131654218140227,
"learning_rate": 3.5359544513893655e-05,
"loss": 0.4044,
"step": 17885
},
{
"epoch": 2.0860541044776117,
"grad_norm": 0.4537637907819424,
"learning_rate": 3.535141438148717e-05,
"loss": 0.4257,
"step": 17890
},
{
"epoch": 2.086637126865672,
"grad_norm": 0.4064131288923466,
"learning_rate": 3.53432830816506e-05,
"loss": 0.414,
"step": 17895
},
{
"epoch": 2.0872201492537314,
"grad_norm": 0.3900103655232811,
"learning_rate": 3.533515061559297e-05,
"loss": 0.396,
"step": 17900
},
{
"epoch": 2.087803171641791,
"grad_norm": 0.42363353505397877,
"learning_rate": 3.532701698452352e-05,
"loss": 0.4104,
"step": 17905
},
{
"epoch": 2.0883861940298507,
"grad_norm": 0.4775771231746664,
"learning_rate": 3.5318882189651635e-05,
"loss": 0.4063,
"step": 17910
},
{
"epoch": 2.0889692164179103,
"grad_norm": 0.45616241876431546,
"learning_rate": 3.531074623218689e-05,
"loss": 0.4081,
"step": 17915
},
{
"epoch": 2.08955223880597,
"grad_norm": 0.42170797688086853,
"learning_rate": 3.5302609113339e-05,
"loss": 0.3844,
"step": 17920
},
{
"epoch": 2.09013526119403,
"grad_norm": 0.439242908636829,
"learning_rate": 3.52944708343179e-05,
"loss": 0.4012,
"step": 17925
},
{
"epoch": 2.0907182835820897,
"grad_norm": 0.46414434867866833,
"learning_rate": 3.5286331396333675e-05,
"loss": 0.4167,
"step": 17930
},
{
"epoch": 2.0913013059701493,
"grad_norm": 0.42598616305498765,
"learning_rate": 3.527819080059657e-05,
"loss": 0.433,
"step": 17935
},
{
"epoch": 2.091884328358209,
"grad_norm": 1.9646733034362873,
"learning_rate": 3.5270049048317016e-05,
"loss": 0.3971,
"step": 17940
},
{
"epoch": 2.0924673507462686,
"grad_norm": 0.442469170939552,
"learning_rate": 3.5261906140705615e-05,
"loss": 0.4112,
"step": 17945
},
{
"epoch": 2.093050373134328,
"grad_norm": 0.4102310568186522,
"learning_rate": 3.525376207897314e-05,
"loss": 0.3888,
"step": 17950
},
{
"epoch": 2.0936333955223883,
"grad_norm": 0.40470558046344235,
"learning_rate": 3.524561686433053e-05,
"loss": 0.39,
"step": 17955
},
{
"epoch": 2.094216417910448,
"grad_norm": 0.4257906695133037,
"learning_rate": 3.5237470497988905e-05,
"loss": 0.4057,
"step": 17960
},
{
"epoch": 2.0947994402985075,
"grad_norm": 0.4570530733280104,
"learning_rate": 3.522932298115955e-05,
"loss": 0.3993,
"step": 17965
},
{
"epoch": 2.095382462686567,
"grad_norm": 0.43448859420515695,
"learning_rate": 3.5221174315053935e-05,
"loss": 0.388,
"step": 17970
},
{
"epoch": 2.0959654850746268,
"grad_norm": 0.45632890224050665,
"learning_rate": 3.5213024500883666e-05,
"loss": 0.3922,
"step": 17975
},
{
"epoch": 2.0965485074626864,
"grad_norm": 0.40012892603825206,
"learning_rate": 3.520487353986056e-05,
"loss": 0.4059,
"step": 17980
},
{
"epoch": 2.0971315298507465,
"grad_norm": 0.41112655096189904,
"learning_rate": 3.5196721433196575e-05,
"loss": 0.4021,
"step": 17985
},
{
"epoch": 2.097714552238806,
"grad_norm": 0.41389615796919993,
"learning_rate": 3.518856818210387e-05,
"loss": 0.3994,
"step": 17990
},
{
"epoch": 2.0982975746268657,
"grad_norm": 0.4131593212558457,
"learning_rate": 3.5180413787794724e-05,
"loss": 0.451,
"step": 17995
},
{
"epoch": 2.0988805970149254,
"grad_norm": 0.42752541290912965,
"learning_rate": 3.517225825148164e-05,
"loss": 0.4025,
"step": 18000
},
{
"epoch": 2.099463619402985,
"grad_norm": 0.502384225015268,
"learning_rate": 3.516410157437727e-05,
"loss": 0.4255,
"step": 18005
},
{
"epoch": 2.1000466417910446,
"grad_norm": 0.42592470933952536,
"learning_rate": 3.515594375769442e-05,
"loss": 0.4134,
"step": 18010
},
{
"epoch": 2.1006296641791047,
"grad_norm": 0.4755593904042131,
"learning_rate": 3.514778480264609e-05,
"loss": 0.4399,
"step": 18015
},
{
"epoch": 2.1012126865671643,
"grad_norm": 0.39159887557747713,
"learning_rate": 3.513962471044543e-05,
"loss": 0.4093,
"step": 18020
},
{
"epoch": 2.101795708955224,
"grad_norm": 0.44229008867276415,
"learning_rate": 3.513146348230578e-05,
"loss": 0.4095,
"step": 18025
},
{
"epoch": 2.1023787313432836,
"grad_norm": 0.4463152827359944,
"learning_rate": 3.512330111944062e-05,
"loss": 0.4183,
"step": 18030
},
{
"epoch": 2.102961753731343,
"grad_norm": 0.4263982192743119,
"learning_rate": 3.511513762306363e-05,
"loss": 0.4039,
"step": 18035
},
{
"epoch": 2.103544776119403,
"grad_norm": 0.3764688587303262,
"learning_rate": 3.510697299438864e-05,
"loss": 0.3985,
"step": 18040
},
{
"epoch": 2.104127798507463,
"grad_norm": 0.7883615400214776,
"learning_rate": 3.509880723462965e-05,
"loss": 0.4069,
"step": 18045
},
{
"epoch": 2.1047108208955225,
"grad_norm": 0.45316947689705445,
"learning_rate": 3.509064034500082e-05,
"loss": 0.4279,
"step": 18050
},
{
"epoch": 2.105293843283582,
"grad_norm": 0.4323162177598834,
"learning_rate": 3.50824723267165e-05,
"loss": 0.4259,
"step": 18055
},
{
"epoch": 2.105876865671642,
"grad_norm": 0.4157293684577551,
"learning_rate": 3.50743031809912e-05,
"loss": 0.4008,
"step": 18060
},
{
"epoch": 2.1064598880597014,
"grad_norm": 0.41443560930906936,
"learning_rate": 3.506613290903958e-05,
"loss": 0.4091,
"step": 18065
},
{
"epoch": 2.107042910447761,
"grad_norm": 1.6540605463548794,
"learning_rate": 3.505796151207651e-05,
"loss": 0.4126,
"step": 18070
},
{
"epoch": 2.1076259328358207,
"grad_norm": 0.45561508384631727,
"learning_rate": 3.504978899131696e-05,
"loss": 0.3996,
"step": 18075
},
{
"epoch": 2.1082089552238807,
"grad_norm": 0.4434447916074442,
"learning_rate": 3.504161534797612e-05,
"loss": 0.4,
"step": 18080
},
{
"epoch": 2.1087919776119404,
"grad_norm": 0.4279181497681623,
"learning_rate": 3.503344058326934e-05,
"loss": 0.4212,
"step": 18085
},
{
"epoch": 2.109375,
"grad_norm": 0.4186326705057981,
"learning_rate": 3.5025264698412126e-05,
"loss": 0.4048,
"step": 18090
},
{
"epoch": 2.1099580223880596,
"grad_norm": 0.42558836872774103,
"learning_rate": 3.5017087694620154e-05,
"loss": 0.4056,
"step": 18095
},
{
"epoch": 2.1105410447761193,
"grad_norm": 0.4585705588952108,
"learning_rate": 3.500890957310926e-05,
"loss": 0.449,
"step": 18100
},
{
"epoch": 2.1111240671641793,
"grad_norm": 0.39868856032720257,
"learning_rate": 3.500073033509546e-05,
"loss": 0.3963,
"step": 18105
},
{
"epoch": 2.111707089552239,
"grad_norm": 0.4518611231636433,
"learning_rate": 3.4992549981794915e-05,
"loss": 0.4087,
"step": 18110
},
{
"epoch": 2.1122901119402986,
"grad_norm": 0.41093552739258665,
"learning_rate": 3.498436851442398e-05,
"loss": 0.4011,
"step": 18115
},
{
"epoch": 2.112873134328358,
"grad_norm": 0.4551342557687998,
"learning_rate": 3.497618593419916e-05,
"loss": 0.4195,
"step": 18120
},
{
"epoch": 2.113456156716418,
"grad_norm": 0.4249640753172037,
"learning_rate": 3.496800224233713e-05,
"loss": 0.4222,
"step": 18125
},
{
"epoch": 2.1140391791044775,
"grad_norm": 0.4317873459107346,
"learning_rate": 3.495981744005471e-05,
"loss": 0.3798,
"step": 18130
},
{
"epoch": 2.114622201492537,
"grad_norm": 0.43776493112473797,
"learning_rate": 3.49516315285689e-05,
"loss": 0.4436,
"step": 18135
},
{
"epoch": 2.115205223880597,
"grad_norm": 0.4536385624082108,
"learning_rate": 3.494344450909689e-05,
"loss": 0.4279,
"step": 18140
},
{
"epoch": 2.115788246268657,
"grad_norm": 0.45679071602941196,
"learning_rate": 3.4935256382855996e-05,
"loss": 0.4218,
"step": 18145
},
{
"epoch": 2.1163712686567164,
"grad_norm": 0.43460120047194506,
"learning_rate": 3.492706715106372e-05,
"loss": 0.3872,
"step": 18150
},
{
"epoch": 2.116954291044776,
"grad_norm": 0.43466514815981416,
"learning_rate": 3.491887681493771e-05,
"loss": 0.4079,
"step": 18155
},
{
"epoch": 2.1175373134328357,
"grad_norm": 0.43627926272393935,
"learning_rate": 3.491068537569581e-05,
"loss": 0.4141,
"step": 18160
},
{
"epoch": 2.1181203358208953,
"grad_norm": 0.400505012948606,
"learning_rate": 3.4902492834555994e-05,
"loss": 0.4082,
"step": 18165
},
{
"epoch": 2.1187033582089554,
"grad_norm": 0.40465376681171294,
"learning_rate": 3.489429919273642e-05,
"loss": 0.3956,
"step": 18170
},
{
"epoch": 2.119286380597015,
"grad_norm": 0.4656852789402069,
"learning_rate": 3.488610445145539e-05,
"loss": 0.4346,
"step": 18175
},
{
"epoch": 2.1198694029850746,
"grad_norm": 0.41113986376765876,
"learning_rate": 3.4877908611931406e-05,
"loss": 0.4083,
"step": 18180
},
{
"epoch": 2.1204524253731343,
"grad_norm": 0.42259438586971515,
"learning_rate": 3.48697116753831e-05,
"loss": 0.4146,
"step": 18185
},
{
"epoch": 2.121035447761194,
"grad_norm": 0.4004898003730871,
"learning_rate": 3.486151364302928e-05,
"loss": 0.392,
"step": 18190
},
{
"epoch": 2.1216184701492535,
"grad_norm": 0.4492710402394599,
"learning_rate": 3.485331451608891e-05,
"loss": 0.406,
"step": 18195
},
{
"epoch": 2.1222014925373136,
"grad_norm": 0.41841886702855924,
"learning_rate": 3.484511429578113e-05,
"loss": 0.4094,
"step": 18200
},
{
"epoch": 2.1227845149253732,
"grad_norm": 0.383432454482924,
"learning_rate": 3.483691298332522e-05,
"loss": 0.3925,
"step": 18205
},
{
"epoch": 2.123367537313433,
"grad_norm": 0.40871040660467367,
"learning_rate": 3.482871057994065e-05,
"loss": 0.4174,
"step": 18210
},
{
"epoch": 2.1239505597014925,
"grad_norm": 0.41203869297679074,
"learning_rate": 3.4820507086847036e-05,
"loss": 0.4183,
"step": 18215
},
{
"epoch": 2.124533582089552,
"grad_norm": 0.39172999651212204,
"learning_rate": 3.481230250526416e-05,
"loss": 0.3882,
"step": 18220
},
{
"epoch": 2.1251166044776117,
"grad_norm": 0.39979542942187735,
"learning_rate": 3.480409683641196e-05,
"loss": 0.3902,
"step": 18225
},
{
"epoch": 2.125699626865672,
"grad_norm": 0.4188582512860355,
"learning_rate": 3.479589008151054e-05,
"loss": 0.4008,
"step": 18230
},
{
"epoch": 2.1262826492537314,
"grad_norm": 0.4018983973265039,
"learning_rate": 3.4787682241780164e-05,
"loss": 0.4247,
"step": 18235
},
{
"epoch": 2.126865671641791,
"grad_norm": 0.492136567359692,
"learning_rate": 3.477947331844127e-05,
"loss": 0.4028,
"step": 18240
},
{
"epoch": 2.1274486940298507,
"grad_norm": 0.4746116975916419,
"learning_rate": 3.477126331271445e-05,
"loss": 0.4032,
"step": 18245
},
{
"epoch": 2.1280317164179103,
"grad_norm": 0.46327889070060085,
"learning_rate": 3.476305222582042e-05,
"loss": 0.4051,
"step": 18250
},
{
"epoch": 2.12861473880597,
"grad_norm": 0.9946701863145109,
"learning_rate": 3.475484005898013e-05,
"loss": 0.419,
"step": 18255
},
{
"epoch": 2.12919776119403,
"grad_norm": 0.457118037673843,
"learning_rate": 3.4746626813414624e-05,
"loss": 0.4194,
"step": 18260
},
{
"epoch": 2.1297807835820897,
"grad_norm": 0.42676681216035084,
"learning_rate": 3.473841249034514e-05,
"loss": 0.3873,
"step": 18265
},
{
"epoch": 2.1303638059701493,
"grad_norm": 0.4175294629764706,
"learning_rate": 3.4730197090993084e-05,
"loss": 0.3949,
"step": 18270
},
{
"epoch": 2.130946828358209,
"grad_norm": 0.4429123065062491,
"learning_rate": 3.4721980616579984e-05,
"loss": 0.3859,
"step": 18275
},
{
"epoch": 2.1315298507462686,
"grad_norm": 0.42960011018651256,
"learning_rate": 3.471376306832756e-05,
"loss": 0.3879,
"step": 18280
},
{
"epoch": 2.132112873134328,
"grad_norm": 0.3977545876838716,
"learning_rate": 3.4705544447457686e-05,
"loss": 0.3659,
"step": 18285
},
{
"epoch": 2.1326958955223883,
"grad_norm": 0.4071063253380223,
"learning_rate": 3.4697324755192387e-05,
"loss": 0.4226,
"step": 18290
},
{
"epoch": 2.133278917910448,
"grad_norm": 0.3928332736265074,
"learning_rate": 3.468910399275387e-05,
"loss": 0.4177,
"step": 18295
},
{
"epoch": 2.1338619402985075,
"grad_norm": 0.4426832092037558,
"learning_rate": 3.468088216136445e-05,
"loss": 0.4196,
"step": 18300
},
{
"epoch": 2.134444962686567,
"grad_norm": 0.4412295617512612,
"learning_rate": 3.467265926224667e-05,
"loss": 0.4098,
"step": 18305
},
{
"epoch": 2.1350279850746268,
"grad_norm": 0.5185023740026237,
"learning_rate": 3.466443529662317e-05,
"loss": 0.4195,
"step": 18310
},
{
"epoch": 2.1356110074626864,
"grad_norm": 0.4505364691967898,
"learning_rate": 3.465621026571679e-05,
"loss": 0.4118,
"step": 18315
},
{
"epoch": 2.1361940298507465,
"grad_norm": 0.40175822152730395,
"learning_rate": 3.4647984170750506e-05,
"loss": 0.3971,
"step": 18320
},
{
"epoch": 2.136777052238806,
"grad_norm": 0.4230475965578641,
"learning_rate": 3.463975701294747e-05,
"loss": 0.4035,
"step": 18325
},
{
"epoch": 2.1373600746268657,
"grad_norm": 0.46159035331101195,
"learning_rate": 3.463152879353097e-05,
"loss": 0.4105,
"step": 18330
},
{
"epoch": 2.1379430970149254,
"grad_norm": 0.4081762936664928,
"learning_rate": 3.462329951372446e-05,
"loss": 0.3938,
"step": 18335
},
{
"epoch": 2.138526119402985,
"grad_norm": 0.4162526437244533,
"learning_rate": 3.4615069174751566e-05,
"loss": 0.4065,
"step": 18340
},
{
"epoch": 2.1391091417910446,
"grad_norm": 0.41395818841603066,
"learning_rate": 3.460683777783605e-05,
"loss": 0.3902,
"step": 18345
},
{
"epoch": 2.1396921641791047,
"grad_norm": 0.4000044137100815,
"learning_rate": 3.459860532420186e-05,
"loss": 0.4215,
"step": 18350
},
{
"epoch": 2.1402751865671643,
"grad_norm": 0.46193139026077534,
"learning_rate": 3.459037181507307e-05,
"loss": 0.3989,
"step": 18355
},
{
"epoch": 2.140858208955224,
"grad_norm": 0.38117231649973093,
"learning_rate": 3.4582137251673916e-05,
"loss": 0.3833,
"step": 18360
},
{
"epoch": 2.1414412313432836,
"grad_norm": 0.4004118722461508,
"learning_rate": 3.4573901635228815e-05,
"loss": 0.4215,
"step": 18365
},
{
"epoch": 2.142024253731343,
"grad_norm": 0.4255852414170699,
"learning_rate": 3.456566496696232e-05,
"loss": 0.4079,
"step": 18370
},
{
"epoch": 2.142607276119403,
"grad_norm": 0.4694852826837166,
"learning_rate": 3.455742724809914e-05,
"loss": 0.4061,
"step": 18375
},
{
"epoch": 2.143190298507463,
"grad_norm": 0.40555512916946207,
"learning_rate": 3.454918847986414e-05,
"loss": 0.3846,
"step": 18380
},
{
"epoch": 2.1437733208955225,
"grad_norm": 0.4686029056448322,
"learning_rate": 3.4540948663482356e-05,
"loss": 0.4174,
"step": 18385
},
{
"epoch": 2.144356343283582,
"grad_norm": 0.4259521402032897,
"learning_rate": 3.453270780017897e-05,
"loss": 0.4211,
"step": 18390
},
{
"epoch": 2.144939365671642,
"grad_norm": 0.42338287414652503,
"learning_rate": 3.452446589117932e-05,
"loss": 0.4064,
"step": 18395
},
{
"epoch": 2.1455223880597014,
"grad_norm": 0.45445782962606823,
"learning_rate": 3.451622293770889e-05,
"loss": 0.4315,
"step": 18400
},
{
"epoch": 2.146105410447761,
"grad_norm": 0.4193864890396944,
"learning_rate": 3.450797894099332e-05,
"loss": 0.4213,
"step": 18405
},
{
"epoch": 2.1466884328358207,
"grad_norm": 0.4045938647292487,
"learning_rate": 3.4499733902258446e-05,
"loss": 0.3765,
"step": 18410
},
{
"epoch": 2.1472714552238807,
"grad_norm": 0.4375966310441991,
"learning_rate": 3.4491487822730194e-05,
"loss": 0.4165,
"step": 18415
},
{
"epoch": 2.1478544776119404,
"grad_norm": 0.4021322452390154,
"learning_rate": 3.448324070363469e-05,
"loss": 0.3987,
"step": 18420
},
{
"epoch": 2.1484375,
"grad_norm": 0.656108292610777,
"learning_rate": 3.447499254619821e-05,
"loss": 0.4061,
"step": 18425
},
{
"epoch": 2.1490205223880596,
"grad_norm": 0.4662551126290562,
"learning_rate": 3.446674335164716e-05,
"loss": 0.416,
"step": 18430
},
{
"epoch": 2.1496035447761193,
"grad_norm": 0.4203431257433708,
"learning_rate": 3.445849312120813e-05,
"loss": 0.3833,
"step": 18435
},
{
"epoch": 2.1501865671641793,
"grad_norm": 0.43701350646468623,
"learning_rate": 3.445024185610783e-05,
"loss": 0.4031,
"step": 18440
},
{
"epoch": 2.150769589552239,
"grad_norm": 0.41464833300465653,
"learning_rate": 3.444198955757316e-05,
"loss": 0.4291,
"step": 18445
},
{
"epoch": 2.1513526119402986,
"grad_norm": 0.38800303739700287,
"learning_rate": 3.443373622683116e-05,
"loss": 0.3794,
"step": 18450
},
{
"epoch": 2.151935634328358,
"grad_norm": 0.5015747533688855,
"learning_rate": 3.442548186510902e-05,
"loss": 0.4224,
"step": 18455
},
{
"epoch": 2.152518656716418,
"grad_norm": 0.44318396438444974,
"learning_rate": 3.441722647363408e-05,
"loss": 0.3991,
"step": 18460
},
{
"epoch": 2.1531016791044775,
"grad_norm": 0.4629739351457824,
"learning_rate": 3.4408970053633835e-05,
"loss": 0.4237,
"step": 18465
},
{
"epoch": 2.153684701492537,
"grad_norm": 0.46756060010063044,
"learning_rate": 3.440071260633594e-05,
"loss": 0.4228,
"step": 18470
},
{
"epoch": 2.154267723880597,
"grad_norm": 0.43654862859242105,
"learning_rate": 3.43924541329682e-05,
"loss": 0.4125,
"step": 18475
},
{
"epoch": 2.154850746268657,
"grad_norm": 0.3975759879864766,
"learning_rate": 3.438419463475857e-05,
"loss": 0.3953,
"step": 18480
},
{
"epoch": 2.1554337686567164,
"grad_norm": 0.43559577817094774,
"learning_rate": 3.437593411293516e-05,
"loss": 0.4237,
"step": 18485
},
{
"epoch": 2.156016791044776,
"grad_norm": 0.4153504610787961,
"learning_rate": 3.436767256872621e-05,
"loss": 0.4057,
"step": 18490
},
{
"epoch": 2.1565998134328357,
"grad_norm": 0.39063787940198125,
"learning_rate": 3.435941000336016e-05,
"loss": 0.399,
"step": 18495
},
{
"epoch": 2.1571828358208953,
"grad_norm": 0.41088878460192013,
"learning_rate": 3.435114641806557e-05,
"loss": 0.4191,
"step": 18500
},
{
"epoch": 2.1577658582089554,
"grad_norm": 0.40787535265956926,
"learning_rate": 3.434288181407114e-05,
"loss": 0.4269,
"step": 18505
},
{
"epoch": 2.158348880597015,
"grad_norm": 0.42008146833643994,
"learning_rate": 3.433461619260575e-05,
"loss": 0.4151,
"step": 18510
},
{
"epoch": 2.1589319029850746,
"grad_norm": 0.4719089973719463,
"learning_rate": 3.432634955489841e-05,
"loss": 0.435,
"step": 18515
},
{
"epoch": 2.1595149253731343,
"grad_norm": 0.4340721315473417,
"learning_rate": 3.43180819021783e-05,
"loss": 0.4143,
"step": 18520
},
{
"epoch": 2.160097947761194,
"grad_norm": 0.412517457467355,
"learning_rate": 3.430981323567475e-05,
"loss": 0.4301,
"step": 18525
},
{
"epoch": 2.1606809701492535,
"grad_norm": 0.4357379435528156,
"learning_rate": 3.4301543556617206e-05,
"loss": 0.396,
"step": 18530
},
{
"epoch": 2.1612639925373136,
"grad_norm": 0.4456223599662238,
"learning_rate": 3.42932728662353e-05,
"loss": 0.4141,
"step": 18535
},
{
"epoch": 2.1618470149253732,
"grad_norm": 0.4304404852290806,
"learning_rate": 3.428500116575881e-05,
"loss": 0.4279,
"step": 18540
},
{
"epoch": 2.162430037313433,
"grad_norm": 0.4180683020284989,
"learning_rate": 3.427672845641765e-05,
"loss": 0.4214,
"step": 18545
},
{
"epoch": 2.1630130597014925,
"grad_norm": 0.4013336763453828,
"learning_rate": 3.42684547394419e-05,
"loss": 0.391,
"step": 18550
},
{
"epoch": 2.163596082089552,
"grad_norm": 0.4293341803829475,
"learning_rate": 3.4260180016061784e-05,
"loss": 0.4085,
"step": 18555
},
{
"epoch": 2.1641791044776117,
"grad_norm": 0.4078104883279643,
"learning_rate": 3.425190428750767e-05,
"loss": 0.3774,
"step": 18560
},
{
"epoch": 2.164762126865672,
"grad_norm": 0.6462307153545634,
"learning_rate": 3.424362755501007e-05,
"loss": 0.3849,
"step": 18565
},
{
"epoch": 2.1653451492537314,
"grad_norm": 0.3935470087549329,
"learning_rate": 3.423534981979968e-05,
"loss": 0.3944,
"step": 18570
},
{
"epoch": 2.165928171641791,
"grad_norm": 0.7172474120312462,
"learning_rate": 3.422707108310729e-05,
"loss": 0.4279,
"step": 18575
},
{
"epoch": 2.1665111940298507,
"grad_norm": 0.4118325792008938,
"learning_rate": 3.4218791346163894e-05,
"loss": 0.4032,
"step": 18580
},
{
"epoch": 2.1670942164179103,
"grad_norm": 0.4233889408435394,
"learning_rate": 3.421051061020059e-05,
"loss": 0.4088,
"step": 18585
},
{
"epoch": 2.16767723880597,
"grad_norm": 0.4072578863626613,
"learning_rate": 3.420222887644866e-05,
"loss": 0.4148,
"step": 18590
},
{
"epoch": 2.16826026119403,
"grad_norm": 0.4626653985981102,
"learning_rate": 3.419394614613951e-05,
"loss": 0.4195,
"step": 18595
},
{
"epoch": 2.1688432835820897,
"grad_norm": 0.4089173557620577,
"learning_rate": 3.41856624205047e-05,
"loss": 0.4023,
"step": 18600
},
{
"epoch": 2.1694263059701493,
"grad_norm": 0.5967881339025768,
"learning_rate": 3.417737770077595e-05,
"loss": 0.4372,
"step": 18605
},
{
"epoch": 2.170009328358209,
"grad_norm": 0.4581423130322887,
"learning_rate": 3.4169091988185106e-05,
"loss": 0.4144,
"step": 18610
},
{
"epoch": 2.1705923507462686,
"grad_norm": 0.43372339823006584,
"learning_rate": 3.4160805283964184e-05,
"loss": 0.4088,
"step": 18615
},
{
"epoch": 2.171175373134328,
"grad_norm": 0.4317014186436621,
"learning_rate": 3.415251758934534e-05,
"loss": 0.4311,
"step": 18620
},
{
"epoch": 2.1717583955223883,
"grad_norm": 0.43081952318192973,
"learning_rate": 3.414422890556087e-05,
"loss": 0.3964,
"step": 18625
},
{
"epoch": 2.172341417910448,
"grad_norm": 0.43556154011233644,
"learning_rate": 3.413593923384321e-05,
"loss": 0.4198,
"step": 18630
},
{
"epoch": 2.1729244402985075,
"grad_norm": 0.40153583373509305,
"learning_rate": 3.4127648575424975e-05,
"loss": 0.3962,
"step": 18635
},
{
"epoch": 2.173507462686567,
"grad_norm": 0.39965651070488356,
"learning_rate": 3.4119356931538894e-05,
"loss": 0.3813,
"step": 18640
},
{
"epoch": 2.1740904850746268,
"grad_norm": 0.42253742875349837,
"learning_rate": 3.411106430341786e-05,
"loss": 0.394,
"step": 18645
},
{
"epoch": 2.1746735074626864,
"grad_norm": 0.467515445902988,
"learning_rate": 3.410277069229491e-05,
"loss": 0.4244,
"step": 18650
},
{
"epoch": 2.1752565298507465,
"grad_norm": 0.3986272829489681,
"learning_rate": 3.409447609940322e-05,
"loss": 0.4061,
"step": 18655
},
{
"epoch": 2.175839552238806,
"grad_norm": 0.40896378231540925,
"learning_rate": 3.408618052597611e-05,
"loss": 0.4067,
"step": 18660
},
{
"epoch": 2.1764225746268657,
"grad_norm": 0.459152029100428,
"learning_rate": 3.407788397324706e-05,
"loss": 0.446,
"step": 18665
},
{
"epoch": 2.1770055970149254,
"grad_norm": 0.4216994814627903,
"learning_rate": 3.4069586442449684e-05,
"loss": 0.3913,
"step": 18670
},
{
"epoch": 2.177588619402985,
"grad_norm": 0.4434271126358292,
"learning_rate": 3.406128793481776e-05,
"loss": 0.41,
"step": 18675
},
{
"epoch": 2.1781716417910446,
"grad_norm": 0.4231177774602946,
"learning_rate": 3.405298845158518e-05,
"loss": 0.3915,
"step": 18680
},
{
"epoch": 2.1787546641791047,
"grad_norm": 0.3950903294347333,
"learning_rate": 3.4044687993985995e-05,
"loss": 0.4027,
"step": 18685
},
{
"epoch": 2.1793376865671643,
"grad_norm": 0.4311258737949541,
"learning_rate": 3.403638656325442e-05,
"loss": 0.3971,
"step": 18690
},
{
"epoch": 2.179920708955224,
"grad_norm": 0.3987552988837497,
"learning_rate": 3.402808416062479e-05,
"loss": 0.4345,
"step": 18695
},
{
"epoch": 2.1805037313432836,
"grad_norm": 0.3977106718559317,
"learning_rate": 3.4019780787331586e-05,
"loss": 0.4056,
"step": 18700
},
{
"epoch": 2.181086753731343,
"grad_norm": 0.4691687312056667,
"learning_rate": 3.4011476444609456e-05,
"loss": 0.4051,
"step": 18705
},
{
"epoch": 2.181669776119403,
"grad_norm": 0.44967777394079034,
"learning_rate": 3.4003171133693154e-05,
"loss": 0.4354,
"step": 18710
},
{
"epoch": 2.182252798507463,
"grad_norm": 0.40526689548409556,
"learning_rate": 3.399486485581762e-05,
"loss": 0.4035,
"step": 18715
},
{
"epoch": 2.1828358208955225,
"grad_norm": 0.4395555616835956,
"learning_rate": 3.3986557612217904e-05,
"loss": 0.392,
"step": 18720
},
{
"epoch": 2.183418843283582,
"grad_norm": 0.4498947483029495,
"learning_rate": 3.3978249404129224e-05,
"loss": 0.3945,
"step": 18725
},
{
"epoch": 2.184001865671642,
"grad_norm": 0.4318157927154423,
"learning_rate": 3.396994023278693e-05,
"loss": 0.4375,
"step": 18730
},
{
"epoch": 2.1845848880597014,
"grad_norm": 0.4374380435409525,
"learning_rate": 3.396163009942651e-05,
"loss": 0.3999,
"step": 18735
},
{
"epoch": 2.185167910447761,
"grad_norm": 0.4468219550216439,
"learning_rate": 3.3953319005283606e-05,
"loss": 0.4004,
"step": 18740
},
{
"epoch": 2.1857509328358207,
"grad_norm": 0.425195951885405,
"learning_rate": 3.3945006951594e-05,
"loss": 0.4284,
"step": 18745
},
{
"epoch": 2.1863339552238807,
"grad_norm": 0.44941379499567136,
"learning_rate": 3.393669393959361e-05,
"loss": 0.4061,
"step": 18750
},
{
"epoch": 2.1869169776119404,
"grad_norm": 0.5269031717232117,
"learning_rate": 3.39283799705185e-05,
"loss": 0.4194,
"step": 18755
},
{
"epoch": 2.1875,
"grad_norm": 0.46465051136448565,
"learning_rate": 3.392006504560487e-05,
"loss": 0.4118,
"step": 18760
},
{
"epoch": 2.1880830223880596,
"grad_norm": 0.3792927685182771,
"learning_rate": 3.391174916608909e-05,
"loss": 0.3958,
"step": 18765
},
{
"epoch": 2.1886660447761193,
"grad_norm": 0.4860036129164203,
"learning_rate": 3.390343233320764e-05,
"loss": 0.4069,
"step": 18770
},
{
"epoch": 2.1892490671641793,
"grad_norm": 0.44189085409945256,
"learning_rate": 3.3895114548197154e-05,
"loss": 0.4101,
"step": 18775
},
{
"epoch": 2.189832089552239,
"grad_norm": 0.43937345374335685,
"learning_rate": 3.388679581229441e-05,
"loss": 0.411,
"step": 18780
},
{
"epoch": 2.1904151119402986,
"grad_norm": 0.4116553804776797,
"learning_rate": 3.3878476126736314e-05,
"loss": 0.4091,
"step": 18785
},
{
"epoch": 2.190998134328358,
"grad_norm": 0.3982822877797406,
"learning_rate": 3.3870155492759936e-05,
"loss": 0.3865,
"step": 18790
},
{
"epoch": 2.191581156716418,
"grad_norm": 0.4679843239066103,
"learning_rate": 3.386183391160248e-05,
"loss": 0.4097,
"step": 18795
},
{
"epoch": 2.1921641791044775,
"grad_norm": 0.41721294970723116,
"learning_rate": 3.3853511384501256e-05,
"loss": 0.3947,
"step": 18800
},
{
"epoch": 2.192747201492537,
"grad_norm": 0.5285480534010507,
"learning_rate": 3.384518791269378e-05,
"loss": 0.4121,
"step": 18805
},
{
"epoch": 2.193330223880597,
"grad_norm": 0.38832671898642174,
"learning_rate": 3.3836863497417645e-05,
"loss": 0.3911,
"step": 18810
},
{
"epoch": 2.193913246268657,
"grad_norm": 0.39413625020813214,
"learning_rate": 3.3828538139910626e-05,
"loss": 0.3847,
"step": 18815
},
{
"epoch": 2.1944962686567164,
"grad_norm": 0.40769783959180844,
"learning_rate": 3.382021184141062e-05,
"loss": 0.3942,
"step": 18820
},
{
"epoch": 2.195079291044776,
"grad_norm": 0.42504062253604796,
"learning_rate": 3.3811884603155665e-05,
"loss": 0.3862,
"step": 18825
},
{
"epoch": 2.1956623134328357,
"grad_norm": 0.46077084076022645,
"learning_rate": 3.3803556426383954e-05,
"loss": 0.4021,
"step": 18830
},
{
"epoch": 2.1962453358208953,
"grad_norm": 0.4951549657984212,
"learning_rate": 3.379522731233379e-05,
"loss": 0.4217,
"step": 18835
},
{
"epoch": 2.1968283582089554,
"grad_norm": 0.4149451628834867,
"learning_rate": 3.378689726224364e-05,
"loss": 0.4025,
"step": 18840
},
{
"epoch": 2.197411380597015,
"grad_norm": 0.4442642390103348,
"learning_rate": 3.377856627735211e-05,
"loss": 0.4019,
"step": 18845
},
{
"epoch": 2.1979944029850746,
"grad_norm": 0.4153845645412846,
"learning_rate": 3.3770234358897926e-05,
"loss": 0.41,
"step": 18850
},
{
"epoch": 2.1985774253731343,
"grad_norm": 0.41527227718583565,
"learning_rate": 3.376190150811997e-05,
"loss": 0.4008,
"step": 18855
},
{
"epoch": 2.199160447761194,
"grad_norm": 0.4702777032654794,
"learning_rate": 3.3753567726257255e-05,
"loss": 0.4328,
"step": 18860
},
{
"epoch": 2.1997434701492535,
"grad_norm": 0.43001582614714884,
"learning_rate": 3.3745233014548936e-05,
"loss": 0.4016,
"step": 18865
},
{
"epoch": 2.2003264925373136,
"grad_norm": 0.4475735867598758,
"learning_rate": 3.373689737423431e-05,
"loss": 0.4133,
"step": 18870
},
{
"epoch": 2.2009095149253732,
"grad_norm": 0.4049807990171297,
"learning_rate": 3.372856080655279e-05,
"loss": 0.4036,
"step": 18875
},
{
"epoch": 2.201492537313433,
"grad_norm": 0.4073750550140848,
"learning_rate": 3.372022331274397e-05,
"loss": 0.4137,
"step": 18880
},
{
"epoch": 2.2020755597014925,
"grad_norm": 0.4655568861000359,
"learning_rate": 3.3711884894047526e-05,
"loss": 0.4142,
"step": 18885
},
{
"epoch": 2.202658582089552,
"grad_norm": 0.4272531176487613,
"learning_rate": 3.3703545551703326e-05,
"loss": 0.403,
"step": 18890
},
{
"epoch": 2.2032416044776117,
"grad_norm": 0.37553555316986753,
"learning_rate": 3.369520528695133e-05,
"loss": 0.376,
"step": 18895
},
{
"epoch": 2.203824626865672,
"grad_norm": 0.39230467817125414,
"learning_rate": 3.368686410103167e-05,
"loss": 0.4148,
"step": 18900
},
{
"epoch": 2.2044076492537314,
"grad_norm": 0.4030207298330371,
"learning_rate": 3.36785219951846e-05,
"loss": 0.3791,
"step": 18905
},
{
"epoch": 2.204990671641791,
"grad_norm": 0.4027915953070148,
"learning_rate": 3.367017897065051e-05,
"loss": 0.3822,
"step": 18910
},
{
"epoch": 2.2055736940298507,
"grad_norm": 0.38773701267559524,
"learning_rate": 3.366183502866991e-05,
"loss": 0.4065,
"step": 18915
},
{
"epoch": 2.2061567164179103,
"grad_norm": 0.40914304941626306,
"learning_rate": 3.3653490170483485e-05,
"loss": 0.3976,
"step": 18920
},
{
"epoch": 2.20673973880597,
"grad_norm": 0.3684464651454588,
"learning_rate": 3.364514439733203e-05,
"loss": 0.3978,
"step": 18925
},
{
"epoch": 2.20732276119403,
"grad_norm": 0.45339020959733384,
"learning_rate": 3.363679771045648e-05,
"loss": 0.3868,
"step": 18930
},
{
"epoch": 2.2079057835820897,
"grad_norm": 0.4640714288006895,
"learning_rate": 3.3628450111097914e-05,
"loss": 0.427,
"step": 18935
},
{
"epoch": 2.2084888059701493,
"grad_norm": 0.3832431727307536,
"learning_rate": 3.3620101600497526e-05,
"loss": 0.3794,
"step": 18940
},
{
"epoch": 2.209071828358209,
"grad_norm": 0.41826149036899857,
"learning_rate": 3.361175217989668e-05,
"loss": 0.3823,
"step": 18945
},
{
"epoch": 2.2096548507462686,
"grad_norm": 0.43461855868836347,
"learning_rate": 3.360340185053683e-05,
"loss": 0.385,
"step": 18950
},
{
"epoch": 2.210237873134328,
"grad_norm": 0.4498995074752811,
"learning_rate": 3.3595050613659605e-05,
"loss": 0.4362,
"step": 18955
},
{
"epoch": 2.2108208955223883,
"grad_norm": 0.4538539856134972,
"learning_rate": 3.358669847050676e-05,
"loss": 0.4102,
"step": 18960
},
{
"epoch": 2.211403917910448,
"grad_norm": 0.48452969901849496,
"learning_rate": 3.3578345422320165e-05,
"loss": 0.4232,
"step": 18965
},
{
"epoch": 2.2119869402985075,
"grad_norm": 0.40585507907959584,
"learning_rate": 3.356999147034184e-05,
"loss": 0.3751,
"step": 18970
},
{
"epoch": 2.212569962686567,
"grad_norm": 0.6602423945002934,
"learning_rate": 3.3561636615813945e-05,
"loss": 0.394,
"step": 18975
},
{
"epoch": 2.2131529850746268,
"grad_norm": 0.3995041964148151,
"learning_rate": 3.355328085997876e-05,
"loss": 0.4013,
"step": 18980
},
{
"epoch": 2.2137360074626864,
"grad_norm": 0.4493928819783629,
"learning_rate": 3.3544924204078715e-05,
"loss": 0.4081,
"step": 18985
},
{
"epoch": 2.2143190298507465,
"grad_norm": 0.4599416014816642,
"learning_rate": 3.3536566649356356e-05,
"loss": 0.4008,
"step": 18990
},
{
"epoch": 2.214902052238806,
"grad_norm": 0.42168687357584556,
"learning_rate": 3.352820819705437e-05,
"loss": 0.4092,
"step": 18995
},
{
"epoch": 2.2154850746268657,
"grad_norm": 0.4315475803234669,
"learning_rate": 3.351984884841558e-05,
"loss": 0.411,
"step": 19000
},
{
"epoch": 2.2160680970149254,
"grad_norm": 0.4085084805547176,
"learning_rate": 3.351148860468295e-05,
"loss": 0.404,
"step": 19005
},
{
"epoch": 2.216651119402985,
"grad_norm": 0.4224188899426308,
"learning_rate": 3.350312746709956e-05,
"loss": 0.4209,
"step": 19010
},
{
"epoch": 2.2172341417910446,
"grad_norm": 0.3903041081792129,
"learning_rate": 3.3494765436908635e-05,
"loss": 0.3881,
"step": 19015
},
{
"epoch": 2.2178171641791047,
"grad_norm": 0.4060882490180791,
"learning_rate": 3.348640251535352e-05,
"loss": 0.4431,
"step": 19020
},
{
"epoch": 2.2184001865671643,
"grad_norm": 0.4287928617213922,
"learning_rate": 3.3478038703677714e-05,
"loss": 0.4297,
"step": 19025
},
{
"epoch": 2.218983208955224,
"grad_norm": 0.4685125464577112,
"learning_rate": 3.346967400312482e-05,
"loss": 0.416,
"step": 19030
},
{
"epoch": 2.2195662313432836,
"grad_norm": 0.4538311902780141,
"learning_rate": 3.346130841493861e-05,
"loss": 0.4115,
"step": 19035
},
{
"epoch": 2.220149253731343,
"grad_norm": 0.41675800397388363,
"learning_rate": 3.3452941940362946e-05,
"loss": 0.4098,
"step": 19040
},
{
"epoch": 2.220732276119403,
"grad_norm": 0.4115809497046314,
"learning_rate": 3.3444574580641855e-05,
"loss": 0.389,
"step": 19045
},
{
"epoch": 2.221315298507463,
"grad_norm": 0.411072064864703,
"learning_rate": 3.343620633701948e-05,
"loss": 0.3959,
"step": 19050
},
{
"epoch": 2.2218983208955225,
"grad_norm": 0.4553105626693565,
"learning_rate": 3.3427837210740096e-05,
"loss": 0.407,
"step": 19055
},
{
"epoch": 2.222481343283582,
"grad_norm": 0.45285812900354233,
"learning_rate": 3.341946720304812e-05,
"loss": 0.4268,
"step": 19060
},
{
"epoch": 2.223064365671642,
"grad_norm": 0.418828723073918,
"learning_rate": 3.341109631518809e-05,
"loss": 0.4411,
"step": 19065
},
{
"epoch": 2.2236473880597014,
"grad_norm": 0.3826327345459468,
"learning_rate": 3.340272454840466e-05,
"loss": 0.4077,
"step": 19070
},
{
"epoch": 2.224230410447761,
"grad_norm": 0.4019863082348389,
"learning_rate": 3.339435190394266e-05,
"loss": 0.3925,
"step": 19075
},
{
"epoch": 2.2248134328358207,
"grad_norm": 0.38540552264775296,
"learning_rate": 3.3385978383046996e-05,
"loss": 0.3667,
"step": 19080
},
{
"epoch": 2.2253964552238807,
"grad_norm": 0.40850546390600095,
"learning_rate": 3.337760398696275e-05,
"loss": 0.4317,
"step": 19085
},
{
"epoch": 2.2259794776119404,
"grad_norm": 0.4075709876647318,
"learning_rate": 3.336922871693509e-05,
"loss": 0.4119,
"step": 19090
},
{
"epoch": 2.2265625,
"grad_norm": 0.4676102128976199,
"learning_rate": 3.3360852574209364e-05,
"loss": 0.401,
"step": 19095
},
{
"epoch": 2.2271455223880596,
"grad_norm": 0.42331464158872506,
"learning_rate": 3.335247556003101e-05,
"loss": 0.4329,
"step": 19100
},
{
"epoch": 2.2277285447761193,
"grad_norm": 0.43176856400343006,
"learning_rate": 3.334409767564562e-05,
"loss": 0.3929,
"step": 19105
},
{
"epoch": 2.2283115671641793,
"grad_norm": 0.4546741054489476,
"learning_rate": 3.33357189222989e-05,
"loss": 0.4298,
"step": 19110
},
{
"epoch": 2.228894589552239,
"grad_norm": 0.4222722235222902,
"learning_rate": 3.3327339301236685e-05,
"loss": 0.3984,
"step": 19115
},
{
"epoch": 2.2294776119402986,
"grad_norm": 0.41156536253900705,
"learning_rate": 3.331895881370495e-05,
"loss": 0.3993,
"step": 19120
},
{
"epoch": 2.230060634328358,
"grad_norm": 0.42397863871609975,
"learning_rate": 3.33105774609498e-05,
"loss": 0.4251,
"step": 19125
},
{
"epoch": 2.230643656716418,
"grad_norm": 0.4094806636161526,
"learning_rate": 3.3302195244217435e-05,
"loss": 0.395,
"step": 19130
},
{
"epoch": 2.2312266791044775,
"grad_norm": 0.4387702217365145,
"learning_rate": 3.329381216475424e-05,
"loss": 0.4125,
"step": 19135
},
{
"epoch": 2.231809701492537,
"grad_norm": 0.44313493486655786,
"learning_rate": 3.32854282238067e-05,
"loss": 0.4234,
"step": 19140
},
{
"epoch": 2.232392723880597,
"grad_norm": 0.42304706417295024,
"learning_rate": 3.327704342262139e-05,
"loss": 0.405,
"step": 19145
},
{
"epoch": 2.232975746268657,
"grad_norm": 0.43077227644053545,
"learning_rate": 3.326865776244509e-05,
"loss": 0.4015,
"step": 19150
},
{
"epoch": 2.2335587686567164,
"grad_norm": 0.40206455569293875,
"learning_rate": 3.326027124452464e-05,
"loss": 0.4033,
"step": 19155
},
{
"epoch": 2.234141791044776,
"grad_norm": 0.3837895983092989,
"learning_rate": 3.3251883870107066e-05,
"loss": 0.386,
"step": 19160
},
{
"epoch": 2.2347248134328357,
"grad_norm": 0.44978271368359235,
"learning_rate": 3.324349564043945e-05,
"loss": 0.4055,
"step": 19165
},
{
"epoch": 2.2353078358208953,
"grad_norm": 0.47717898363254724,
"learning_rate": 3.323510655676906e-05,
"loss": 0.4109,
"step": 19170
},
{
"epoch": 2.2358908582089554,
"grad_norm": 0.44735628156790574,
"learning_rate": 3.322671662034329e-05,
"loss": 0.4376,
"step": 19175
},
{
"epoch": 2.236473880597015,
"grad_norm": 0.4502648812475679,
"learning_rate": 3.3218325832409616e-05,
"loss": 0.4336,
"step": 19180
},
{
"epoch": 2.2370569029850746,
"grad_norm": 0.4267332030835794,
"learning_rate": 3.3209934194215685e-05,
"loss": 0.3839,
"step": 19185
},
{
"epoch": 2.2376399253731343,
"grad_norm": 0.4899680668403373,
"learning_rate": 3.320154170700925e-05,
"loss": 0.4115,
"step": 19190
},
{
"epoch": 2.238222947761194,
"grad_norm": 0.4372955169554078,
"learning_rate": 3.3193148372038186e-05,
"loss": 0.4369,
"step": 19195
},
{
"epoch": 2.2388059701492535,
"grad_norm": 0.40104463684894454,
"learning_rate": 3.3184754190550506e-05,
"loss": 0.3798,
"step": 19200
},
{
"epoch": 2.2393889925373136,
"grad_norm": 0.4458017841998383,
"learning_rate": 3.317635916379434e-05,
"loss": 0.4321,
"step": 19205
},
{
"epoch": 2.2399720149253732,
"grad_norm": 0.46399213984482857,
"learning_rate": 3.316796329301796e-05,
"loss": 0.4135,
"step": 19210
},
{
"epoch": 2.240555037313433,
"grad_norm": 0.470985654242471,
"learning_rate": 3.315956657946975e-05,
"loss": 0.4332,
"step": 19215
},
{
"epoch": 2.2411380597014925,
"grad_norm": 0.4012839214934904,
"learning_rate": 3.31511690243982e-05,
"loss": 0.387,
"step": 19220
},
{
"epoch": 2.241721082089552,
"grad_norm": 0.37874214937375866,
"learning_rate": 3.314277062905197e-05,
"loss": 0.4104,
"step": 19225
},
{
"epoch": 2.2423041044776117,
"grad_norm": 0.38189244637664627,
"learning_rate": 3.3134371394679806e-05,
"loss": 0.406,
"step": 19230
},
{
"epoch": 2.242887126865672,
"grad_norm": 0.4073493151668433,
"learning_rate": 3.3125971322530605e-05,
"loss": 0.3941,
"step": 19235
},
{
"epoch": 2.2434701492537314,
"grad_norm": 0.41295329233596706,
"learning_rate": 3.3117570413853373e-05,
"loss": 0.4375,
"step": 19240
},
{
"epoch": 2.244053171641791,
"grad_norm": 0.4113591289942066,
"learning_rate": 3.3109168669897234e-05,
"loss": 0.4334,
"step": 19245
},
{
"epoch": 2.2446361940298507,
"grad_norm": 0.3948951613485122,
"learning_rate": 3.3100766091911464e-05,
"loss": 0.4223,
"step": 19250
},
{
"epoch": 2.2452192164179103,
"grad_norm": 0.47305209701104506,
"learning_rate": 3.309236268114543e-05,
"loss": 0.3993,
"step": 19255
},
{
"epoch": 2.24580223880597,
"grad_norm": 0.4063313365683372,
"learning_rate": 3.308395843884866e-05,
"loss": 0.3894,
"step": 19260
},
{
"epoch": 2.24638526119403,
"grad_norm": 0.4966297547212204,
"learning_rate": 3.3075553366270765e-05,
"loss": 0.4113,
"step": 19265
},
{
"epoch": 2.2469682835820897,
"grad_norm": 0.42660858880974023,
"learning_rate": 3.30671474646615e-05,
"loss": 0.4103,
"step": 19270
},
{
"epoch": 2.2475513059701493,
"grad_norm": 0.3921026144813095,
"learning_rate": 3.305874073527076e-05,
"loss": 0.3976,
"step": 19275
},
{
"epoch": 2.248134328358209,
"grad_norm": 0.40947245891887135,
"learning_rate": 3.305033317934852e-05,
"loss": 0.4195,
"step": 19280
},
{
"epoch": 2.2487173507462686,
"grad_norm": 0.39824547295686524,
"learning_rate": 3.304192479814492e-05,
"loss": 0.3951,
"step": 19285
},
{
"epoch": 2.249300373134328,
"grad_norm": 0.4605956512328903,
"learning_rate": 3.30335155929102e-05,
"loss": 0.4179,
"step": 19290
},
{
"epoch": 2.2498833955223883,
"grad_norm": 0.4235960354555694,
"learning_rate": 3.3025105564894726e-05,
"loss": 0.4074,
"step": 19295
},
{
"epoch": 2.250466417910448,
"grad_norm": 0.7789025397419326,
"learning_rate": 3.301669471534899e-05,
"loss": 0.4059,
"step": 19300
},
{
"epoch": 2.2510494402985075,
"grad_norm": 0.41723162788172374,
"learning_rate": 3.300828304552362e-05,
"loss": 0.4225,
"step": 19305
},
{
"epoch": 2.251632462686567,
"grad_norm": 0.4106349523920227,
"learning_rate": 3.299987055666932e-05,
"loss": 0.4052,
"step": 19310
},
{
"epoch": 2.2522154850746268,
"grad_norm": 0.4283657768425172,
"learning_rate": 3.299145725003698e-05,
"loss": 0.4132,
"step": 19315
},
{
"epoch": 2.2527985074626864,
"grad_norm": 0.42135169549612134,
"learning_rate": 3.298304312687754e-05,
"loss": 0.4057,
"step": 19320
},
{
"epoch": 2.2533815298507465,
"grad_norm": 0.4343994879957478,
"learning_rate": 3.2974628188442135e-05,
"loss": 0.4162,
"step": 19325
},
{
"epoch": 2.253964552238806,
"grad_norm": 0.38418503116395103,
"learning_rate": 3.2966212435981975e-05,
"loss": 0.4028,
"step": 19330
},
{
"epoch": 2.2545475746268657,
"grad_norm": 0.4386121996784451,
"learning_rate": 3.295779587074839e-05,
"loss": 0.4076,
"step": 19335
},
{
"epoch": 2.2551305970149254,
"grad_norm": 0.4261264480921298,
"learning_rate": 3.2949378493992854e-05,
"loss": 0.4434,
"step": 19340
},
{
"epoch": 2.255713619402985,
"grad_norm": 0.4202596274567211,
"learning_rate": 3.294096030696695e-05,
"loss": 0.4288,
"step": 19345
},
{
"epoch": 2.2562966417910446,
"grad_norm": 0.4023981550455131,
"learning_rate": 3.293254131092238e-05,
"loss": 0.3901,
"step": 19350
},
{
"epoch": 2.2568796641791042,
"grad_norm": 0.43842212014715376,
"learning_rate": 3.292412150711096e-05,
"loss": 0.4076,
"step": 19355
},
{
"epoch": 2.2574626865671643,
"grad_norm": 0.4235541619911344,
"learning_rate": 3.2915700896784655e-05,
"loss": 0.41,
"step": 19360
},
{
"epoch": 2.258045708955224,
"grad_norm": 0.3924033689792878,
"learning_rate": 3.29072794811955e-05,
"loss": 0.4163,
"step": 19365
},
{
"epoch": 2.2586287313432836,
"grad_norm": 0.5602310911698453,
"learning_rate": 3.28988572615957e-05,
"loss": 0.4159,
"step": 19370
},
{
"epoch": 2.259211753731343,
"grad_norm": 0.44134197341260173,
"learning_rate": 3.289043423923756e-05,
"loss": 0.4145,
"step": 19375
},
{
"epoch": 2.259794776119403,
"grad_norm": 0.4355846163833659,
"learning_rate": 3.288201041537348e-05,
"loss": 0.4081,
"step": 19380
},
{
"epoch": 2.260377798507463,
"grad_norm": 0.37596200221279935,
"learning_rate": 3.2873585791256026e-05,
"loss": 0.4149,
"step": 19385
},
{
"epoch": 2.2609608208955225,
"grad_norm": 0.429895697615993,
"learning_rate": 3.286516036813785e-05,
"loss": 0.4211,
"step": 19390
},
{
"epoch": 2.261543843283582,
"grad_norm": 0.4110188978590114,
"learning_rate": 3.2856734147271734e-05,
"loss": 0.4115,
"step": 19395
},
{
"epoch": 2.262126865671642,
"grad_norm": 0.4219592976595445,
"learning_rate": 3.284830712991057e-05,
"loss": 0.4149,
"step": 19400
},
{
"epoch": 2.2627098880597014,
"grad_norm": 0.3974568668430782,
"learning_rate": 3.283987931730738e-05,
"loss": 0.4092,
"step": 19405
},
{
"epoch": 2.263292910447761,
"grad_norm": 0.41645173989415424,
"learning_rate": 3.28314507107153e-05,
"loss": 0.4021,
"step": 19410
},
{
"epoch": 2.2638759328358207,
"grad_norm": 0.41635428855051715,
"learning_rate": 3.282302131138758e-05,
"loss": 0.3773,
"step": 19415
},
{
"epoch": 2.2644589552238807,
"grad_norm": 0.3923843382277778,
"learning_rate": 3.281459112057759e-05,
"loss": 0.4427,
"step": 19420
},
{
"epoch": 2.2650419776119404,
"grad_norm": 0.40199122879201077,
"learning_rate": 3.280616013953882e-05,
"loss": 0.3949,
"step": 19425
},
{
"epoch": 2.265625,
"grad_norm": 0.47769311176420254,
"learning_rate": 3.2797728369524875e-05,
"loss": 0.3995,
"step": 19430
},
{
"epoch": 2.2662080223880596,
"grad_norm": 0.4425848431188337,
"learning_rate": 3.2789295811789486e-05,
"loss": 0.4143,
"step": 19435
},
{
"epoch": 2.2667910447761193,
"grad_norm": 0.4324374383716641,
"learning_rate": 3.2780862467586486e-05,
"loss": 0.4028,
"step": 19440
},
{
"epoch": 2.2673740671641793,
"grad_norm": 0.43786139719568246,
"learning_rate": 3.2772428338169835e-05,
"loss": 0.4266,
"step": 19445
},
{
"epoch": 2.267957089552239,
"grad_norm": 0.5747528957501068,
"learning_rate": 3.2763993424793604e-05,
"loss": 0.4308,
"step": 19450
},
{
"epoch": 2.2685401119402986,
"grad_norm": 0.3870883426706857,
"learning_rate": 3.275555772871198e-05,
"loss": 0.3695,
"step": 19455
},
{
"epoch": 2.269123134328358,
"grad_norm": 0.44081356534981314,
"learning_rate": 3.2747121251179294e-05,
"loss": 0.4262,
"step": 19460
},
{
"epoch": 2.269706156716418,
"grad_norm": 4.048079343708381,
"learning_rate": 3.2738683993449937e-05,
"loss": 0.424,
"step": 19465
},
{
"epoch": 2.2702891791044775,
"grad_norm": 0.44792548660329046,
"learning_rate": 3.273024595677846e-05,
"loss": 0.4231,
"step": 19470
},
{
"epoch": 2.270872201492537,
"grad_norm": 0.4550582978354379,
"learning_rate": 3.272180714241954e-05,
"loss": 0.3964,
"step": 19475
},
{
"epoch": 2.271455223880597,
"grad_norm": 0.4093528141036578,
"learning_rate": 3.271336755162792e-05,
"loss": 0.3874,
"step": 19480
},
{
"epoch": 2.272038246268657,
"grad_norm": 0.4258837139029889,
"learning_rate": 3.270492718565848e-05,
"loss": 0.4188,
"step": 19485
},
{
"epoch": 2.2726212686567164,
"grad_norm": 0.4492579967645142,
"learning_rate": 3.269648604576625e-05,
"loss": 0.4329,
"step": 19490
},
{
"epoch": 2.273204291044776,
"grad_norm": 0.4299121285612161,
"learning_rate": 3.2688044133206336e-05,
"loss": 0.3899,
"step": 19495
},
{
"epoch": 2.2737873134328357,
"grad_norm": 0.4475759072761559,
"learning_rate": 3.267960144923397e-05,
"loss": 0.4271,
"step": 19500
},
{
"epoch": 2.2743703358208958,
"grad_norm": 0.47668206129734714,
"learning_rate": 3.267115799510448e-05,
"loss": 0.4259,
"step": 19505
},
{
"epoch": 2.2749533582089554,
"grad_norm": 0.42450594454386936,
"learning_rate": 3.266271377207335e-05,
"loss": 0.4242,
"step": 19510
},
{
"epoch": 2.275536380597015,
"grad_norm": 0.431683491675114,
"learning_rate": 3.265426878139615e-05,
"loss": 0.4076,
"step": 19515
},
{
"epoch": 2.2761194029850746,
"grad_norm": 0.43830855687851744,
"learning_rate": 3.264582302432856e-05,
"loss": 0.4024,
"step": 19520
},
{
"epoch": 2.2767024253731343,
"grad_norm": 0.4386244681679389,
"learning_rate": 3.263737650212639e-05,
"loss": 0.3781,
"step": 19525
},
{
"epoch": 2.277285447761194,
"grad_norm": 0.43014543463612237,
"learning_rate": 3.262892921604556e-05,
"loss": 0.4089,
"step": 19530
},
{
"epoch": 2.2778684701492535,
"grad_norm": 0.4343757511308402,
"learning_rate": 3.262048116734209e-05,
"loss": 0.4129,
"step": 19535
},
{
"epoch": 2.2784514925373136,
"grad_norm": 0.40237337145907276,
"learning_rate": 3.261203235727214e-05,
"loss": 0.3891,
"step": 19540
},
{
"epoch": 2.2790345149253732,
"grad_norm": 0.45091683200498744,
"learning_rate": 3.2603582787091954e-05,
"loss": 0.4146,
"step": 19545
},
{
"epoch": 2.279617537313433,
"grad_norm": 0.3974180501483524,
"learning_rate": 3.259513245805791e-05,
"loss": 0.3896,
"step": 19550
},
{
"epoch": 2.2802005597014925,
"grad_norm": 0.3903372551099371,
"learning_rate": 3.258668137142648e-05,
"loss": 0.4153,
"step": 19555
},
{
"epoch": 2.280783582089552,
"grad_norm": 0.44473419209702447,
"learning_rate": 3.2578229528454266e-05,
"loss": 0.4477,
"step": 19560
},
{
"epoch": 2.2813666044776117,
"grad_norm": 0.42338089189107436,
"learning_rate": 3.2569776930397986e-05,
"loss": 0.4107,
"step": 19565
},
{
"epoch": 2.281949626865672,
"grad_norm": 0.4271095839124686,
"learning_rate": 3.256132357851445e-05,
"loss": 0.4111,
"step": 19570
},
{
"epoch": 2.2825326492537314,
"grad_norm": 0.4240361012694304,
"learning_rate": 3.255286947406058e-05,
"loss": 0.4256,
"step": 19575
},
{
"epoch": 2.283115671641791,
"grad_norm": 0.43112816927698716,
"learning_rate": 3.254441461829344e-05,
"loss": 0.4476,
"step": 19580
},
{
"epoch": 2.2836986940298507,
"grad_norm": 0.4452325290170443,
"learning_rate": 3.253595901247017e-05,
"loss": 0.4232,
"step": 19585
},
{
"epoch": 2.2842817164179103,
"grad_norm": 0.422485076532205,
"learning_rate": 3.252750265784806e-05,
"loss": 0.3957,
"step": 19590
},
{
"epoch": 2.28486473880597,
"grad_norm": 0.44535347407515513,
"learning_rate": 3.251904555568446e-05,
"loss": 0.3979,
"step": 19595
},
{
"epoch": 2.28544776119403,
"grad_norm": 0.442577024845191,
"learning_rate": 3.251058770723688e-05,
"loss": 0.3925,
"step": 19600
},
{
"epoch": 2.2860307835820897,
"grad_norm": 0.4029571005295842,
"learning_rate": 3.2502129113762916e-05,
"loss": 0.3969,
"step": 19605
},
{
"epoch": 2.2866138059701493,
"grad_norm": 0.4409108505001848,
"learning_rate": 3.249366977652028e-05,
"loss": 0.3964,
"step": 19610
},
{
"epoch": 2.287196828358209,
"grad_norm": 0.43439371315893266,
"learning_rate": 3.248520969676679e-05,
"loss": 0.4023,
"step": 19615
},
{
"epoch": 2.2877798507462686,
"grad_norm": 0.45017403790988,
"learning_rate": 3.247674887576038e-05,
"loss": 0.4053,
"step": 19620
},
{
"epoch": 2.288362873134328,
"grad_norm": 0.41301240031394465,
"learning_rate": 3.24682873147591e-05,
"loss": 0.4034,
"step": 19625
},
{
"epoch": 2.2889458955223883,
"grad_norm": 0.43559973356243425,
"learning_rate": 3.24598250150211e-05,
"loss": 0.4009,
"step": 19630
},
{
"epoch": 2.289528917910448,
"grad_norm": 0.3955797638011971,
"learning_rate": 3.245136197780464e-05,
"loss": 0.4003,
"step": 19635
},
{
"epoch": 2.2901119402985075,
"grad_norm": 0.5378091429648244,
"learning_rate": 3.2442898204368086e-05,
"loss": 0.412,
"step": 19640
},
{
"epoch": 2.290694962686567,
"grad_norm": 0.41044291869817934,
"learning_rate": 3.243443369596994e-05,
"loss": 0.4,
"step": 19645
},
{
"epoch": 2.2912779850746268,
"grad_norm": 0.5155671788654731,
"learning_rate": 3.242596845386878e-05,
"loss": 0.4272,
"step": 19650
},
{
"epoch": 2.2918610074626864,
"grad_norm": 0.46689772618771636,
"learning_rate": 3.24175024793233e-05,
"loss": 0.4553,
"step": 19655
},
{
"epoch": 2.2924440298507465,
"grad_norm": 0.44880295534744935,
"learning_rate": 3.240903577359232e-05,
"loss": 0.3942,
"step": 19660
},
{
"epoch": 2.293027052238806,
"grad_norm": 0.4386249304117945,
"learning_rate": 3.2400568337934753e-05,
"loss": 0.4148,
"step": 19665
},
{
"epoch": 2.2936100746268657,
"grad_norm": 0.3841084140882872,
"learning_rate": 3.239210017360963e-05,
"loss": 0.4006,
"step": 19670
},
{
"epoch": 2.2941930970149254,
"grad_norm": 0.4270348488186706,
"learning_rate": 3.238363128187609e-05,
"loss": 0.4257,
"step": 19675
},
{
"epoch": 2.294776119402985,
"grad_norm": 0.48423965482498266,
"learning_rate": 3.237516166399336e-05,
"loss": 0.3961,
"step": 19680
},
{
"epoch": 2.2953591417910446,
"grad_norm": 0.39634698752788855,
"learning_rate": 3.23666913212208e-05,
"loss": 0.3969,
"step": 19685
},
{
"epoch": 2.2959421641791042,
"grad_norm": 0.45634558141168363,
"learning_rate": 3.2358220254817874e-05,
"loss": 0.3999,
"step": 19690
},
{
"epoch": 2.2965251865671643,
"grad_norm": 0.3964112052788775,
"learning_rate": 3.234974846604414e-05,
"loss": 0.3696,
"step": 19695
},
{
"epoch": 2.297108208955224,
"grad_norm": 0.4906474377698196,
"learning_rate": 3.234127595615927e-05,
"loss": 0.4101,
"step": 19700
},
{
"epoch": 2.2976912313432836,
"grad_norm": 0.4264425204370442,
"learning_rate": 3.233280272642306e-05,
"loss": 0.4013,
"step": 19705
},
{
"epoch": 2.298274253731343,
"grad_norm": 0.43759040567702656,
"learning_rate": 3.232432877809538e-05,
"loss": 0.4417,
"step": 19710
},
{
"epoch": 2.298857276119403,
"grad_norm": 0.3948352991061603,
"learning_rate": 3.231585411243625e-05,
"loss": 0.4268,
"step": 19715
},
{
"epoch": 2.299440298507463,
"grad_norm": 0.4726401560011215,
"learning_rate": 3.230737873070574e-05,
"loss": 0.4239,
"step": 19720
},
{
"epoch": 2.3000233208955225,
"grad_norm": 0.4633160273854487,
"learning_rate": 3.2298902634164076e-05,
"loss": 0.4466,
"step": 19725
},
{
"epoch": 2.300606343283582,
"grad_norm": 0.3866952226690696,
"learning_rate": 3.229042582407157e-05,
"loss": 0.4167,
"step": 19730
},
{
"epoch": 2.301189365671642,
"grad_norm": 0.4311186362441962,
"learning_rate": 3.2281948301688644e-05,
"loss": 0.4065,
"step": 19735
},
{
"epoch": 2.3017723880597014,
"grad_norm": 0.4348609745059275,
"learning_rate": 3.2273470068275816e-05,
"loss": 0.3859,
"step": 19740
},
{
"epoch": 2.302355410447761,
"grad_norm": 0.40747328514484904,
"learning_rate": 3.2264991125093734e-05,
"loss": 0.4135,
"step": 19745
},
{
"epoch": 2.3029384328358207,
"grad_norm": 0.4222247568672412,
"learning_rate": 3.225651147340312e-05,
"loss": 0.4136,
"step": 19750
},
{
"epoch": 2.3035214552238807,
"grad_norm": 0.4285394092682662,
"learning_rate": 3.224803111446482e-05,
"loss": 0.4054,
"step": 19755
},
{
"epoch": 2.3041044776119404,
"grad_norm": 0.4484351780405356,
"learning_rate": 3.223955004953979e-05,
"loss": 0.4058,
"step": 19760
},
{
"epoch": 2.3046875,
"grad_norm": 0.7772680349991993,
"learning_rate": 3.223106827988908e-05,
"loss": 0.4178,
"step": 19765
},
{
"epoch": 2.3052705223880596,
"grad_norm": 0.42053916584824075,
"learning_rate": 3.222258580677385e-05,
"loss": 0.424,
"step": 19770
},
{
"epoch": 2.3058535447761193,
"grad_norm": 0.4617068198682325,
"learning_rate": 3.221410263145534e-05,
"loss": 0.3985,
"step": 19775
},
{
"epoch": 2.3064365671641793,
"grad_norm": 0.4161972544147272,
"learning_rate": 3.220561875519495e-05,
"loss": 0.4045,
"step": 19780
},
{
"epoch": 2.307019589552239,
"grad_norm": 0.45098332375509076,
"learning_rate": 3.219713417925414e-05,
"loss": 0.3895,
"step": 19785
},
{
"epoch": 2.3076026119402986,
"grad_norm": 0.39429093395594705,
"learning_rate": 3.218864890489446e-05,
"loss": 0.4099,
"step": 19790
},
{
"epoch": 2.308185634328358,
"grad_norm": 0.4441370005884668,
"learning_rate": 3.218016293337763e-05,
"loss": 0.4148,
"step": 19795
},
{
"epoch": 2.308768656716418,
"grad_norm": 0.5063774882183846,
"learning_rate": 3.2171676265965415e-05,
"loss": 0.3917,
"step": 19800
},
{
"epoch": 2.3093516791044775,
"grad_norm": 0.41287549927129186,
"learning_rate": 3.216318890391969e-05,
"loss": 0.4159,
"step": 19805
},
{
"epoch": 2.309934701492537,
"grad_norm": 0.39958654092936824,
"learning_rate": 3.2154700848502454e-05,
"loss": 0.3951,
"step": 19810
},
{
"epoch": 2.310517723880597,
"grad_norm": 0.4457276425526031,
"learning_rate": 3.21462121009758e-05,
"loss": 0.3834,
"step": 19815
},
{
"epoch": 2.311100746268657,
"grad_norm": 0.3999870840682877,
"learning_rate": 3.2137722662601934e-05,
"loss": 0.3951,
"step": 19820
},
{
"epoch": 2.3116837686567164,
"grad_norm": 0.45784680226701713,
"learning_rate": 3.212923253464314e-05,
"loss": 0.4181,
"step": 19825
},
{
"epoch": 2.312266791044776,
"grad_norm": 0.4197518751738806,
"learning_rate": 3.212074171836181e-05,
"loss": 0.4343,
"step": 19830
},
{
"epoch": 2.3128498134328357,
"grad_norm": 0.42816680566531606,
"learning_rate": 3.211225021502047e-05,
"loss": 0.3756,
"step": 19835
},
{
"epoch": 2.3134328358208958,
"grad_norm": 0.40880001600894555,
"learning_rate": 3.21037580258817e-05,
"loss": 0.3867,
"step": 19840
},
{
"epoch": 2.3140158582089554,
"grad_norm": 0.3973850915227583,
"learning_rate": 3.209526515220824e-05,
"loss": 0.4087,
"step": 19845
},
{
"epoch": 2.314598880597015,
"grad_norm": 0.43025167101520595,
"learning_rate": 3.208677159526287e-05,
"loss": 0.4039,
"step": 19850
},
{
"epoch": 2.3151819029850746,
"grad_norm": 0.43292192466058066,
"learning_rate": 3.207827735630851e-05,
"loss": 0.4189,
"step": 19855
},
{
"epoch": 2.3157649253731343,
"grad_norm": 0.3870544571777672,
"learning_rate": 3.206978243660817e-05,
"loss": 0.3742,
"step": 19860
},
{
"epoch": 2.316347947761194,
"grad_norm": 0.40163379164131247,
"learning_rate": 3.206128683742497e-05,
"loss": 0.381,
"step": 19865
},
{
"epoch": 2.3169309701492535,
"grad_norm": 0.436758508652368,
"learning_rate": 3.205279056002212e-05,
"loss": 0.4012,
"step": 19870
},
{
"epoch": 2.3175139925373136,
"grad_norm": 0.39860721757852535,
"learning_rate": 3.204429360566293e-05,
"loss": 0.3733,
"step": 19875
},
{
"epoch": 2.3180970149253732,
"grad_norm": 0.42656335971174303,
"learning_rate": 3.203579597561082e-05,
"loss": 0.395,
"step": 19880
},
{
"epoch": 2.318680037313433,
"grad_norm": 0.4497528973413777,
"learning_rate": 3.202729767112931e-05,
"loss": 0.3933,
"step": 19885
},
{
"epoch": 2.3192630597014925,
"grad_norm": 0.48962635772772106,
"learning_rate": 3.2018798693482015e-05,
"loss": 0.4005,
"step": 19890
},
{
"epoch": 2.319846082089552,
"grad_norm": 0.43087445360519433,
"learning_rate": 3.201029904393265e-05,
"loss": 0.3916,
"step": 19895
},
{
"epoch": 2.3204291044776117,
"grad_norm": 0.4290395291230286,
"learning_rate": 3.200179872374503e-05,
"loss": 0.4111,
"step": 19900
},
{
"epoch": 2.321012126865672,
"grad_norm": 0.43822200286103785,
"learning_rate": 3.199329773418307e-05,
"loss": 0.4117,
"step": 19905
},
{
"epoch": 2.3215951492537314,
"grad_norm": 0.4313418370656167,
"learning_rate": 3.198479607651079e-05,
"loss": 0.4198,
"step": 19910
},
{
"epoch": 2.322178171641791,
"grad_norm": 0.4505798881150213,
"learning_rate": 3.1976293751992295e-05,
"loss": 0.4264,
"step": 19915
},
{
"epoch": 2.3227611940298507,
"grad_norm": 0.4342457919802341,
"learning_rate": 3.1967790761891826e-05,
"loss": 0.3948,
"step": 19920
},
{
"epoch": 2.3233442164179103,
"grad_norm": 0.43034316120632954,
"learning_rate": 3.195928710747366e-05,
"loss": 0.3945,
"step": 19925
},
{
"epoch": 2.32392723880597,
"grad_norm": 0.3794488846030449,
"learning_rate": 3.1950782790002236e-05,
"loss": 0.3983,
"step": 19930
},
{
"epoch": 2.32451026119403,
"grad_norm": 0.43556043158438223,
"learning_rate": 3.194227781074205e-05,
"loss": 0.4076,
"step": 19935
},
{
"epoch": 2.3250932835820897,
"grad_norm": 0.42871794072011205,
"learning_rate": 3.1933772170957716e-05,
"loss": 0.404,
"step": 19940
},
{
"epoch": 2.3256763059701493,
"grad_norm": 0.44074119013671076,
"learning_rate": 3.192526587191395e-05,
"loss": 0.4035,
"step": 19945
},
{
"epoch": 2.326259328358209,
"grad_norm": 0.4359971713746372,
"learning_rate": 3.191675891487554e-05,
"loss": 0.3908,
"step": 19950
},
{
"epoch": 2.3268423507462686,
"grad_norm": 0.45675933115702533,
"learning_rate": 3.190825130110739e-05,
"loss": 0.3975,
"step": 19955
},
{
"epoch": 2.327425373134328,
"grad_norm": 0.4129999570499641,
"learning_rate": 3.189974303187452e-05,
"loss": 0.3941,
"step": 19960
},
{
"epoch": 2.3280083955223883,
"grad_norm": 0.473532918140509,
"learning_rate": 3.1891234108442007e-05,
"loss": 0.3914,
"step": 19965
},
{
"epoch": 2.328591417910448,
"grad_norm": 0.3960313726920244,
"learning_rate": 3.188272453207507e-05,
"loss": 0.3835,
"step": 19970
},
{
"epoch": 2.3291744402985075,
"grad_norm": 0.4139472126774143,
"learning_rate": 3.1874214304038965e-05,
"loss": 0.3835,
"step": 19975
},
{
"epoch": 2.329757462686567,
"grad_norm": 0.41494669031781867,
"learning_rate": 3.186570342559912e-05,
"loss": 0.4275,
"step": 19980
},
{
"epoch": 2.3303404850746268,
"grad_norm": 0.4178039006108638,
"learning_rate": 3.1857191898020996e-05,
"loss": 0.3807,
"step": 19985
},
{
"epoch": 2.3309235074626864,
"grad_norm": 0.44004003162212,
"learning_rate": 3.184867972257019e-05,
"loss": 0.4094,
"step": 19990
},
{
"epoch": 2.3315065298507465,
"grad_norm": 0.4388893382580093,
"learning_rate": 3.184016690051237e-05,
"loss": 0.4354,
"step": 19995
},
{
"epoch": 2.332089552238806,
"grad_norm": 0.4123871451644767,
"learning_rate": 3.1831653433113317e-05,
"loss": 0.3927,
"step": 20000
},
{
"epoch": 2.3326725746268657,
"grad_norm": 0.5086366333897813,
"learning_rate": 3.18231393216389e-05,
"loss": 0.4365,
"step": 20005
},
{
"epoch": 2.3332555970149254,
"grad_norm": 0.4244222110279215,
"learning_rate": 3.1814624567355087e-05,
"loss": 0.4105,
"step": 20010
},
{
"epoch": 2.333838619402985,
"grad_norm": 0.483139716666001,
"learning_rate": 3.180610917152795e-05,
"loss": 0.4058,
"step": 20015
},
{
"epoch": 2.3344216417910446,
"grad_norm": 0.47751784102029826,
"learning_rate": 3.179759313542362e-05,
"loss": 0.4326,
"step": 20020
},
{
"epoch": 2.3350046641791042,
"grad_norm": 0.4530366603227296,
"learning_rate": 3.178907646030838e-05,
"loss": 0.427,
"step": 20025
},
{
"epoch": 2.3355876865671643,
"grad_norm": 0.4131765323720608,
"learning_rate": 3.1780559147448554e-05,
"loss": 0.3869,
"step": 20030
},
{
"epoch": 2.336170708955224,
"grad_norm": 0.3909592535404698,
"learning_rate": 3.1772041198110604e-05,
"loss": 0.4155,
"step": 20035
},
{
"epoch": 2.3367537313432836,
"grad_norm": 0.392337753019418,
"learning_rate": 3.176352261356105e-05,
"loss": 0.403,
"step": 20040
},
{
"epoch": 2.337336753731343,
"grad_norm": 0.40748285662078076,
"learning_rate": 3.1755003395066546e-05,
"loss": 0.3931,
"step": 20045
},
{
"epoch": 2.337919776119403,
"grad_norm": 0.4215948719992,
"learning_rate": 3.17464835438938e-05,
"loss": 0.3992,
"step": 20050
},
{
"epoch": 2.338502798507463,
"grad_norm": 0.4107600451026482,
"learning_rate": 3.173796306130963e-05,
"loss": 0.385,
"step": 20055
},
{
"epoch": 2.3390858208955225,
"grad_norm": 0.4112967006629501,
"learning_rate": 3.172944194858096e-05,
"loss": 0.3813,
"step": 20060
},
{
"epoch": 2.339668843283582,
"grad_norm": 0.4050618756862867,
"learning_rate": 3.17209202069748e-05,
"loss": 0.4194,
"step": 20065
},
{
"epoch": 2.340251865671642,
"grad_norm": 0.4395181870541969,
"learning_rate": 3.171239783775825e-05,
"loss": 0.4291,
"step": 20070
},
{
"epoch": 2.3408348880597014,
"grad_norm": 0.41947129455782817,
"learning_rate": 3.170387484219849e-05,
"loss": 0.4266,
"step": 20075
},
{
"epoch": 2.341417910447761,
"grad_norm": 0.4102958164069878,
"learning_rate": 3.169535122156283e-05,
"loss": 0.4082,
"step": 20080
},
{
"epoch": 2.3420009328358207,
"grad_norm": 0.4671712493198971,
"learning_rate": 3.1686826977118635e-05,
"loss": 0.4312,
"step": 20085
},
{
"epoch": 2.3425839552238807,
"grad_norm": 0.4552248865176427,
"learning_rate": 3.167830211013338e-05,
"loss": 0.3904,
"step": 20090
},
{
"epoch": 2.3431669776119404,
"grad_norm": 0.4303843947549127,
"learning_rate": 3.166977662187464e-05,
"loss": 0.3861,
"step": 20095
},
{
"epoch": 2.34375,
"grad_norm": 0.4309237307936687,
"learning_rate": 3.166125051361007e-05,
"loss": 0.3873,
"step": 20100
},
{
"epoch": 2.3443330223880596,
"grad_norm": 0.3994652686476947,
"learning_rate": 3.165272378660741e-05,
"loss": 0.3848,
"step": 20105
},
{
"epoch": 2.3449160447761193,
"grad_norm": 0.4284802061767505,
"learning_rate": 3.164419644213451e-05,
"loss": 0.4149,
"step": 20110
},
{
"epoch": 2.3454990671641793,
"grad_norm": 0.42777765319194583,
"learning_rate": 3.1635668481459306e-05,
"loss": 0.4066,
"step": 20115
},
{
"epoch": 2.346082089552239,
"grad_norm": 0.44010404715579654,
"learning_rate": 3.162713990584983e-05,
"loss": 0.425,
"step": 20120
},
{
"epoch": 2.3466651119402986,
"grad_norm": 0.4205388245648607,
"learning_rate": 3.1618610716574196e-05,
"loss": 0.4062,
"step": 20125
},
{
"epoch": 2.347248134328358,
"grad_norm": 0.437076758831132,
"learning_rate": 3.1610080914900604e-05,
"loss": 0.4409,
"step": 20130
},
{
"epoch": 2.347831156716418,
"grad_norm": 0.4623130929754608,
"learning_rate": 3.1601550502097356e-05,
"loss": 0.4162,
"step": 20135
},
{
"epoch": 2.3484141791044775,
"grad_norm": 0.40550332345951196,
"learning_rate": 3.159301947943285e-05,
"loss": 0.3974,
"step": 20140
},
{
"epoch": 2.348997201492537,
"grad_norm": 0.427271702948713,
"learning_rate": 3.158448784817556e-05,
"loss": 0.4172,
"step": 20145
},
{
"epoch": 2.349580223880597,
"grad_norm": 0.406859620580034,
"learning_rate": 3.157595560959407e-05,
"loss": 0.3981,
"step": 20150
},
{
"epoch": 2.350163246268657,
"grad_norm": 0.39776364849500506,
"learning_rate": 3.156742276495702e-05,
"loss": 0.3899,
"step": 20155
},
{
"epoch": 2.3507462686567164,
"grad_norm": 0.4510281901026045,
"learning_rate": 3.155888931553319e-05,
"loss": 0.4406,
"step": 20160
},
{
"epoch": 2.351329291044776,
"grad_norm": 0.4365965942805441,
"learning_rate": 3.155035526259139e-05,
"loss": 0.3817,
"step": 20165
},
{
"epoch": 2.3519123134328357,
"grad_norm": 0.4545022679422716,
"learning_rate": 3.154182060740058e-05,
"loss": 0.4329,
"step": 20170
},
{
"epoch": 2.3524953358208958,
"grad_norm": 0.3934986770325854,
"learning_rate": 3.153328535122978e-05,
"loss": 0.3961,
"step": 20175
},
{
"epoch": 2.3530783582089554,
"grad_norm": 0.4282677145322305,
"learning_rate": 3.152474949534808e-05,
"loss": 0.4105,
"step": 20180
},
{
"epoch": 2.353661380597015,
"grad_norm": 0.38176053687568606,
"learning_rate": 3.15162130410247e-05,
"loss": 0.3885,
"step": 20185
},
{
"epoch": 2.3542444029850746,
"grad_norm": 0.4681680530843617,
"learning_rate": 3.1507675989528915e-05,
"loss": 0.4028,
"step": 20190
},
{
"epoch": 2.3548274253731343,
"grad_norm": 0.43310488195784286,
"learning_rate": 3.1499138342130114e-05,
"loss": 0.3849,
"step": 20195
},
{
"epoch": 2.355410447761194,
"grad_norm": 0.39181491044361416,
"learning_rate": 3.1490600100097746e-05,
"loss": 0.3933,
"step": 20200
},
{
"epoch": 2.3559934701492535,
"grad_norm": 0.3984123549110816,
"learning_rate": 3.148206126470138e-05,
"loss": 0.3849,
"step": 20205
},
{
"epoch": 2.3565764925373136,
"grad_norm": 0.4185418729161508,
"learning_rate": 3.147352183721067e-05,
"loss": 0.4053,
"step": 20210
},
{
"epoch": 2.3571595149253732,
"grad_norm": 0.4153936523137314,
"learning_rate": 3.1464981818895325e-05,
"loss": 0.4045,
"step": 20215
},
{
"epoch": 2.357742537313433,
"grad_norm": 0.4215852959550067,
"learning_rate": 3.145644121102517e-05,
"loss": 0.4319,
"step": 20220
},
{
"epoch": 2.3583255597014925,
"grad_norm": 0.3932563933434031,
"learning_rate": 3.1447900014870125e-05,
"loss": 0.4,
"step": 20225
},
{
"epoch": 2.358908582089552,
"grad_norm": 0.42232225989119876,
"learning_rate": 3.1439358231700165e-05,
"loss": 0.4123,
"step": 20230
},
{
"epoch": 2.3594916044776117,
"grad_norm": 0.38271574851859447,
"learning_rate": 3.143081586278539e-05,
"loss": 0.4145,
"step": 20235
},
{
"epoch": 2.360074626865672,
"grad_norm": 0.39680852459169036,
"learning_rate": 3.142227290939595e-05,
"loss": 0.4048,
"step": 20240
},
{
"epoch": 2.3606576492537314,
"grad_norm": 0.3908631021160493,
"learning_rate": 3.14137293728021e-05,
"loss": 0.3967,
"step": 20245
},
{
"epoch": 2.361240671641791,
"grad_norm": 0.38489036732200466,
"learning_rate": 3.14051852542742e-05,
"loss": 0.3806,
"step": 20250
},
{
"epoch": 2.3618236940298507,
"grad_norm": 0.414997237373693,
"learning_rate": 3.1396640555082665e-05,
"loss": 0.4315,
"step": 20255
},
{
"epoch": 2.3624067164179103,
"grad_norm": 0.399401219366884,
"learning_rate": 3.1388095276498013e-05,
"loss": 0.4272,
"step": 20260
},
{
"epoch": 2.36298973880597,
"grad_norm": 0.43657244502649967,
"learning_rate": 3.137954941979085e-05,
"loss": 0.3999,
"step": 20265
},
{
"epoch": 2.36357276119403,
"grad_norm": 0.5097443478752106,
"learning_rate": 3.1371002986231855e-05,
"loss": 0.3874,
"step": 20270
},
{
"epoch": 2.3641557835820897,
"grad_norm": 0.4216243846227505,
"learning_rate": 3.136245597709181e-05,
"loss": 0.4382,
"step": 20275
},
{
"epoch": 2.3647388059701493,
"grad_norm": 0.3834158761183634,
"learning_rate": 3.1353908393641574e-05,
"loss": 0.3912,
"step": 20280
},
{
"epoch": 2.365321828358209,
"grad_norm": 0.4730070704366648,
"learning_rate": 3.134536023715207e-05,
"loss": 0.4071,
"step": 20285
},
{
"epoch": 2.3659048507462686,
"grad_norm": 0.42336283969259897,
"learning_rate": 3.133681150889434e-05,
"loss": 0.4033,
"step": 20290
},
{
"epoch": 2.366487873134328,
"grad_norm": 0.41487489597089183,
"learning_rate": 3.1328262210139515e-05,
"loss": 0.4041,
"step": 20295
},
{
"epoch": 2.3670708955223883,
"grad_norm": 0.4143196669283807,
"learning_rate": 3.131971234215877e-05,
"loss": 0.4204,
"step": 20300
},
{
"epoch": 2.367653917910448,
"grad_norm": 0.4192282635638447,
"learning_rate": 3.13111619062234e-05,
"loss": 0.4249,
"step": 20305
},
{
"epoch": 2.3682369402985075,
"grad_norm": 0.4281030980622985,
"learning_rate": 3.1302610903604775e-05,
"loss": 0.3837,
"step": 20310
},
{
"epoch": 2.368819962686567,
"grad_norm": 0.42012529862171527,
"learning_rate": 3.129405933557433e-05,
"loss": 0.4065,
"step": 20315
},
{
"epoch": 2.3694029850746268,
"grad_norm": 0.4280964987224729,
"learning_rate": 3.128550720340362e-05,
"loss": 0.4199,
"step": 20320
},
{
"epoch": 2.3699860074626864,
"grad_norm": 0.40572184861344535,
"learning_rate": 3.127695450836426e-05,
"loss": 0.386,
"step": 20325
},
{
"epoch": 2.3705690298507465,
"grad_norm": 0.3987095331284861,
"learning_rate": 3.126840125172795e-05,
"loss": 0.4256,
"step": 20330
},
{
"epoch": 2.371152052238806,
"grad_norm": 0.4041729121660031,
"learning_rate": 3.125984743476648e-05,
"loss": 0.4079,
"step": 20335
},
{
"epoch": 2.3717350746268657,
"grad_norm": 0.40515378758058834,
"learning_rate": 3.125129305875172e-05,
"loss": 0.4276,
"step": 20340
},
{
"epoch": 2.3723180970149254,
"grad_norm": 0.4367449436443719,
"learning_rate": 3.1242738124955624e-05,
"loss": 0.4252,
"step": 20345
},
{
"epoch": 2.372901119402985,
"grad_norm": 0.4232215775395212,
"learning_rate": 3.1234182634650234e-05,
"loss": 0.4135,
"step": 20350
},
{
"epoch": 2.3734841417910446,
"grad_norm": 0.49297842768262384,
"learning_rate": 3.122562658910765e-05,
"loss": 0.4111,
"step": 20355
},
{
"epoch": 2.3740671641791042,
"grad_norm": 0.4097608023873178,
"learning_rate": 3.1217069989600097e-05,
"loss": 0.4213,
"step": 20360
},
{
"epoch": 2.3746501865671643,
"grad_norm": 0.4085027008391505,
"learning_rate": 3.1208512837399856e-05,
"loss": 0.3976,
"step": 20365
},
{
"epoch": 2.375233208955224,
"grad_norm": 0.4185490070440442,
"learning_rate": 3.119995513377928e-05,
"loss": 0.4022,
"step": 20370
},
{
"epoch": 2.3758162313432836,
"grad_norm": 0.3782251945072346,
"learning_rate": 3.119139688001082e-05,
"loss": 0.4073,
"step": 20375
},
{
"epoch": 2.376399253731343,
"grad_norm": 0.43857261590896524,
"learning_rate": 3.118283807736703e-05,
"loss": 0.4336,
"step": 20380
},
{
"epoch": 2.376982276119403,
"grad_norm": 0.4625141637612915,
"learning_rate": 3.1174278727120496e-05,
"loss": 0.405,
"step": 20385
},
{
"epoch": 2.377565298507463,
"grad_norm": 0.4151689515085908,
"learning_rate": 3.1165718830543914e-05,
"loss": 0.4239,
"step": 20390
},
{
"epoch": 2.3781483208955225,
"grad_norm": 0.41081737455149997,
"learning_rate": 3.115715838891007e-05,
"loss": 0.4171,
"step": 20395
},
{
"epoch": 2.378731343283582,
"grad_norm": 0.4750578098368241,
"learning_rate": 3.1148597403491816e-05,
"loss": 0.4378,
"step": 20400
},
{
"epoch": 2.379314365671642,
"grad_norm": 0.4357376291208132,
"learning_rate": 3.114003587556208e-05,
"loss": 0.4245,
"step": 20405
},
{
"epoch": 2.3798973880597014,
"grad_norm": 0.4011573642351387,
"learning_rate": 3.1131473806393876e-05,
"loss": 0.3972,
"step": 20410
},
{
"epoch": 2.380480410447761,
"grad_norm": 0.397300079143492,
"learning_rate": 3.112291119726032e-05,
"loss": 0.4254,
"step": 20415
},
{
"epoch": 2.3810634328358207,
"grad_norm": 0.4069344376856726,
"learning_rate": 3.1114348049434583e-05,
"loss": 0.3848,
"step": 20420
},
{
"epoch": 2.3816464552238807,
"grad_norm": 0.3974674651436214,
"learning_rate": 3.110578436418992e-05,
"loss": 0.3926,
"step": 20425
},
{
"epoch": 2.3822294776119404,
"grad_norm": 0.4521801060123563,
"learning_rate": 3.109722014279967e-05,
"loss": 0.4047,
"step": 20430
},
{
"epoch": 2.3828125,
"grad_norm": 0.39620834393615184,
"learning_rate": 3.108865538653725e-05,
"loss": 0.3763,
"step": 20435
},
{
"epoch": 2.3833955223880596,
"grad_norm": 0.4379716188799694,
"learning_rate": 3.108009009667615e-05,
"loss": 0.3971,
"step": 20440
},
{
"epoch": 2.3839785447761193,
"grad_norm": 0.40168351093725196,
"learning_rate": 3.1071524274489966e-05,
"loss": 0.4279,
"step": 20445
},
{
"epoch": 2.3845615671641793,
"grad_norm": 0.4470968221721542,
"learning_rate": 3.106295792125233e-05,
"loss": 0.4263,
"step": 20450
},
{
"epoch": 2.385144589552239,
"grad_norm": 0.45990971240797274,
"learning_rate": 3.1054391038237e-05,
"loss": 0.4429,
"step": 20455
},
{
"epoch": 2.3857276119402986,
"grad_norm": 0.42201629816681707,
"learning_rate": 3.104582362671778e-05,
"loss": 0.4311,
"step": 20460
},
{
"epoch": 2.386310634328358,
"grad_norm": 0.3751079860934871,
"learning_rate": 3.103725568796854e-05,
"loss": 0.388,
"step": 20465
},
{
"epoch": 2.386893656716418,
"grad_norm": 0.4332472386759332,
"learning_rate": 3.102868722326328e-05,
"loss": 0.4032,
"step": 20470
},
{
"epoch": 2.3874766791044775,
"grad_norm": 0.45049407808338693,
"learning_rate": 3.102011823387605e-05,
"loss": 0.401,
"step": 20475
},
{
"epoch": 2.388059701492537,
"grad_norm": 0.4071153872700178,
"learning_rate": 3.1011548721080955e-05,
"loss": 0.4025,
"step": 20480
},
{
"epoch": 2.388642723880597,
"grad_norm": 0.4776890821807428,
"learning_rate": 3.100297868615222e-05,
"loss": 0.4094,
"step": 20485
},
{
"epoch": 2.389225746268657,
"grad_norm": 0.42967822194216654,
"learning_rate": 3.099440813036411e-05,
"loss": 0.4142,
"step": 20490
},
{
"epoch": 2.3898087686567164,
"grad_norm": 0.39970921355549716,
"learning_rate": 3.0985837054990983e-05,
"loss": 0.4014,
"step": 20495
},
{
"epoch": 2.390391791044776,
"grad_norm": 0.40907692527245637,
"learning_rate": 3.097726546130729e-05,
"loss": 0.3832,
"step": 20500
},
{
"epoch": 2.3909748134328357,
"grad_norm": 0.42182387212132105,
"learning_rate": 3.096869335058755e-05,
"loss": 0.4126,
"step": 20505
},
{
"epoch": 2.3915578358208958,
"grad_norm": 0.3907488142464014,
"learning_rate": 3.096012072410633e-05,
"loss": 0.4078,
"step": 20510
},
{
"epoch": 2.3921408582089554,
"grad_norm": 0.42905064578366786,
"learning_rate": 3.095154758313831e-05,
"loss": 0.4209,
"step": 20515
},
{
"epoch": 2.392723880597015,
"grad_norm": 0.42243047199337486,
"learning_rate": 3.094297392895825e-05,
"loss": 0.3888,
"step": 20520
},
{
"epoch": 2.3933069029850746,
"grad_norm": 0.45036621292222845,
"learning_rate": 3.093439976284094e-05,
"loss": 0.4333,
"step": 20525
},
{
"epoch": 2.3938899253731343,
"grad_norm": 0.4137249071571011,
"learning_rate": 3.0925825086061295e-05,
"loss": 0.4103,
"step": 20530
},
{
"epoch": 2.394472947761194,
"grad_norm": 0.463925441043269,
"learning_rate": 3.0917249899894285e-05,
"loss": 0.403,
"step": 20535
},
{
"epoch": 2.3950559701492535,
"grad_norm": 0.44466459696270055,
"learning_rate": 3.090867420561495e-05,
"loss": 0.4264,
"step": 20540
},
{
"epoch": 2.3956389925373136,
"grad_norm": 0.4789706272314867,
"learning_rate": 3.090009800449842e-05,
"loss": 0.4343,
"step": 20545
},
{
"epoch": 2.3962220149253732,
"grad_norm": 0.4677081585453051,
"learning_rate": 3.0891521297819906e-05,
"loss": 0.4168,
"step": 20550
},
{
"epoch": 2.396805037313433,
"grad_norm": 0.4394275863034301,
"learning_rate": 3.088294408685466e-05,
"loss": 0.4149,
"step": 20555
},
{
"epoch": 2.3973880597014925,
"grad_norm": 0.44940923406140465,
"learning_rate": 3.0874366372878036e-05,
"loss": 0.4437,
"step": 20560
},
{
"epoch": 2.397971082089552,
"grad_norm": 0.4048519685878374,
"learning_rate": 3.086578815716548e-05,
"loss": 0.4055,
"step": 20565
},
{
"epoch": 2.3985541044776117,
"grad_norm": 0.39322982180209676,
"learning_rate": 3.085720944099246e-05,
"loss": 0.3879,
"step": 20570
},
{
"epoch": 2.399137126865672,
"grad_norm": 0.4292387757119189,
"learning_rate": 3.0848630225634564e-05,
"loss": 0.3966,
"step": 20575
},
{
"epoch": 2.3997201492537314,
"grad_norm": 0.39519598752823326,
"learning_rate": 3.0840050512367444e-05,
"loss": 0.3793,
"step": 20580
},
{
"epoch": 2.400303171641791,
"grad_norm": 0.4441287436004875,
"learning_rate": 3.08314703024668e-05,
"loss": 0.39,
"step": 20585
},
{
"epoch": 2.4008861940298507,
"grad_norm": 0.3926175895055523,
"learning_rate": 3.082288959720845e-05,
"loss": 0.4133,
"step": 20590
},
{
"epoch": 2.4014692164179103,
"grad_norm": 0.452675435761681,
"learning_rate": 3.081430839786825e-05,
"loss": 0.4103,
"step": 20595
},
{
"epoch": 2.40205223880597,
"grad_norm": 0.4211404487727711,
"learning_rate": 3.0805726705722156e-05,
"loss": 0.42,
"step": 20600
},
{
"epoch": 2.40263526119403,
"grad_norm": 0.42564064678254915,
"learning_rate": 3.079714452204617e-05,
"loss": 0.3877,
"step": 20605
},
{
"epoch": 2.4032182835820897,
"grad_norm": 0.3929680554251429,
"learning_rate": 3.078856184811638e-05,
"loss": 0.3891,
"step": 20610
},
{
"epoch": 2.4038013059701493,
"grad_norm": 0.4530599295376277,
"learning_rate": 3.0779978685208956e-05,
"loss": 0.3993,
"step": 20615
},
{
"epoch": 2.404384328358209,
"grad_norm": 0.4036391934769732,
"learning_rate": 3.077139503460012e-05,
"loss": 0.4197,
"step": 20620
},
{
"epoch": 2.4049673507462686,
"grad_norm": 0.40603035443045027,
"learning_rate": 3.0762810897566184e-05,
"loss": 0.3807,
"step": 20625
},
{
"epoch": 2.405550373134328,
"grad_norm": 0.48643919266582186,
"learning_rate": 3.0754226275383546e-05,
"loss": 0.4486,
"step": 20630
},
{
"epoch": 2.4061333955223883,
"grad_norm": 0.44987864557727897,
"learning_rate": 3.0745641169328627e-05,
"loss": 0.4137,
"step": 20635
},
{
"epoch": 2.406716417910448,
"grad_norm": 0.3967160264721621,
"learning_rate": 3.073705558067797e-05,
"loss": 0.4116,
"step": 20640
},
{
"epoch": 2.4072994402985075,
"grad_norm": 0.43400882918751754,
"learning_rate": 3.072846951070816e-05,
"loss": 0.4211,
"step": 20645
},
{
"epoch": 2.407882462686567,
"grad_norm": 0.43332322513148747,
"learning_rate": 3.071988296069586e-05,
"loss": 0.4159,
"step": 20650
},
{
"epoch": 2.4084654850746268,
"grad_norm": 0.40809729263624933,
"learning_rate": 3.071129593191783e-05,
"loss": 0.4201,
"step": 20655
},
{
"epoch": 2.4090485074626864,
"grad_norm": 0.441248257372105,
"learning_rate": 3.070270842565084e-05,
"loss": 0.4051,
"step": 20660
},
{
"epoch": 2.4096315298507465,
"grad_norm": 0.4013710020223029,
"learning_rate": 3.069412044317181e-05,
"loss": 0.3836,
"step": 20665
},
{
"epoch": 2.410214552238806,
"grad_norm": 0.41433068535226436,
"learning_rate": 3.068553198575767e-05,
"loss": 0.4328,
"step": 20670
},
{
"epoch": 2.4107975746268657,
"grad_norm": 0.4493111978580927,
"learning_rate": 3.0676943054685445e-05,
"loss": 0.3994,
"step": 20675
},
{
"epoch": 2.4113805970149254,
"grad_norm": 0.46036109843736306,
"learning_rate": 3.0668353651232226e-05,
"loss": 0.417,
"step": 20680
},
{
"epoch": 2.411963619402985,
"grad_norm": 0.41046467832975453,
"learning_rate": 3.065976377667517e-05,
"loss": 0.3921,
"step": 20685
},
{
"epoch": 2.4125466417910446,
"grad_norm": 0.41393769625874877,
"learning_rate": 3.065117343229153e-05,
"loss": 0.3999,
"step": 20690
},
{
"epoch": 2.4131296641791042,
"grad_norm": 0.4683255889770547,
"learning_rate": 3.0642582619358576e-05,
"loss": 0.4093,
"step": 20695
},
{
"epoch": 2.4137126865671643,
"grad_norm": 0.43977471510009114,
"learning_rate": 3.063399133915371e-05,
"loss": 0.4025,
"step": 20700
},
{
"epoch": 2.414295708955224,
"grad_norm": 0.41258514848682676,
"learning_rate": 3.0625399592954346e-05,
"loss": 0.3768,
"step": 20705
},
{
"epoch": 2.4148787313432836,
"grad_norm": 0.4084008696090954,
"learning_rate": 3.0616807382038016e-05,
"loss": 0.3983,
"step": 20710
},
{
"epoch": 2.415461753731343,
"grad_norm": 0.4795776151542908,
"learning_rate": 3.0608214707682286e-05,
"loss": 0.3944,
"step": 20715
},
{
"epoch": 2.416044776119403,
"grad_norm": 0.43468743305130253,
"learning_rate": 3.059962157116481e-05,
"loss": 0.3776,
"step": 20720
},
{
"epoch": 2.416627798507463,
"grad_norm": 0.39826732371846674,
"learning_rate": 3.059102797376331e-05,
"loss": 0.4066,
"step": 20725
},
{
"epoch": 2.4172108208955225,
"grad_norm": 0.4738603577348725,
"learning_rate": 3.058243391675557e-05,
"loss": 0.4203,
"step": 20730
},
{
"epoch": 2.417793843283582,
"grad_norm": 0.4621376024557958,
"learning_rate": 3.0573839401419426e-05,
"loss": 0.4172,
"step": 20735
},
{
"epoch": 2.418376865671642,
"grad_norm": 0.4307459637540691,
"learning_rate": 3.056524442903282e-05,
"loss": 0.4082,
"step": 20740
},
{
"epoch": 2.4189598880597014,
"grad_norm": 0.41965207734947385,
"learning_rate": 3.055664900087374e-05,
"loss": 0.4192,
"step": 20745
},
{
"epoch": 2.419542910447761,
"grad_norm": 0.4079039339608424,
"learning_rate": 3.054805311822023e-05,
"loss": 0.4219,
"step": 20750
},
{
"epoch": 2.4201259328358207,
"grad_norm": 0.4156154188703871,
"learning_rate": 3.0539456782350436e-05,
"loss": 0.3912,
"step": 20755
},
{
"epoch": 2.4207089552238807,
"grad_norm": 0.37832895205507644,
"learning_rate": 3.053085999454254e-05,
"loss": 0.4041,
"step": 20760
},
{
"epoch": 2.4212919776119404,
"grad_norm": 0.4387467174365725,
"learning_rate": 3.0522262756074796e-05,
"loss": 0.4251,
"step": 20765
},
{
"epoch": 2.421875,
"grad_norm": 0.4026709072983937,
"learning_rate": 3.051366506822554e-05,
"loss": 0.4225,
"step": 20770
},
{
"epoch": 2.4224580223880596,
"grad_norm": 0.399075731089508,
"learning_rate": 3.0505066932273157e-05,
"loss": 0.3929,
"step": 20775
},
{
"epoch": 2.4230410447761193,
"grad_norm": 0.4104673891549445,
"learning_rate": 3.0496468349496115e-05,
"loss": 0.4161,
"step": 20780
},
{
"epoch": 2.4236240671641793,
"grad_norm": 0.44960082229866133,
"learning_rate": 3.0487869321172947e-05,
"loss": 0.4242,
"step": 20785
},
{
"epoch": 2.424207089552239,
"grad_norm": 0.46803264518164983,
"learning_rate": 3.047926984858223e-05,
"loss": 0.4248,
"step": 20790
},
{
"epoch": 2.4247901119402986,
"grad_norm": 0.4458071023091181,
"learning_rate": 3.047066993300264e-05,
"loss": 0.4182,
"step": 20795
},
{
"epoch": 2.425373134328358,
"grad_norm": 0.4146868850445553,
"learning_rate": 3.046206957571288e-05,
"loss": 0.3912,
"step": 20800
},
{
"epoch": 2.425956156716418,
"grad_norm": 0.39914570066824695,
"learning_rate": 3.0453468777991768e-05,
"loss": 0.4008,
"step": 20805
},
{
"epoch": 2.4265391791044775,
"grad_norm": 0.4182645339921344,
"learning_rate": 3.0444867541118145e-05,
"loss": 0.4193,
"step": 20810
},
{
"epoch": 2.427122201492537,
"grad_norm": 0.4320660646550807,
"learning_rate": 3.0436265866370922e-05,
"loss": 0.4025,
"step": 20815
},
{
"epoch": 2.427705223880597,
"grad_norm": 0.4129958951278084,
"learning_rate": 3.0427663755029108e-05,
"loss": 0.3835,
"step": 20820
},
{
"epoch": 2.428288246268657,
"grad_norm": 0.4526823821048915,
"learning_rate": 3.041906120837174e-05,
"loss": 0.4081,
"step": 20825
},
{
"epoch": 2.4288712686567164,
"grad_norm": 0.43042241753095895,
"learning_rate": 3.0410458227677934e-05,
"loss": 0.4141,
"step": 20830
},
{
"epoch": 2.429454291044776,
"grad_norm": 0.4220190332194512,
"learning_rate": 3.040185481422689e-05,
"loss": 0.4066,
"step": 20835
},
{
"epoch": 2.4300373134328357,
"grad_norm": 0.4289775459163234,
"learning_rate": 3.0393250969297826e-05,
"loss": 0.4104,
"step": 20840
},
{
"epoch": 2.4306203358208958,
"grad_norm": 0.4199362912881698,
"learning_rate": 3.0384646694170073e-05,
"loss": 0.4292,
"step": 20845
},
{
"epoch": 2.4312033582089554,
"grad_norm": 0.41385623867411475,
"learning_rate": 3.0376041990122983e-05,
"loss": 0.404,
"step": 20850
},
{
"epoch": 2.431786380597015,
"grad_norm": 0.40814181771152663,
"learning_rate": 3.036743685843601e-05,
"loss": 0.439,
"step": 20855
},
{
"epoch": 2.4323694029850746,
"grad_norm": 0.4162244955391998,
"learning_rate": 3.0358831300388657e-05,
"loss": 0.4025,
"step": 20860
},
{
"epoch": 2.4329524253731343,
"grad_norm": 0.4072332592261159,
"learning_rate": 3.035022531726047e-05,
"loss": 0.3916,
"step": 20865
},
{
"epoch": 2.433535447761194,
"grad_norm": 0.4274678052433978,
"learning_rate": 3.0341618910331093e-05,
"loss": 0.407,
"step": 20870
},
{
"epoch": 2.4341184701492535,
"grad_norm": 0.4094881177576757,
"learning_rate": 3.0333012080880207e-05,
"loss": 0.3993,
"step": 20875
},
{
"epoch": 2.4347014925373136,
"grad_norm": 0.4313901078848929,
"learning_rate": 3.0324404830187564e-05,
"loss": 0.42,
"step": 20880
},
{
"epoch": 2.4352845149253732,
"grad_norm": 0.4549308035397354,
"learning_rate": 3.0315797159532995e-05,
"loss": 0.4303,
"step": 20885
},
{
"epoch": 2.435867537313433,
"grad_norm": 0.40035506096441525,
"learning_rate": 3.0307189070196358e-05,
"loss": 0.3914,
"step": 20890
},
{
"epoch": 2.4364505597014925,
"grad_norm": 0.39996703863221555,
"learning_rate": 3.0298580563457606e-05,
"loss": 0.3742,
"step": 20895
},
{
"epoch": 2.437033582089552,
"grad_norm": 0.4314370923694186,
"learning_rate": 3.0289971640596737e-05,
"loss": 0.409,
"step": 20900
},
{
"epoch": 2.4376166044776117,
"grad_norm": 0.42296438302299455,
"learning_rate": 3.0281362302893822e-05,
"loss": 0.4043,
"step": 20905
},
{
"epoch": 2.438199626865672,
"grad_norm": 0.43148416696910497,
"learning_rate": 3.0272752551628975e-05,
"loss": 0.3851,
"step": 20910
},
{
"epoch": 2.4387826492537314,
"grad_norm": 0.4305457157633837,
"learning_rate": 3.026414238808239e-05,
"loss": 0.3934,
"step": 20915
},
{
"epoch": 2.439365671641791,
"grad_norm": 0.4237369702223498,
"learning_rate": 3.0255531813534322e-05,
"loss": 0.4,
"step": 20920
},
{
"epoch": 2.4399486940298507,
"grad_norm": 0.44021761828619876,
"learning_rate": 3.0246920829265067e-05,
"loss": 0.4381,
"step": 20925
},
{
"epoch": 2.4405317164179103,
"grad_norm": 0.4255673618813955,
"learning_rate": 3.0238309436555e-05,
"loss": 0.4051,
"step": 20930
},
{
"epoch": 2.44111473880597,
"grad_norm": 0.39474819727816296,
"learning_rate": 3.0229697636684568e-05,
"loss": 0.389,
"step": 20935
},
{
"epoch": 2.44169776119403,
"grad_norm": 0.4081173436729887,
"learning_rate": 3.022108543093425e-05,
"loss": 0.4282,
"step": 20940
},
{
"epoch": 2.4422807835820897,
"grad_norm": 0.44756554485405237,
"learning_rate": 3.0212472820584587e-05,
"loss": 0.3982,
"step": 20945
},
{
"epoch": 2.4428638059701493,
"grad_norm": 0.4379530237401978,
"learning_rate": 3.020385980691621e-05,
"loss": 0.4291,
"step": 20950
},
{
"epoch": 2.443446828358209,
"grad_norm": 0.38755292941765784,
"learning_rate": 3.019524639120979e-05,
"loss": 0.4263,
"step": 20955
},
{
"epoch": 2.4440298507462686,
"grad_norm": 0.4410023360525595,
"learning_rate": 3.0186632574746055e-05,
"loss": 0.3812,
"step": 20960
},
{
"epoch": 2.444612873134328,
"grad_norm": 0.41835013471968363,
"learning_rate": 3.0178018358805793e-05,
"loss": 0.3962,
"step": 20965
},
{
"epoch": 2.4451958955223883,
"grad_norm": 0.4288347118406861,
"learning_rate": 3.016940374466986e-05,
"loss": 0.407,
"step": 20970
},
{
"epoch": 2.445778917910448,
"grad_norm": 0.40810092040353635,
"learning_rate": 3.0160788733619167e-05,
"loss": 0.409,
"step": 20975
},
{
"epoch": 2.4463619402985075,
"grad_norm": 0.38803359970739254,
"learning_rate": 3.0152173326934692e-05,
"loss": 0.4163,
"step": 20980
},
{
"epoch": 2.446944962686567,
"grad_norm": 0.44456045122082916,
"learning_rate": 3.0143557525897444e-05,
"loss": 0.4199,
"step": 20985
},
{
"epoch": 2.4475279850746268,
"grad_norm": 0.5765165346297749,
"learning_rate": 3.0134941331788525e-05,
"loss": 0.3912,
"step": 20990
},
{
"epoch": 2.4481110074626864,
"grad_norm": 0.39749837135133814,
"learning_rate": 3.0126324745889067e-05,
"loss": 0.3937,
"step": 20995
},
{
"epoch": 2.4486940298507465,
"grad_norm": 0.41658626971422624,
"learning_rate": 3.0117707769480285e-05,
"loss": 0.3956,
"step": 21000
},
{
"epoch": 2.449277052238806,
"grad_norm": 0.4174717512204487,
"learning_rate": 3.0109090403843448e-05,
"loss": 0.3994,
"step": 21005
},
{
"epoch": 2.4498600746268657,
"grad_norm": 0.3668055516956151,
"learning_rate": 3.0100472650259866e-05,
"loss": 0.383,
"step": 21010
},
{
"epoch": 2.4504430970149254,
"grad_norm": 0.46167332370755626,
"learning_rate": 3.0091854510010907e-05,
"loss": 0.4617,
"step": 21015
},
{
"epoch": 2.451026119402985,
"grad_norm": 0.41320660992640296,
"learning_rate": 3.008323598437802e-05,
"loss": 0.3861,
"step": 21020
},
{
"epoch": 2.4516091417910446,
"grad_norm": 0.4173103635473449,
"learning_rate": 3.0074617074642693e-05,
"loss": 0.4062,
"step": 21025
},
{
"epoch": 2.4521921641791042,
"grad_norm": 0.415988371828538,
"learning_rate": 3.006599778208647e-05,
"loss": 0.3943,
"step": 21030
},
{
"epoch": 2.4527751865671643,
"grad_norm": 0.43831470089612806,
"learning_rate": 3.005737810799097e-05,
"loss": 0.3961,
"step": 21035
},
{
"epoch": 2.453358208955224,
"grad_norm": 0.44906331098002017,
"learning_rate": 3.0048758053637844e-05,
"loss": 0.4064,
"step": 21040
},
{
"epoch": 2.4539412313432836,
"grad_norm": 0.39363363505423504,
"learning_rate": 3.0040137620308812e-05,
"loss": 0.4004,
"step": 21045
},
{
"epoch": 2.454524253731343,
"grad_norm": 0.4259684359070059,
"learning_rate": 3.0031516809285658e-05,
"loss": 0.4237,
"step": 21050
},
{
"epoch": 2.455107276119403,
"grad_norm": 0.4536016468944285,
"learning_rate": 3.0022895621850207e-05,
"loss": 0.4114,
"step": 21055
},
{
"epoch": 2.455690298507463,
"grad_norm": 0.43620568966654055,
"learning_rate": 3.001427405928435e-05,
"loss": 0.4059,
"step": 21060
},
{
"epoch": 2.4562733208955225,
"grad_norm": 0.4021623529310093,
"learning_rate": 3.0005652122870032e-05,
"loss": 0.4138,
"step": 21065
},
{
"epoch": 2.456856343283582,
"grad_norm": 0.441213535522327,
"learning_rate": 2.999702981388925e-05,
"loss": 0.4089,
"step": 21070
},
{
"epoch": 2.457439365671642,
"grad_norm": 0.46302774269362845,
"learning_rate": 2.9988407133624057e-05,
"loss": 0.4504,
"step": 21075
},
{
"epoch": 2.4580223880597014,
"grad_norm": 0.40340865628322886,
"learning_rate": 2.9979784083356567e-05,
"loss": 0.3934,
"step": 21080
},
{
"epoch": 2.458605410447761,
"grad_norm": 0.4825397672970459,
"learning_rate": 2.9971160664368946e-05,
"loss": 0.4174,
"step": 21085
},
{
"epoch": 2.4591884328358207,
"grad_norm": 0.4433655216883402,
"learning_rate": 2.996253687794341e-05,
"loss": 0.4306,
"step": 21090
},
{
"epoch": 2.4597714552238807,
"grad_norm": 0.3525179960536248,
"learning_rate": 2.9953912725362225e-05,
"loss": 0.3879,
"step": 21095
},
{
"epoch": 2.4603544776119404,
"grad_norm": 0.45390894009834903,
"learning_rate": 2.994528820790774e-05,
"loss": 0.4135,
"step": 21100
},
{
"epoch": 2.4609375,
"grad_norm": 0.4805990267234585,
"learning_rate": 2.9936663326862323e-05,
"loss": 0.4009,
"step": 21105
},
{
"epoch": 2.4615205223880596,
"grad_norm": 0.45127157246378136,
"learning_rate": 2.9928038083508415e-05,
"loss": 0.4159,
"step": 21110
},
{
"epoch": 2.4621035447761193,
"grad_norm": 0.3990167290637741,
"learning_rate": 2.9919412479128513e-05,
"loss": 0.4213,
"step": 21115
},
{
"epoch": 2.4626865671641793,
"grad_norm": 0.4237021041873273,
"learning_rate": 2.9910786515005146e-05,
"loss": 0.3822,
"step": 21120
},
{
"epoch": 2.463269589552239,
"grad_norm": 0.4116002355791405,
"learning_rate": 2.990216019242093e-05,
"loss": 0.4163,
"step": 21125
},
{
"epoch": 2.4638526119402986,
"grad_norm": 0.43391250160781775,
"learning_rate": 2.9893533512658507e-05,
"loss": 0.4002,
"step": 21130
},
{
"epoch": 2.464435634328358,
"grad_norm": 0.40413378811255374,
"learning_rate": 2.988490647700058e-05,
"loss": 0.4124,
"step": 21135
},
{
"epoch": 2.465018656716418,
"grad_norm": 0.38966204506682905,
"learning_rate": 2.987627908672992e-05,
"loss": 0.3958,
"step": 21140
},
{
"epoch": 2.4656016791044775,
"grad_norm": 0.4629556379149426,
"learning_rate": 2.9867651343129315e-05,
"loss": 0.4225,
"step": 21145
},
{
"epoch": 2.466184701492537,
"grad_norm": 0.394750285059733,
"learning_rate": 2.9859023247481644e-05,
"loss": 0.4123,
"step": 21150
},
{
"epoch": 2.466767723880597,
"grad_norm": 0.3900887470754676,
"learning_rate": 2.985039480106982e-05,
"loss": 0.3894,
"step": 21155
},
{
"epoch": 2.467350746268657,
"grad_norm": 0.44726068748955106,
"learning_rate": 2.9841766005176808e-05,
"loss": 0.4169,
"step": 21160
},
{
"epoch": 2.4679337686567164,
"grad_norm": 0.40366600935594904,
"learning_rate": 2.983313686108563e-05,
"loss": 0.4198,
"step": 21165
},
{
"epoch": 2.468516791044776,
"grad_norm": 0.4251537162507513,
"learning_rate": 2.982450737007935e-05,
"loss": 0.4009,
"step": 21170
},
{
"epoch": 2.4690998134328357,
"grad_norm": 0.413453973855533,
"learning_rate": 2.9815877533441107e-05,
"loss": 0.3987,
"step": 21175
},
{
"epoch": 2.4696828358208958,
"grad_norm": 0.4164854573390885,
"learning_rate": 2.9807247352454055e-05,
"loss": 0.3866,
"step": 21180
},
{
"epoch": 2.4702658582089554,
"grad_norm": 0.42380988885977333,
"learning_rate": 2.9798616828401428e-05,
"loss": 0.4139,
"step": 21185
},
{
"epoch": 2.470848880597015,
"grad_norm": 0.3943042370713538,
"learning_rate": 2.9789985962566503e-05,
"loss": 0.4144,
"step": 21190
},
{
"epoch": 2.4714319029850746,
"grad_norm": 0.43740794486958573,
"learning_rate": 2.9781354756232604e-05,
"loss": 0.4135,
"step": 21195
},
{
"epoch": 2.4720149253731343,
"grad_norm": 0.41163586037059013,
"learning_rate": 2.977272321068311e-05,
"loss": 0.3922,
"step": 21200
},
{
"epoch": 2.472597947761194,
"grad_norm": 0.4222108796080354,
"learning_rate": 2.9764091327201456e-05,
"loss": 0.3814,
"step": 21205
},
{
"epoch": 2.4731809701492535,
"grad_norm": 0.4245857820065939,
"learning_rate": 2.975545910707111e-05,
"loss": 0.3924,
"step": 21210
},
{
"epoch": 2.4737639925373136,
"grad_norm": 0.468031541203365,
"learning_rate": 2.9746826551575606e-05,
"loss": 0.4196,
"step": 21215
},
{
"epoch": 2.4743470149253732,
"grad_norm": 0.47749102565899015,
"learning_rate": 2.9738193661998526e-05,
"loss": 0.4182,
"step": 21220
},
{
"epoch": 2.474930037313433,
"grad_norm": 0.41588474180728796,
"learning_rate": 2.9729560439623484e-05,
"loss": 0.3983,
"step": 21225
},
{
"epoch": 2.4755130597014925,
"grad_norm": 0.4155839342233435,
"learning_rate": 2.9720926885734167e-05,
"loss": 0.4143,
"step": 21230
},
{
"epoch": 2.476096082089552,
"grad_norm": 0.4085486152492274,
"learning_rate": 2.97122930016143e-05,
"loss": 0.3968,
"step": 21235
},
{
"epoch": 2.4766791044776117,
"grad_norm": 0.4216718598522093,
"learning_rate": 2.9703658788547674e-05,
"loss": 0.3694,
"step": 21240
},
{
"epoch": 2.477262126865672,
"grad_norm": 0.4102876666531931,
"learning_rate": 2.9695024247818088e-05,
"loss": 0.42,
"step": 21245
},
{
"epoch": 2.4778451492537314,
"grad_norm": 0.40350132209784795,
"learning_rate": 2.968638938070942e-05,
"loss": 0.4171,
"step": 21250
},
{
"epoch": 2.478428171641791,
"grad_norm": 0.40859983213586976,
"learning_rate": 2.9677754188505614e-05,
"loss": 0.4041,
"step": 21255
},
{
"epoch": 2.4790111940298507,
"grad_norm": 0.3986772690785753,
"learning_rate": 2.9669118672490627e-05,
"loss": 0.395,
"step": 21260
},
{
"epoch": 2.4795942164179103,
"grad_norm": 0.46926280945104276,
"learning_rate": 2.9660482833948466e-05,
"loss": 0.4449,
"step": 21265
},
{
"epoch": 2.48017723880597,
"grad_norm": 0.42229644993874677,
"learning_rate": 2.9651846674163208e-05,
"loss": 0.4274,
"step": 21270
},
{
"epoch": 2.48076026119403,
"grad_norm": 0.4801079403961941,
"learning_rate": 2.964321019441898e-05,
"loss": 0.4282,
"step": 21275
},
{
"epoch": 2.4813432835820897,
"grad_norm": 0.4190103918754208,
"learning_rate": 2.9634573395999916e-05,
"loss": 0.4421,
"step": 21280
},
{
"epoch": 2.4819263059701493,
"grad_norm": 0.3943065446933593,
"learning_rate": 2.962593628019024e-05,
"loss": 0.3797,
"step": 21285
},
{
"epoch": 2.482509328358209,
"grad_norm": 0.5370051602481157,
"learning_rate": 2.9617298848274223e-05,
"loss": 0.4006,
"step": 21290
},
{
"epoch": 2.4830923507462686,
"grad_norm": 0.7458422677624169,
"learning_rate": 2.960866110153614e-05,
"loss": 0.4127,
"step": 21295
},
{
"epoch": 2.483675373134328,
"grad_norm": 0.3882811930877767,
"learning_rate": 2.9600023041260355e-05,
"loss": 0.3721,
"step": 21300
},
{
"epoch": 2.4842583955223883,
"grad_norm": 0.44801344023130407,
"learning_rate": 2.9591384668731264e-05,
"loss": 0.402,
"step": 21305
},
{
"epoch": 2.484841417910448,
"grad_norm": 0.4531884106552282,
"learning_rate": 2.9582745985233312e-05,
"loss": 0.43,
"step": 21310
},
{
"epoch": 2.4854244402985075,
"grad_norm": 0.38787451045843185,
"learning_rate": 2.9574106992050993e-05,
"loss": 0.4139,
"step": 21315
},
{
"epoch": 2.486007462686567,
"grad_norm": 0.4231218200420361,
"learning_rate": 2.9565467690468834e-05,
"loss": 0.3824,
"step": 21320
},
{
"epoch": 2.4865904850746268,
"grad_norm": 0.4236931079009935,
"learning_rate": 2.9556828081771413e-05,
"loss": 0.3923,
"step": 21325
},
{
"epoch": 2.4871735074626864,
"grad_norm": 0.41850927778702807,
"learning_rate": 2.9548188167243372e-05,
"loss": 0.3824,
"step": 21330
},
{
"epoch": 2.4877565298507465,
"grad_norm": 0.4570955094695527,
"learning_rate": 2.953954794816937e-05,
"loss": 0.4263,
"step": 21335
},
{
"epoch": 2.488339552238806,
"grad_norm": 0.41066886065497704,
"learning_rate": 2.953090742583413e-05,
"loss": 0.3996,
"step": 21340
},
{
"epoch": 2.4889225746268657,
"grad_norm": 0.4023001356178296,
"learning_rate": 2.952226660152242e-05,
"loss": 0.4348,
"step": 21345
},
{
"epoch": 2.4895055970149254,
"grad_norm": 0.37656867274528566,
"learning_rate": 2.951362547651903e-05,
"loss": 0.418,
"step": 21350
},
{
"epoch": 2.490088619402985,
"grad_norm": 0.4609871369769133,
"learning_rate": 2.950498405210883e-05,
"loss": 0.4038,
"step": 21355
},
{
"epoch": 2.4906716417910446,
"grad_norm": 0.40441805010449483,
"learning_rate": 2.949634232957671e-05,
"loss": 0.4375,
"step": 21360
},
{
"epoch": 2.4912546641791042,
"grad_norm": 0.3979780416205592,
"learning_rate": 2.9487700310207618e-05,
"loss": 0.4188,
"step": 21365
},
{
"epoch": 2.4918376865671643,
"grad_norm": 0.41771420923256114,
"learning_rate": 2.9479057995286528e-05,
"loss": 0.4149,
"step": 21370
},
{
"epoch": 2.492420708955224,
"grad_norm": 0.5541303106519052,
"learning_rate": 2.947041538609848e-05,
"loss": 0.385,
"step": 21375
},
{
"epoch": 2.4930037313432836,
"grad_norm": 0.4486629116161178,
"learning_rate": 2.9461772483928547e-05,
"loss": 0.411,
"step": 21380
},
{
"epoch": 2.493586753731343,
"grad_norm": 0.3995070406681024,
"learning_rate": 2.9453129290061832e-05,
"loss": 0.4122,
"step": 21385
},
{
"epoch": 2.494169776119403,
"grad_norm": 0.4204055617537802,
"learning_rate": 2.944448580578351e-05,
"loss": 0.408,
"step": 21390
},
{
"epoch": 2.494752798507463,
"grad_norm": 0.40183549257717155,
"learning_rate": 2.9435842032378778e-05,
"loss": 0.3994,
"step": 21395
},
{
"epoch": 2.4953358208955225,
"grad_norm": 0.4664775080112823,
"learning_rate": 2.9427197971132886e-05,
"loss": 0.4167,
"step": 21400
},
{
"epoch": 2.495918843283582,
"grad_norm": 0.4184877981207242,
"learning_rate": 2.941855362333112e-05,
"loss": 0.396,
"step": 21405
},
{
"epoch": 2.496501865671642,
"grad_norm": 0.41181796305763557,
"learning_rate": 2.9409908990258812e-05,
"loss": 0.4087,
"step": 21410
},
{
"epoch": 2.4970848880597014,
"grad_norm": 0.430701979025966,
"learning_rate": 2.9401264073201333e-05,
"loss": 0.4299,
"step": 21415
},
{
"epoch": 2.497667910447761,
"grad_norm": 0.4027967505927966,
"learning_rate": 2.9392618873444112e-05,
"loss": 0.39,
"step": 21420
},
{
"epoch": 2.4982509328358207,
"grad_norm": 0.4355200232474978,
"learning_rate": 2.938397339227259e-05,
"loss": 0.4288,
"step": 21425
},
{
"epoch": 2.4988339552238807,
"grad_norm": 0.4619928752591043,
"learning_rate": 2.937532763097227e-05,
"loss": 0.4215,
"step": 21430
},
{
"epoch": 2.4994169776119404,
"grad_norm": 0.4095081145525622,
"learning_rate": 2.936668159082871e-05,
"loss": 0.3808,
"step": 21435
},
{
"epoch": 2.5,
"grad_norm": 0.4215106073434386,
"learning_rate": 2.9358035273127483e-05,
"loss": 0.3971,
"step": 21440
},
{
"epoch": 2.5005830223880596,
"grad_norm": 0.38973346358103333,
"learning_rate": 2.9349388679154206e-05,
"loss": 0.3973,
"step": 21445
},
{
"epoch": 2.5011660447761193,
"grad_norm": 0.41835149099010577,
"learning_rate": 2.934074181019455e-05,
"loss": 0.3942,
"step": 21450
},
{
"epoch": 2.5017490671641793,
"grad_norm": 0.4046609572605303,
"learning_rate": 2.9332094667534238e-05,
"loss": 0.3975,
"step": 21455
},
{
"epoch": 2.502332089552239,
"grad_norm": 0.39363303197131777,
"learning_rate": 2.9323447252458986e-05,
"loss": 0.4111,
"step": 21460
},
{
"epoch": 2.5029151119402986,
"grad_norm": 0.4110521497065824,
"learning_rate": 2.9314799566254603e-05,
"loss": 0.4108,
"step": 21465
},
{
"epoch": 2.503498134328358,
"grad_norm": 0.43182190112430263,
"learning_rate": 2.9306151610206916e-05,
"loss": 0.4267,
"step": 21470
},
{
"epoch": 2.504081156716418,
"grad_norm": 0.4301973632379895,
"learning_rate": 2.9297503385601788e-05,
"loss": 0.3874,
"step": 21475
},
{
"epoch": 2.5046641791044775,
"grad_norm": 0.4293896023715753,
"learning_rate": 2.9288854893725128e-05,
"loss": 0.4208,
"step": 21480
},
{
"epoch": 2.505247201492537,
"grad_norm": 0.4416509369399998,
"learning_rate": 2.928020613586288e-05,
"loss": 0.4042,
"step": 21485
},
{
"epoch": 2.505830223880597,
"grad_norm": 0.41110271954851013,
"learning_rate": 2.9271557113301047e-05,
"loss": 0.4046,
"step": 21490
},
{
"epoch": 2.506413246268657,
"grad_norm": 0.4247547878930185,
"learning_rate": 2.9262907827325638e-05,
"loss": 0.4093,
"step": 21495
},
{
"epoch": 2.5069962686567164,
"grad_norm": 0.46742801105554627,
"learning_rate": 2.9254258279222724e-05,
"loss": 0.4068,
"step": 21500
},
{
"epoch": 2.507579291044776,
"grad_norm": 0.43897513187606063,
"learning_rate": 2.9245608470278417e-05,
"loss": 0.3935,
"step": 21505
},
{
"epoch": 2.5081623134328357,
"grad_norm": 0.3855522137341133,
"learning_rate": 2.9236958401778854e-05,
"loss": 0.4267,
"step": 21510
},
{
"epoch": 2.5087453358208958,
"grad_norm": 0.4411823143289941,
"learning_rate": 2.9228308075010213e-05,
"loss": 0.441,
"step": 21515
},
{
"epoch": 2.5093283582089554,
"grad_norm": 0.4196947934879396,
"learning_rate": 2.921965749125873e-05,
"loss": 0.4117,
"step": 21520
},
{
"epoch": 2.509911380597015,
"grad_norm": 0.44916711873645215,
"learning_rate": 2.9211006651810645e-05,
"loss": 0.4058,
"step": 21525
},
{
"epoch": 2.5104944029850746,
"grad_norm": 0.46830549267752497,
"learning_rate": 2.920235555795227e-05,
"loss": 0.4155,
"step": 21530
},
{
"epoch": 2.5110774253731343,
"grad_norm": 0.4322880218878378,
"learning_rate": 2.919370421096993e-05,
"loss": 0.3937,
"step": 21535
},
{
"epoch": 2.511660447761194,
"grad_norm": 0.41689030667715465,
"learning_rate": 2.9185052612150004e-05,
"loss": 0.4028,
"step": 21540
},
{
"epoch": 2.5122434701492535,
"grad_norm": 0.44283367320919953,
"learning_rate": 2.9176400762778906e-05,
"loss": 0.4324,
"step": 21545
},
{
"epoch": 2.5128264925373136,
"grad_norm": 0.4462371229480639,
"learning_rate": 2.9167748664143067e-05,
"loss": 0.3823,
"step": 21550
},
{
"epoch": 2.5134095149253732,
"grad_norm": 0.36489194518166884,
"learning_rate": 2.9159096317528985e-05,
"loss": 0.3788,
"step": 21555
},
{
"epoch": 2.513992537313433,
"grad_norm": 0.5003846722632077,
"learning_rate": 2.9150443724223174e-05,
"loss": 0.4258,
"step": 21560
},
{
"epoch": 2.5145755597014925,
"grad_norm": 0.4205153403610952,
"learning_rate": 2.91417908855122e-05,
"loss": 0.4135,
"step": 21565
},
{
"epoch": 2.515158582089552,
"grad_norm": 0.4036636588531879,
"learning_rate": 2.9133137802682646e-05,
"loss": 0.3993,
"step": 21570
},
{
"epoch": 2.515741604477612,
"grad_norm": 0.41459542944198186,
"learning_rate": 2.912448447702115e-05,
"loss": 0.4058,
"step": 21575
},
{
"epoch": 2.5163246268656714,
"grad_norm": 0.40454346028708793,
"learning_rate": 2.9115830909814374e-05,
"loss": 0.4187,
"step": 21580
},
{
"epoch": 2.5169076492537314,
"grad_norm": 0.4227574492714799,
"learning_rate": 2.9107177102349026e-05,
"loss": 0.412,
"step": 21585
},
{
"epoch": 2.517490671641791,
"grad_norm": 0.3820073360239675,
"learning_rate": 2.909852305591184e-05,
"loss": 0.3882,
"step": 21590
},
{
"epoch": 2.5180736940298507,
"grad_norm": 0.41820926089204913,
"learning_rate": 2.9089868771789598e-05,
"loss": 0.4028,
"step": 21595
},
{
"epoch": 2.5186567164179103,
"grad_norm": 0.4368346520661654,
"learning_rate": 2.9081214251269095e-05,
"loss": 0.4418,
"step": 21600
},
{
"epoch": 2.51923973880597,
"grad_norm": 0.5258252316148309,
"learning_rate": 2.9072559495637187e-05,
"loss": 0.4111,
"step": 21605
},
{
"epoch": 2.51982276119403,
"grad_norm": 0.4398525033401105,
"learning_rate": 2.9063904506180746e-05,
"loss": 0.4026,
"step": 21610
},
{
"epoch": 2.5204057835820897,
"grad_norm": 0.4429162914251529,
"learning_rate": 2.9055249284186686e-05,
"loss": 0.4085,
"step": 21615
},
{
"epoch": 2.5209888059701493,
"grad_norm": 0.42203105398839463,
"learning_rate": 2.904659383094197e-05,
"loss": 0.4344,
"step": 21620
},
{
"epoch": 2.521571828358209,
"grad_norm": 0.40620566810256226,
"learning_rate": 2.9037938147733557e-05,
"loss": 0.3826,
"step": 21625
},
{
"epoch": 2.5221548507462686,
"grad_norm": 0.41829004160654043,
"learning_rate": 2.902928223584848e-05,
"loss": 0.4044,
"step": 21630
},
{
"epoch": 2.5227378731343286,
"grad_norm": 0.42124429434039856,
"learning_rate": 2.9020626096573793e-05,
"loss": 0.4184,
"step": 21635
},
{
"epoch": 2.523320895522388,
"grad_norm": 0.41292177384683004,
"learning_rate": 2.9011969731196565e-05,
"loss": 0.3945,
"step": 21640
},
{
"epoch": 2.523903917910448,
"grad_norm": 0.42538107659879604,
"learning_rate": 2.9003313141003934e-05,
"loss": 0.3956,
"step": 21645
},
{
"epoch": 2.5244869402985075,
"grad_norm": 0.4263449207400961,
"learning_rate": 2.8994656327283036e-05,
"loss": 0.388,
"step": 21650
},
{
"epoch": 2.525069962686567,
"grad_norm": 0.43570400848495505,
"learning_rate": 2.898599929132107e-05,
"loss": 0.4091,
"step": 21655
},
{
"epoch": 2.5256529850746268,
"grad_norm": 0.3685028513907922,
"learning_rate": 2.897734203440524e-05,
"loss": 0.3859,
"step": 21660
},
{
"epoch": 2.5262360074626864,
"grad_norm": 0.42144367982032993,
"learning_rate": 2.8968684557822806e-05,
"loss": 0.3989,
"step": 21665
},
{
"epoch": 2.5268190298507465,
"grad_norm": 0.42879147975007814,
"learning_rate": 2.8960026862861057e-05,
"loss": 0.416,
"step": 21670
},
{
"epoch": 2.527402052238806,
"grad_norm": 0.42972426861921337,
"learning_rate": 2.8951368950807288e-05,
"loss": 0.416,
"step": 21675
},
{
"epoch": 2.5279850746268657,
"grad_norm": 0.41098597858844427,
"learning_rate": 2.894271082294887e-05,
"loss": 0.4061,
"step": 21680
},
{
"epoch": 2.5285680970149254,
"grad_norm": 0.41669927137715934,
"learning_rate": 2.8934052480573175e-05,
"loss": 0.3919,
"step": 21685
},
{
"epoch": 2.529151119402985,
"grad_norm": 0.5075636109195275,
"learning_rate": 2.8925393924967615e-05,
"loss": 0.3878,
"step": 21690
},
{
"epoch": 2.5297341417910446,
"grad_norm": 0.43172849798857615,
"learning_rate": 2.891673515741964e-05,
"loss": 0.4003,
"step": 21695
},
{
"epoch": 2.5303171641791042,
"grad_norm": 0.4138559841501898,
"learning_rate": 2.8908076179216715e-05,
"loss": 0.398,
"step": 21700
},
{
"epoch": 2.5309001865671643,
"grad_norm": 0.4433376163763771,
"learning_rate": 2.8899416991646354e-05,
"loss": 0.4245,
"step": 21705
},
{
"epoch": 2.531483208955224,
"grad_norm": 0.4345974572000521,
"learning_rate": 2.88907575959961e-05,
"loss": 0.3999,
"step": 21710
},
{
"epoch": 2.5320662313432836,
"grad_norm": 0.40454268596395815,
"learning_rate": 2.8882097993553504e-05,
"loss": 0.41,
"step": 21715
},
{
"epoch": 2.532649253731343,
"grad_norm": 0.3980572891511395,
"learning_rate": 2.8873438185606194e-05,
"loss": 0.403,
"step": 21720
},
{
"epoch": 2.533232276119403,
"grad_norm": 0.42317910525564006,
"learning_rate": 2.8864778173441775e-05,
"loss": 0.3967,
"step": 21725
},
{
"epoch": 2.533815298507463,
"grad_norm": 0.4563297646016866,
"learning_rate": 2.8856117958347923e-05,
"loss": 0.4272,
"step": 21730
},
{
"epoch": 2.5343983208955225,
"grad_norm": 0.4018884227123111,
"learning_rate": 2.884745754161232e-05,
"loss": 0.4066,
"step": 21735
},
{
"epoch": 2.534981343283582,
"grad_norm": 0.4242066161997057,
"learning_rate": 2.8838796924522694e-05,
"loss": 0.3932,
"step": 21740
},
{
"epoch": 2.535564365671642,
"grad_norm": 0.401542307992814,
"learning_rate": 2.883013610836679e-05,
"loss": 0.4204,
"step": 21745
},
{
"epoch": 2.5361473880597014,
"grad_norm": 0.4216314766589857,
"learning_rate": 2.8821475094432393e-05,
"loss": 0.4115,
"step": 21750
},
{
"epoch": 2.536730410447761,
"grad_norm": 0.38714484182977127,
"learning_rate": 2.8812813884007306e-05,
"loss": 0.4157,
"step": 21755
},
{
"epoch": 2.5373134328358207,
"grad_norm": 0.401423487689331,
"learning_rate": 2.8804152478379377e-05,
"loss": 0.4004,
"step": 21760
},
{
"epoch": 2.5378964552238807,
"grad_norm": 0.39028263032288135,
"learning_rate": 2.8795490878836468e-05,
"loss": 0.3918,
"step": 21765
},
{
"epoch": 2.5384794776119404,
"grad_norm": 0.4065366877513189,
"learning_rate": 2.8786829086666483e-05,
"loss": 0.3895,
"step": 21770
},
{
"epoch": 2.5390625,
"grad_norm": 0.38717226865681914,
"learning_rate": 2.877816710315734e-05,
"loss": 0.3943,
"step": 21775
},
{
"epoch": 2.5396455223880596,
"grad_norm": 0.42458318342762813,
"learning_rate": 2.8769504929596986e-05,
"loss": 0.4244,
"step": 21780
},
{
"epoch": 2.5402285447761193,
"grad_norm": 0.5479384180330955,
"learning_rate": 2.876084256727342e-05,
"loss": 0.462,
"step": 21785
},
{
"epoch": 2.5408115671641793,
"grad_norm": 0.41665436063921335,
"learning_rate": 2.8752180017474646e-05,
"loss": 0.4203,
"step": 21790
},
{
"epoch": 2.541394589552239,
"grad_norm": 0.42340080438974775,
"learning_rate": 2.8743517281488703e-05,
"loss": 0.417,
"step": 21795
},
{
"epoch": 2.5419776119402986,
"grad_norm": 0.4255330334928534,
"learning_rate": 2.8734854360603646e-05,
"loss": 0.4148,
"step": 21800
},
{
"epoch": 2.542560634328358,
"grad_norm": 0.45592826028439914,
"learning_rate": 2.8726191256107582e-05,
"loss": 0.3922,
"step": 21805
},
{
"epoch": 2.543143656716418,
"grad_norm": 0.403101059759272,
"learning_rate": 2.8717527969288632e-05,
"loss": 0.387,
"step": 21810
},
{
"epoch": 2.5437266791044775,
"grad_norm": 0.4021919699524545,
"learning_rate": 2.870886450143493e-05,
"loss": 0.4124,
"step": 21815
},
{
"epoch": 2.544309701492537,
"grad_norm": 0.422454164200704,
"learning_rate": 2.870020085383466e-05,
"loss": 0.4013,
"step": 21820
},
{
"epoch": 2.544892723880597,
"grad_norm": 0.45287185780633143,
"learning_rate": 2.8691537027776022e-05,
"loss": 0.4303,
"step": 21825
},
{
"epoch": 2.545475746268657,
"grad_norm": 0.4108350468375693,
"learning_rate": 2.868287302454725e-05,
"loss": 0.4208,
"step": 21830
},
{
"epoch": 2.5460587686567164,
"grad_norm": 0.42982432907901225,
"learning_rate": 2.867420884543659e-05,
"loss": 0.4029,
"step": 21835
},
{
"epoch": 2.546641791044776,
"grad_norm": 0.4086358957532511,
"learning_rate": 2.8665544491732315e-05,
"loss": 0.4133,
"step": 21840
},
{
"epoch": 2.5472248134328357,
"grad_norm": 0.45077563953403066,
"learning_rate": 2.8656879964722753e-05,
"loss": 0.4225,
"step": 21845
},
{
"epoch": 2.5478078358208958,
"grad_norm": 0.49688042020521955,
"learning_rate": 2.8648215265696227e-05,
"loss": 0.4206,
"step": 21850
},
{
"epoch": 2.5483908582089554,
"grad_norm": 0.4286001722564965,
"learning_rate": 2.8639550395941085e-05,
"loss": 0.4105,
"step": 21855
},
{
"epoch": 2.548973880597015,
"grad_norm": 0.39585660153808594,
"learning_rate": 2.8630885356745716e-05,
"loss": 0.4275,
"step": 21860
},
{
"epoch": 2.5495569029850746,
"grad_norm": 0.39916539236865917,
"learning_rate": 2.8622220149398533e-05,
"loss": 0.3737,
"step": 21865
},
{
"epoch": 2.5501399253731343,
"grad_norm": 0.3998433747834968,
"learning_rate": 2.8613554775187962e-05,
"loss": 0.3922,
"step": 21870
},
{
"epoch": 2.550722947761194,
"grad_norm": 0.37614164356161306,
"learning_rate": 2.860488923540247e-05,
"loss": 0.4006,
"step": 21875
},
{
"epoch": 2.5513059701492535,
"grad_norm": 0.4446740096753151,
"learning_rate": 2.859622353133054e-05,
"loss": 0.4203,
"step": 21880
},
{
"epoch": 2.5518889925373136,
"grad_norm": 0.42220583522566685,
"learning_rate": 2.8587557664260662e-05,
"loss": 0.4038,
"step": 21885
},
{
"epoch": 2.5524720149253732,
"grad_norm": 0.4180017378797911,
"learning_rate": 2.8578891635481387e-05,
"loss": 0.429,
"step": 21890
},
{
"epoch": 2.553055037313433,
"grad_norm": 0.42970196216904194,
"learning_rate": 2.857022544628126e-05,
"loss": 0.4237,
"step": 21895
},
{
"epoch": 2.5536380597014925,
"grad_norm": 0.407121658567729,
"learning_rate": 2.8561559097948863e-05,
"loss": 0.401,
"step": 21900
},
{
"epoch": 2.554221082089552,
"grad_norm": 0.3922610519380279,
"learning_rate": 2.8552892591772806e-05,
"loss": 0.3913,
"step": 21905
},
{
"epoch": 2.554804104477612,
"grad_norm": 0.4098739412970042,
"learning_rate": 2.8544225929041697e-05,
"loss": 0.4026,
"step": 21910
},
{
"epoch": 2.5553871268656714,
"grad_norm": 0.40623477535739433,
"learning_rate": 2.8535559111044206e-05,
"loss": 0.4082,
"step": 21915
},
{
"epoch": 2.5559701492537314,
"grad_norm": 0.4016199061212181,
"learning_rate": 2.852689213906899e-05,
"loss": 0.4331,
"step": 21920
},
{
"epoch": 2.556553171641791,
"grad_norm": 0.4455436046189285,
"learning_rate": 2.851822501440476e-05,
"loss": 0.382,
"step": 21925
},
{
"epoch": 2.5571361940298507,
"grad_norm": 0.409219008525592,
"learning_rate": 2.850955773834022e-05,
"loss": 0.3984,
"step": 21930
},
{
"epoch": 2.5577192164179103,
"grad_norm": 0.39634426680336415,
"learning_rate": 2.850089031216412e-05,
"loss": 0.3869,
"step": 21935
},
{
"epoch": 2.55830223880597,
"grad_norm": 0.46628954030010084,
"learning_rate": 2.849222273716522e-05,
"loss": 0.395,
"step": 21940
},
{
"epoch": 2.55888526119403,
"grad_norm": 0.4401779440320999,
"learning_rate": 2.84835550146323e-05,
"loss": 0.4189,
"step": 21945
},
{
"epoch": 2.5594682835820897,
"grad_norm": 0.44388153967136124,
"learning_rate": 2.8474887145854183e-05,
"loss": 0.408,
"step": 21950
},
{
"epoch": 2.5600513059701493,
"grad_norm": 0.4036687461098782,
"learning_rate": 2.8466219132119688e-05,
"loss": 0.4039,
"step": 21955
},
{
"epoch": 2.560634328358209,
"grad_norm": 0.4032150130986403,
"learning_rate": 2.8457550974717655e-05,
"loss": 0.4208,
"step": 21960
},
{
"epoch": 2.5612173507462686,
"grad_norm": 0.4125258192596562,
"learning_rate": 2.8448882674936973e-05,
"loss": 0.3951,
"step": 21965
},
{
"epoch": 2.5618003731343286,
"grad_norm": 0.4104849801208035,
"learning_rate": 2.8440214234066524e-05,
"loss": 0.4074,
"step": 21970
},
{
"epoch": 2.562383395522388,
"grad_norm": 0.39023451394189007,
"learning_rate": 2.8431545653395236e-05,
"loss": 0.3778,
"step": 21975
},
{
"epoch": 2.562966417910448,
"grad_norm": 0.43210189811340993,
"learning_rate": 2.8422876934212027e-05,
"loss": 0.4341,
"step": 21980
},
{
"epoch": 2.5635494402985075,
"grad_norm": 0.4380309452516506,
"learning_rate": 2.841420807780586e-05,
"loss": 0.4221,
"step": 21985
},
{
"epoch": 2.564132462686567,
"grad_norm": 0.4332141557386373,
"learning_rate": 2.8405539085465717e-05,
"loss": 0.4323,
"step": 21990
},
{
"epoch": 2.5647154850746268,
"grad_norm": 0.46081240369833126,
"learning_rate": 2.8396869958480587e-05,
"loss": 0.3924,
"step": 21995
},
{
"epoch": 2.5652985074626864,
"grad_norm": 0.4397813502612307,
"learning_rate": 2.8388200698139484e-05,
"loss": 0.4371,
"step": 22000
},
{
"epoch": 2.5658815298507465,
"grad_norm": 0.4130395762991295,
"learning_rate": 2.837953130573145e-05,
"loss": 0.3803,
"step": 22005
},
{
"epoch": 2.566464552238806,
"grad_norm": 0.46598689677584026,
"learning_rate": 2.8370861782545537e-05,
"loss": 0.4045,
"step": 22010
},
{
"epoch": 2.5670475746268657,
"grad_norm": 0.36310078298500686,
"learning_rate": 2.8362192129870817e-05,
"loss": 0.4044,
"step": 22015
},
{
"epoch": 2.5676305970149254,
"grad_norm": 0.4385515606137188,
"learning_rate": 2.8353522348996388e-05,
"loss": 0.4217,
"step": 22020
},
{
"epoch": 2.568213619402985,
"grad_norm": 0.40335608705844234,
"learning_rate": 2.8344852441211367e-05,
"loss": 0.4213,
"step": 22025
},
{
"epoch": 2.5687966417910446,
"grad_norm": 0.42055530755961007,
"learning_rate": 2.8336182407804886e-05,
"loss": 0.4073,
"step": 22030
},
{
"epoch": 2.5693796641791042,
"grad_norm": 0.4021555512781502,
"learning_rate": 2.8327512250066083e-05,
"loss": 0.433,
"step": 22035
},
{
"epoch": 2.5699626865671643,
"grad_norm": 0.4013356366914958,
"learning_rate": 2.8318841969284145e-05,
"loss": 0.3689,
"step": 22040
},
{
"epoch": 2.570545708955224,
"grad_norm": 0.40531603837072133,
"learning_rate": 2.8310171566748243e-05,
"loss": 0.4037,
"step": 22045
},
{
"epoch": 2.5711287313432836,
"grad_norm": 0.41115410771966315,
"learning_rate": 2.8301501043747608e-05,
"loss": 0.4164,
"step": 22050
},
{
"epoch": 2.571711753731343,
"grad_norm": 0.42752996731676995,
"learning_rate": 2.829283040157143e-05,
"loss": 0.4021,
"step": 22055
},
{
"epoch": 2.572294776119403,
"grad_norm": 0.44069654554966997,
"learning_rate": 2.8284159641508972e-05,
"loss": 0.4024,
"step": 22060
},
{
"epoch": 2.572877798507463,
"grad_norm": 0.4311906079588823,
"learning_rate": 2.827548876484949e-05,
"loss": 0.4185,
"step": 22065
},
{
"epoch": 2.5734608208955225,
"grad_norm": 0.39731550212635675,
"learning_rate": 2.826681777288226e-05,
"loss": 0.3763,
"step": 22070
},
{
"epoch": 2.574043843283582,
"grad_norm": 0.42071353294179487,
"learning_rate": 2.825814666689658e-05,
"loss": 0.394,
"step": 22075
},
{
"epoch": 2.574626865671642,
"grad_norm": 0.42675361843367593,
"learning_rate": 2.824947544818175e-05,
"loss": 0.422,
"step": 22080
},
{
"epoch": 2.5752098880597014,
"grad_norm": 0.4205779890422091,
"learning_rate": 2.8240804118027092e-05,
"loss": 0.3994,
"step": 22085
},
{
"epoch": 2.575792910447761,
"grad_norm": 0.3833468747199891,
"learning_rate": 2.8232132677721972e-05,
"loss": 0.406,
"step": 22090
},
{
"epoch": 2.5763759328358207,
"grad_norm": 0.4054730357193972,
"learning_rate": 2.8223461128555727e-05,
"loss": 0.3922,
"step": 22095
},
{
"epoch": 2.5769589552238807,
"grad_norm": 0.4460472381553438,
"learning_rate": 2.8214789471817754e-05,
"loss": 0.4172,
"step": 22100
},
{
"epoch": 2.5775419776119404,
"grad_norm": 0.4247270949273739,
"learning_rate": 2.8206117708797432e-05,
"loss": 0.4344,
"step": 22105
},
{
"epoch": 2.578125,
"grad_norm": 0.4021692662185986,
"learning_rate": 2.819744584078417e-05,
"loss": 0.4083,
"step": 22110
},
{
"epoch": 2.5787080223880596,
"grad_norm": 0.4014781071940521,
"learning_rate": 2.818877386906739e-05,
"loss": 0.4134,
"step": 22115
},
{
"epoch": 2.5792910447761193,
"grad_norm": 0.40459568724162415,
"learning_rate": 2.8180101794936542e-05,
"loss": 0.4081,
"step": 22120
},
{
"epoch": 2.5798740671641793,
"grad_norm": 0.41212136983585784,
"learning_rate": 2.8171429619681073e-05,
"loss": 0.4111,
"step": 22125
},
{
"epoch": 2.580457089552239,
"grad_norm": 0.4577787954720089,
"learning_rate": 2.8162757344590445e-05,
"loss": 0.4203,
"step": 22130
},
{
"epoch": 2.5810401119402986,
"grad_norm": 0.39595967976502433,
"learning_rate": 2.815408497095416e-05,
"loss": 0.4018,
"step": 22135
},
{
"epoch": 2.581623134328358,
"grad_norm": 0.39472005263470467,
"learning_rate": 2.8145412500061702e-05,
"loss": 0.3734,
"step": 22140
},
{
"epoch": 2.582206156716418,
"grad_norm": 0.4172146465184219,
"learning_rate": 2.813673993320259e-05,
"loss": 0.4212,
"step": 22145
},
{
"epoch": 2.5827891791044775,
"grad_norm": 0.3902573137975402,
"learning_rate": 2.812806727166635e-05,
"loss": 0.4147,
"step": 22150
},
{
"epoch": 2.583372201492537,
"grad_norm": 0.43001278044711233,
"learning_rate": 2.811939451674252e-05,
"loss": 0.4102,
"step": 22155
},
{
"epoch": 2.583955223880597,
"grad_norm": 0.3730688765244184,
"learning_rate": 2.8110721669720663e-05,
"loss": 0.3709,
"step": 22160
},
{
"epoch": 2.584538246268657,
"grad_norm": 0.4299995367289496,
"learning_rate": 2.8102048731890345e-05,
"loss": 0.391,
"step": 22165
},
{
"epoch": 2.5851212686567164,
"grad_norm": 0.4118428597575234,
"learning_rate": 2.8093375704541158e-05,
"loss": 0.3908,
"step": 22170
},
{
"epoch": 2.585704291044776,
"grad_norm": 0.43560160129228553,
"learning_rate": 2.808470258896268e-05,
"loss": 0.4025,
"step": 22175
},
{
"epoch": 2.5862873134328357,
"grad_norm": 0.41686734968264216,
"learning_rate": 2.8076029386444524e-05,
"loss": 0.4024,
"step": 22180
},
{
"epoch": 2.5868703358208958,
"grad_norm": 0.37646393896873553,
"learning_rate": 2.806735609827633e-05,
"loss": 0.3895,
"step": 22185
},
{
"epoch": 2.5874533582089554,
"grad_norm": 0.41617694394390714,
"learning_rate": 2.805868272574771e-05,
"loss": 0.3886,
"step": 22190
},
{
"epoch": 2.588036380597015,
"grad_norm": 0.4455298031267967,
"learning_rate": 2.8050009270148326e-05,
"loss": 0.432,
"step": 22195
},
{
"epoch": 2.5886194029850746,
"grad_norm": 0.44424990770711303,
"learning_rate": 2.804133573276783e-05,
"loss": 0.4439,
"step": 22200
},
{
"epoch": 2.5892024253731343,
"grad_norm": 0.4357978692348376,
"learning_rate": 2.803266211489591e-05,
"loss": 0.4076,
"step": 22205
},
{
"epoch": 2.589785447761194,
"grad_norm": 0.4062729433625467,
"learning_rate": 2.8023988417822222e-05,
"loss": 0.3905,
"step": 22210
},
{
"epoch": 2.5903684701492535,
"grad_norm": 0.4152482478748045,
"learning_rate": 2.8015314642836476e-05,
"loss": 0.3906,
"step": 22215
},
{
"epoch": 2.5909514925373136,
"grad_norm": 0.42204159354622933,
"learning_rate": 2.800664079122839e-05,
"loss": 0.423,
"step": 22220
},
{
"epoch": 2.5915345149253732,
"grad_norm": 0.40821190774312066,
"learning_rate": 2.799796686428766e-05,
"loss": 0.4179,
"step": 22225
},
{
"epoch": 2.592117537313433,
"grad_norm": 0.4052933237396047,
"learning_rate": 2.7989292863304045e-05,
"loss": 0.3832,
"step": 22230
},
{
"epoch": 2.5927005597014925,
"grad_norm": 0.4229739013081971,
"learning_rate": 2.7980618789567258e-05,
"loss": 0.4054,
"step": 22235
},
{
"epoch": 2.593283582089552,
"grad_norm": 0.42874283632249127,
"learning_rate": 2.7971944644367066e-05,
"loss": 0.4123,
"step": 22240
},
{
"epoch": 2.593866604477612,
"grad_norm": 0.3909236414679042,
"learning_rate": 2.796327042899322e-05,
"loss": 0.4103,
"step": 22245
},
{
"epoch": 2.5944496268656714,
"grad_norm": 0.3975325102935005,
"learning_rate": 2.7954596144735512e-05,
"loss": 0.382,
"step": 22250
},
{
"epoch": 2.5950326492537314,
"grad_norm": 0.4088920813258165,
"learning_rate": 2.7945921792883707e-05,
"loss": 0.3917,
"step": 22255
},
{
"epoch": 2.595615671641791,
"grad_norm": 0.4530264932143675,
"learning_rate": 2.79372473747276e-05,
"loss": 0.4198,
"step": 22260
},
{
"epoch": 2.5961986940298507,
"grad_norm": 0.4434793559794076,
"learning_rate": 2.7928572891557003e-05,
"loss": 0.4291,
"step": 22265
},
{
"epoch": 2.5967817164179103,
"grad_norm": 0.40488246331631406,
"learning_rate": 2.7919898344661723e-05,
"loss": 0.4139,
"step": 22270
},
{
"epoch": 2.59736473880597,
"grad_norm": 0.3983270611357439,
"learning_rate": 2.791122373533157e-05,
"loss": 0.4037,
"step": 22275
},
{
"epoch": 2.59794776119403,
"grad_norm": 0.45657914992253346,
"learning_rate": 2.7902549064856405e-05,
"loss": 0.4098,
"step": 22280
},
{
"epoch": 2.5985307835820897,
"grad_norm": 0.4575063731920724,
"learning_rate": 2.7893874334526043e-05,
"loss": 0.423,
"step": 22285
},
{
"epoch": 2.5991138059701493,
"grad_norm": 0.4133352840595252,
"learning_rate": 2.7885199545630343e-05,
"loss": 0.4223,
"step": 22290
},
{
"epoch": 2.599696828358209,
"grad_norm": 0.42850769313939163,
"learning_rate": 2.7876524699459163e-05,
"loss": 0.4342,
"step": 22295
},
{
"epoch": 2.6002798507462686,
"grad_norm": 0.4077259449427239,
"learning_rate": 2.7867849797302357e-05,
"loss": 0.3952,
"step": 22300
},
{
"epoch": 2.6008628731343286,
"grad_norm": 0.4841522503520805,
"learning_rate": 2.7859174840449826e-05,
"loss": 0.4095,
"step": 22305
},
{
"epoch": 2.601445895522388,
"grad_norm": 0.4242614014407421,
"learning_rate": 2.785049983019143e-05,
"loss": 0.4199,
"step": 22310
},
{
"epoch": 2.602028917910448,
"grad_norm": 0.4716478244582767,
"learning_rate": 2.7841824767817065e-05,
"loss": 0.403,
"step": 22315
},
{
"epoch": 2.6026119402985075,
"grad_norm": 0.43420482892718326,
"learning_rate": 2.7833149654616637e-05,
"loss": 0.4232,
"step": 22320
},
{
"epoch": 2.603194962686567,
"grad_norm": 0.40058425131317626,
"learning_rate": 2.782447449188004e-05,
"loss": 0.3793,
"step": 22325
},
{
"epoch": 2.6037779850746268,
"grad_norm": 0.39819558111581616,
"learning_rate": 2.7815799280897202e-05,
"loss": 0.4095,
"step": 22330
},
{
"epoch": 2.6043610074626864,
"grad_norm": 0.39928022261189117,
"learning_rate": 2.780712402295803e-05,
"loss": 0.3788,
"step": 22335
},
{
"epoch": 2.6049440298507465,
"grad_norm": 0.44112096336092194,
"learning_rate": 2.7798448719352467e-05,
"loss": 0.4381,
"step": 22340
},
{
"epoch": 2.605527052238806,
"grad_norm": 0.3987578265750149,
"learning_rate": 2.778977337137044e-05,
"loss": 0.3785,
"step": 22345
},
{
"epoch": 2.6061100746268657,
"grad_norm": 0.4212063703933318,
"learning_rate": 2.7781097980301878e-05,
"loss": 0.4011,
"step": 22350
},
{
"epoch": 2.6066930970149254,
"grad_norm": 0.44446745041834496,
"learning_rate": 2.777242254743675e-05,
"loss": 0.4113,
"step": 22355
},
{
"epoch": 2.607276119402985,
"grad_norm": 0.4460450398839979,
"learning_rate": 2.7763747074065e-05,
"loss": 0.4231,
"step": 22360
},
{
"epoch": 2.6078591417910446,
"grad_norm": 0.4233552010343455,
"learning_rate": 2.775507156147658e-05,
"loss": 0.4226,
"step": 22365
},
{
"epoch": 2.6084421641791042,
"grad_norm": 0.4212752781517472,
"learning_rate": 2.7746396010961462e-05,
"loss": 0.3786,
"step": 22370
},
{
"epoch": 2.6090251865671643,
"grad_norm": 0.4767806851482472,
"learning_rate": 2.773772042380962e-05,
"loss": 0.4075,
"step": 22375
},
{
"epoch": 2.609608208955224,
"grad_norm": 0.48360042803705594,
"learning_rate": 2.7729044801311032e-05,
"loss": 0.3988,
"step": 22380
},
{
"epoch": 2.6101912313432836,
"grad_norm": 0.38229719290658104,
"learning_rate": 2.772036914475567e-05,
"loss": 0.3905,
"step": 22385
},
{
"epoch": 2.610774253731343,
"grad_norm": 0.4133302946992925,
"learning_rate": 2.7711693455433534e-05,
"loss": 0.3953,
"step": 22390
},
{
"epoch": 2.611357276119403,
"grad_norm": 0.4165574527008948,
"learning_rate": 2.7703017734634608e-05,
"loss": 0.3945,
"step": 22395
},
{
"epoch": 2.611940298507463,
"grad_norm": 0.4210804912903456,
"learning_rate": 2.7694341983648884e-05,
"loss": 0.4313,
"step": 22400
},
{
"epoch": 2.6125233208955225,
"grad_norm": 0.4367388363947852,
"learning_rate": 2.768566620376638e-05,
"loss": 0.4329,
"step": 22405
},
{
"epoch": 2.613106343283582,
"grad_norm": 0.424959420072146,
"learning_rate": 2.7676990396277085e-05,
"loss": 0.4094,
"step": 22410
},
{
"epoch": 2.613689365671642,
"grad_norm": 0.39468901509057047,
"learning_rate": 2.7668314562471008e-05,
"loss": 0.3875,
"step": 22415
},
{
"epoch": 2.6142723880597014,
"grad_norm": 0.4262353191509631,
"learning_rate": 2.7659638703638173e-05,
"loss": 0.4233,
"step": 22420
},
{
"epoch": 2.614855410447761,
"grad_norm": 0.4076459301141559,
"learning_rate": 2.7650962821068595e-05,
"loss": 0.3985,
"step": 22425
},
{
"epoch": 2.6154384328358207,
"grad_norm": 0.46846289670042857,
"learning_rate": 2.764228691605229e-05,
"loss": 0.4245,
"step": 22430
},
{
"epoch": 2.6160214552238807,
"grad_norm": 0.3901685618490997,
"learning_rate": 2.7633610989879284e-05,
"loss": 0.3936,
"step": 22435
},
{
"epoch": 2.6166044776119404,
"grad_norm": 0.40547142558021154,
"learning_rate": 2.76249350438396e-05,
"loss": 0.3963,
"step": 22440
},
{
"epoch": 2.6171875,
"grad_norm": 0.4070049551386447,
"learning_rate": 2.761625907922328e-05,
"loss": 0.4323,
"step": 22445
},
{
"epoch": 2.6177705223880596,
"grad_norm": 0.40403875423687535,
"learning_rate": 2.7607583097320345e-05,
"loss": 0.393,
"step": 22450
},
{
"epoch": 2.6183535447761193,
"grad_norm": 0.3826446833329266,
"learning_rate": 2.7598907099420835e-05,
"loss": 0.3775,
"step": 22455
},
{
"epoch": 2.6189365671641793,
"grad_norm": 0.472760481953975,
"learning_rate": 2.7590231086814782e-05,
"loss": 0.4497,
"step": 22460
},
{
"epoch": 2.619519589552239,
"grad_norm": 0.4931381898829188,
"learning_rate": 2.758155506079223e-05,
"loss": 0.4046,
"step": 22465
},
{
"epoch": 2.6201026119402986,
"grad_norm": 0.3828620701724777,
"learning_rate": 2.7572879022643228e-05,
"loss": 0.3692,
"step": 22470
},
{
"epoch": 2.620685634328358,
"grad_norm": 0.4408182639542891,
"learning_rate": 2.7564202973657815e-05,
"loss": 0.3907,
"step": 22475
},
{
"epoch": 2.621268656716418,
"grad_norm": 0.39359600706846304,
"learning_rate": 2.7555526915126033e-05,
"loss": 0.4072,
"step": 22480
},
{
"epoch": 2.6218516791044775,
"grad_norm": 0.38439887698226066,
"learning_rate": 2.754685084833793e-05,
"loss": 0.3863,
"step": 22485
},
{
"epoch": 2.622434701492537,
"grad_norm": 0.4157945776321559,
"learning_rate": 2.7538174774583552e-05,
"loss": 0.3726,
"step": 22490
},
{
"epoch": 2.623017723880597,
"grad_norm": 0.49145189075517376,
"learning_rate": 2.752949869515295e-05,
"loss": 0.4228,
"step": 22495
},
{
"epoch": 2.623600746268657,
"grad_norm": 0.4141780869591111,
"learning_rate": 2.7520822611336176e-05,
"loss": 0.4133,
"step": 22500
},
{
"epoch": 2.6241837686567164,
"grad_norm": 0.4119506326344177,
"learning_rate": 2.7512146524423288e-05,
"loss": 0.3876,
"step": 22505
},
{
"epoch": 2.624766791044776,
"grad_norm": 0.4187381085954075,
"learning_rate": 2.7503470435704322e-05,
"loss": 0.394,
"step": 22510
},
{
"epoch": 2.6253498134328357,
"grad_norm": 0.4090135178615731,
"learning_rate": 2.7494794346469327e-05,
"loss": 0.4223,
"step": 22515
},
{
"epoch": 2.6259328358208958,
"grad_norm": 0.4895679200073971,
"learning_rate": 2.7486118258008374e-05,
"loss": 0.391,
"step": 22520
},
{
"epoch": 2.6265158582089554,
"grad_norm": 0.41769312414868964,
"learning_rate": 2.74774421716115e-05,
"loss": 0.4008,
"step": 22525
},
{
"epoch": 2.627098880597015,
"grad_norm": 0.43231608436340224,
"learning_rate": 2.746876608856876e-05,
"loss": 0.4026,
"step": 22530
},
{
"epoch": 2.6276819029850746,
"grad_norm": 0.37945509445185577,
"learning_rate": 2.7460090010170197e-05,
"loss": 0.3772,
"step": 22535
},
{
"epoch": 2.6282649253731343,
"grad_norm": 0.41771793432698484,
"learning_rate": 2.7451413937705878e-05,
"loss": 0.3921,
"step": 22540
},
{
"epoch": 2.628847947761194,
"grad_norm": 0.4172517343337237,
"learning_rate": 2.7442737872465835e-05,
"loss": 0.4105,
"step": 22545
},
{
"epoch": 2.6294309701492535,
"grad_norm": 0.40019931215224414,
"learning_rate": 2.743406181574012e-05,
"loss": 0.4056,
"step": 22550
},
{
"epoch": 2.6300139925373136,
"grad_norm": 0.3876915080045103,
"learning_rate": 2.7425385768818784e-05,
"loss": 0.4061,
"step": 22555
},
{
"epoch": 2.6305970149253732,
"grad_norm": 0.4244692317339967,
"learning_rate": 2.7416709732991863e-05,
"loss": 0.4166,
"step": 22560
},
{
"epoch": 2.631180037313433,
"grad_norm": 0.4714360469640444,
"learning_rate": 2.7408033709549413e-05,
"loss": 0.4269,
"step": 22565
},
{
"epoch": 2.6317630597014925,
"grad_norm": 0.4583986896646736,
"learning_rate": 2.7399357699781477e-05,
"loss": 0.4063,
"step": 22570
},
{
"epoch": 2.632346082089552,
"grad_norm": 0.4053176001235881,
"learning_rate": 2.739068170497807e-05,
"loss": 0.4134,
"step": 22575
},
{
"epoch": 2.632929104477612,
"grad_norm": 0.38513463011362853,
"learning_rate": 2.7382005726429256e-05,
"loss": 0.3723,
"step": 22580
},
{
"epoch": 2.6335121268656714,
"grad_norm": 0.4170962101723745,
"learning_rate": 2.7373329765425053e-05,
"loss": 0.4114,
"step": 22585
},
{
"epoch": 2.6340951492537314,
"grad_norm": 0.42612121152467314,
"learning_rate": 2.736465382325551e-05,
"loss": 0.3895,
"step": 22590
},
{
"epoch": 2.634678171641791,
"grad_norm": 0.44297163836302395,
"learning_rate": 2.7355977901210646e-05,
"loss": 0.4147,
"step": 22595
},
{
"epoch": 2.6352611940298507,
"grad_norm": 0.3967897810112757,
"learning_rate": 2.7347302000580475e-05,
"loss": 0.4174,
"step": 22600
},
{
"epoch": 2.6358442164179103,
"grad_norm": 0.43077554945593344,
"learning_rate": 2.7338626122655047e-05,
"loss": 0.394,
"step": 22605
},
{
"epoch": 2.63642723880597,
"grad_norm": 0.7404115339398403,
"learning_rate": 2.7329950268724358e-05,
"loss": 0.4053,
"step": 22610
},
{
"epoch": 2.63701026119403,
"grad_norm": 0.40749654219568165,
"learning_rate": 2.7321274440078442e-05,
"loss": 0.3827,
"step": 22615
},
{
"epoch": 2.6375932835820897,
"grad_norm": 0.453240222390189,
"learning_rate": 2.7312598638007308e-05,
"loss": 0.4143,
"step": 22620
},
{
"epoch": 2.6381763059701493,
"grad_norm": 0.40924706417700796,
"learning_rate": 2.7303922863800952e-05,
"loss": 0.406,
"step": 22625
},
{
"epoch": 2.638759328358209,
"grad_norm": 0.4287701356191956,
"learning_rate": 2.7295247118749395e-05,
"loss": 0.4277,
"step": 22630
},
{
"epoch": 2.6393423507462686,
"grad_norm": 0.4429404967354247,
"learning_rate": 2.728657140414262e-05,
"loss": 0.4176,
"step": 22635
},
{
"epoch": 2.6399253731343286,
"grad_norm": 0.42899805695313875,
"learning_rate": 2.727789572127064e-05,
"loss": 0.4077,
"step": 22640
},
{
"epoch": 2.640508395522388,
"grad_norm": 0.4433381158083287,
"learning_rate": 2.726922007142344e-05,
"loss": 0.4034,
"step": 22645
},
{
"epoch": 2.641091417910448,
"grad_norm": 0.4250267655283454,
"learning_rate": 2.7260544455890996e-05,
"loss": 0.3969,
"step": 22650
},
{
"epoch": 2.6416744402985075,
"grad_norm": 0.4217829392477543,
"learning_rate": 2.7251868875963312e-05,
"loss": 0.4214,
"step": 22655
},
{
"epoch": 2.642257462686567,
"grad_norm": 0.39776670231864214,
"learning_rate": 2.724319333293033e-05,
"loss": 0.375,
"step": 22660
},
{
"epoch": 2.6428404850746268,
"grad_norm": 0.4490704964111275,
"learning_rate": 2.723451782808205e-05,
"loss": 0.4137,
"step": 22665
},
{
"epoch": 2.6434235074626864,
"grad_norm": 0.42191925987309814,
"learning_rate": 2.7225842362708427e-05,
"loss": 0.3901,
"step": 22670
},
{
"epoch": 2.6440065298507465,
"grad_norm": 0.38597234858085877,
"learning_rate": 2.7217166938099402e-05,
"loss": 0.4004,
"step": 22675
},
{
"epoch": 2.644589552238806,
"grad_norm": 0.4017520282470813,
"learning_rate": 2.7208491555544964e-05,
"loss": 0.3971,
"step": 22680
},
{
"epoch": 2.6451725746268657,
"grad_norm": 0.4006865608636846,
"learning_rate": 2.7199816216335024e-05,
"loss": 0.4131,
"step": 22685
},
{
"epoch": 2.6457555970149254,
"grad_norm": 0.4380554997393179,
"learning_rate": 2.7191140921759546e-05,
"loss": 0.4302,
"step": 22690
},
{
"epoch": 2.646338619402985,
"grad_norm": 0.43207188614631753,
"learning_rate": 2.7182465673108454e-05,
"loss": 0.4141,
"step": 22695
},
{
"epoch": 2.6469216417910446,
"grad_norm": 0.42491075705628634,
"learning_rate": 2.7173790471671662e-05,
"loss": 0.3825,
"step": 22700
},
{
"epoch": 2.6475046641791042,
"grad_norm": 0.40472752807362644,
"learning_rate": 2.7165115318739114e-05,
"loss": 0.3965,
"step": 22705
},
{
"epoch": 2.6480876865671643,
"grad_norm": 0.41736521217127226,
"learning_rate": 2.7156440215600703e-05,
"loss": 0.4186,
"step": 22710
},
{
"epoch": 2.648670708955224,
"grad_norm": 0.41016883747121546,
"learning_rate": 2.7147765163546347e-05,
"loss": 0.4239,
"step": 22715
},
{
"epoch": 2.6492537313432836,
"grad_norm": 0.3966850260485352,
"learning_rate": 2.7139090163865932e-05,
"loss": 0.394,
"step": 22720
},
{
"epoch": 2.649836753731343,
"grad_norm": 0.4372589987881457,
"learning_rate": 2.7130415217849354e-05,
"loss": 0.3926,
"step": 22725
},
{
"epoch": 2.650419776119403,
"grad_norm": 0.4133710259665845,
"learning_rate": 2.712174032678648e-05,
"loss": 0.4281,
"step": 22730
},
{
"epoch": 2.651002798507463,
"grad_norm": 0.42125956473126536,
"learning_rate": 2.711306549196721e-05,
"loss": 0.415,
"step": 22735
},
{
"epoch": 2.6515858208955225,
"grad_norm": 0.42151885559638086,
"learning_rate": 2.7104390714681393e-05,
"loss": 0.4255,
"step": 22740
},
{
"epoch": 2.652168843283582,
"grad_norm": 0.4312968876015643,
"learning_rate": 2.7095715996218875e-05,
"loss": 0.4187,
"step": 22745
},
{
"epoch": 2.652751865671642,
"grad_norm": 0.39117445238789644,
"learning_rate": 2.7087041337869522e-05,
"loss": 0.4134,
"step": 22750
},
{
"epoch": 2.6533348880597014,
"grad_norm": 0.4040788468456531,
"learning_rate": 2.7078366740923173e-05,
"loss": 0.3892,
"step": 22755
},
{
"epoch": 2.653917910447761,
"grad_norm": 0.48078558835702323,
"learning_rate": 2.7069692206669633e-05,
"loss": 0.4212,
"step": 22760
},
{
"epoch": 2.6545009328358207,
"grad_norm": 0.4110706737895837,
"learning_rate": 2.7061017736398752e-05,
"loss": 0.4115,
"step": 22765
},
{
"epoch": 2.6550839552238807,
"grad_norm": 0.47303699942361366,
"learning_rate": 2.7052343331400322e-05,
"loss": 0.4071,
"step": 22770
},
{
"epoch": 2.6556669776119404,
"grad_norm": 0.4362663665645726,
"learning_rate": 2.7043668992964157e-05,
"loss": 0.4389,
"step": 22775
},
{
"epoch": 2.65625,
"grad_norm": 0.41176850375937024,
"learning_rate": 2.7034994722380036e-05,
"loss": 0.3869,
"step": 22780
},
{
"epoch": 2.6568330223880596,
"grad_norm": 0.4634401237951629,
"learning_rate": 2.7026320520937738e-05,
"loss": 0.4003,
"step": 22785
},
{
"epoch": 2.6574160447761193,
"grad_norm": 0.390507280135157,
"learning_rate": 2.701764638992705e-05,
"loss": 0.3876,
"step": 22790
},
{
"epoch": 2.6579990671641793,
"grad_norm": 0.45156977381832825,
"learning_rate": 2.700897233063771e-05,
"loss": 0.3873,
"step": 22795
},
{
"epoch": 2.658582089552239,
"grad_norm": 0.44211322039557294,
"learning_rate": 2.7000298344359494e-05,
"loss": 0.3986,
"step": 22800
},
{
"epoch": 2.6591651119402986,
"grad_norm": 0.40439680647060877,
"learning_rate": 2.6991624432382123e-05,
"loss": 0.4075,
"step": 22805
},
{
"epoch": 2.659748134328358,
"grad_norm": 0.4232521145995904,
"learning_rate": 2.6982950595995315e-05,
"loss": 0.4343,
"step": 22810
},
{
"epoch": 2.660331156716418,
"grad_norm": 0.4007371468963951,
"learning_rate": 2.6974276836488815e-05,
"loss": 0.3863,
"step": 22815
},
{
"epoch": 2.6609141791044775,
"grad_norm": 0.44755798649366285,
"learning_rate": 2.6965603155152302e-05,
"loss": 0.4087,
"step": 22820
},
{
"epoch": 2.661497201492537,
"grad_norm": 0.40849892329817666,
"learning_rate": 2.695692955327549e-05,
"loss": 0.4253,
"step": 22825
},
{
"epoch": 2.662080223880597,
"grad_norm": 0.3991343216495092,
"learning_rate": 2.6948256032148052e-05,
"loss": 0.4024,
"step": 22830
},
{
"epoch": 2.662663246268657,
"grad_norm": 0.4280138181137737,
"learning_rate": 2.6939582593059638e-05,
"loss": 0.4184,
"step": 22835
},
{
"epoch": 2.6632462686567164,
"grad_norm": 0.417897238898508,
"learning_rate": 2.6930909237299934e-05,
"loss": 0.4023,
"step": 22840
},
{
"epoch": 2.663829291044776,
"grad_norm": 0.4099105982377711,
"learning_rate": 2.692223596615857e-05,
"loss": 0.3971,
"step": 22845
},
{
"epoch": 2.6644123134328357,
"grad_norm": 0.4247214587174987,
"learning_rate": 2.691356278092519e-05,
"loss": 0.3981,
"step": 22850
},
{
"epoch": 2.6649953358208958,
"grad_norm": 0.42057720588431785,
"learning_rate": 2.6904889682889404e-05,
"loss": 0.4088,
"step": 22855
},
{
"epoch": 2.6655783582089554,
"grad_norm": 0.40337693004799835,
"learning_rate": 2.6896216673340814e-05,
"loss": 0.4156,
"step": 22860
},
{
"epoch": 2.666161380597015,
"grad_norm": 0.44041729987186246,
"learning_rate": 2.6887543753569022e-05,
"loss": 0.3958,
"step": 22865
},
{
"epoch": 2.6667444029850746,
"grad_norm": 0.3992974735678961,
"learning_rate": 2.687887092486361e-05,
"loss": 0.3951,
"step": 22870
},
{
"epoch": 2.6673274253731343,
"grad_norm": 0.49044311970394866,
"learning_rate": 2.6870198188514135e-05,
"loss": 0.4412,
"step": 22875
},
{
"epoch": 2.667910447761194,
"grad_norm": 0.45914136697215857,
"learning_rate": 2.686152554581016e-05,
"loss": 0.4483,
"step": 22880
},
{
"epoch": 2.6684934701492535,
"grad_norm": 0.42236530528271476,
"learning_rate": 2.6852852998041216e-05,
"loss": 0.3997,
"step": 22885
},
{
"epoch": 2.6690764925373136,
"grad_norm": 0.4006640978829436,
"learning_rate": 2.6844180546496833e-05,
"loss": 0.4116,
"step": 22890
},
{
"epoch": 2.6696595149253732,
"grad_norm": 0.4244299860535475,
"learning_rate": 2.6835508192466512e-05,
"loss": 0.4146,
"step": 22895
},
{
"epoch": 2.670242537313433,
"grad_norm": 0.43229327696854875,
"learning_rate": 2.682683593723977e-05,
"loss": 0.4009,
"step": 22900
},
{
"epoch": 2.6708255597014925,
"grad_norm": 0.4255475594996611,
"learning_rate": 2.6818163782106065e-05,
"loss": 0.3926,
"step": 22905
},
{
"epoch": 2.671408582089552,
"grad_norm": 0.41387623183724126,
"learning_rate": 2.680949172835487e-05,
"loss": 0.3794,
"step": 22910
},
{
"epoch": 2.671991604477612,
"grad_norm": 0.39916971201033524,
"learning_rate": 2.680081977727565e-05,
"loss": 0.4106,
"step": 22915
},
{
"epoch": 2.6725746268656714,
"grad_norm": 0.4225506725510269,
"learning_rate": 2.6792147930157812e-05,
"loss": 0.4047,
"step": 22920
},
{
"epoch": 2.6731576492537314,
"grad_norm": 0.4441389573144737,
"learning_rate": 2.6783476188290813e-05,
"loss": 0.4155,
"step": 22925
},
{
"epoch": 2.673740671641791,
"grad_norm": 0.5024459162447518,
"learning_rate": 2.6774804552964034e-05,
"loss": 0.4014,
"step": 22930
},
{
"epoch": 2.6743236940298507,
"grad_norm": 0.45680744890787905,
"learning_rate": 2.676613302546686e-05,
"loss": 0.4145,
"step": 22935
},
{
"epoch": 2.6749067164179103,
"grad_norm": 0.4168266138016353,
"learning_rate": 2.6757461607088692e-05,
"loss": 0.4058,
"step": 22940
},
{
"epoch": 2.67548973880597,
"grad_norm": 0.41624027321880297,
"learning_rate": 2.6748790299118846e-05,
"loss": 0.4018,
"step": 22945
},
{
"epoch": 2.67607276119403,
"grad_norm": 0.41227579319805346,
"learning_rate": 2.6740119102846707e-05,
"loss": 0.4348,
"step": 22950
},
{
"epoch": 2.6766557835820897,
"grad_norm": 0.40388939645917765,
"learning_rate": 2.6731448019561566e-05,
"loss": 0.4115,
"step": 22955
},
{
"epoch": 2.6772388059701493,
"grad_norm": 0.449562641306428,
"learning_rate": 2.6722777050552737e-05,
"loss": 0.4224,
"step": 22960
},
{
"epoch": 2.677821828358209,
"grad_norm": 0.42693867980569006,
"learning_rate": 2.6714106197109522e-05,
"loss": 0.4092,
"step": 22965
},
{
"epoch": 2.6784048507462686,
"grad_norm": 0.43130612603178214,
"learning_rate": 2.6705435460521177e-05,
"loss": 0.4101,
"step": 22970
},
{
"epoch": 2.6789878731343286,
"grad_norm": 0.4370067229452783,
"learning_rate": 2.6696764842076978e-05,
"loss": 0.4065,
"step": 22975
},
{
"epoch": 2.679570895522388,
"grad_norm": 0.4143678191065438,
"learning_rate": 2.668809434306615e-05,
"loss": 0.4137,
"step": 22980
},
{
"epoch": 2.680153917910448,
"grad_norm": 0.43136723228740675,
"learning_rate": 2.66794239647779e-05,
"loss": 0.3968,
"step": 22985
},
{
"epoch": 2.6807369402985075,
"grad_norm": 0.40280454269379085,
"learning_rate": 2.6670753708501454e-05,
"loss": 0.4125,
"step": 22990
},
{
"epoch": 2.681319962686567,
"grad_norm": 0.391809902282067,
"learning_rate": 2.6662083575525986e-05,
"loss": 0.3959,
"step": 22995
},
{
"epoch": 2.6819029850746268,
"grad_norm": 0.5427471445296358,
"learning_rate": 2.6653413567140668e-05,
"loss": 0.4388,
"step": 23000
},
{
"epoch": 2.6824860074626864,
"grad_norm": 0.49937611659196496,
"learning_rate": 2.6644743684634637e-05,
"loss": 0.4058,
"step": 23005
},
{
"epoch": 2.6830690298507465,
"grad_norm": 0.4167771466896609,
"learning_rate": 2.6636073929297018e-05,
"loss": 0.3937,
"step": 23010
},
{
"epoch": 2.683652052238806,
"grad_norm": 0.4024336174643316,
"learning_rate": 2.6627404302416935e-05,
"loss": 0.3864,
"step": 23015
},
{
"epoch": 2.6842350746268657,
"grad_norm": 0.4391183889059748,
"learning_rate": 2.661873480528347e-05,
"loss": 0.4209,
"step": 23020
},
{
"epoch": 2.6848180970149254,
"grad_norm": 0.44775313421072555,
"learning_rate": 2.66100654391857e-05,
"loss": 0.4197,
"step": 23025
},
{
"epoch": 2.685401119402985,
"grad_norm": 0.4284044805741468,
"learning_rate": 2.660139620541267e-05,
"loss": 0.4227,
"step": 23030
},
{
"epoch": 2.6859841417910446,
"grad_norm": 0.4200272710208144,
"learning_rate": 2.6592727105253413e-05,
"loss": 0.4262,
"step": 23035
},
{
"epoch": 2.6865671641791042,
"grad_norm": 0.4354503482949555,
"learning_rate": 2.6584058139996942e-05,
"loss": 0.4063,
"step": 23040
},
{
"epoch": 2.6871501865671643,
"grad_norm": 0.48701117877310707,
"learning_rate": 2.6575389310932252e-05,
"loss": 0.4203,
"step": 23045
},
{
"epoch": 2.687733208955224,
"grad_norm": 0.39316205216981354,
"learning_rate": 2.656672061934831e-05,
"loss": 0.3962,
"step": 23050
},
{
"epoch": 2.6883162313432836,
"grad_norm": 0.4453379823230853,
"learning_rate": 2.6558052066534077e-05,
"loss": 0.4195,
"step": 23055
},
{
"epoch": 2.688899253731343,
"grad_norm": 0.40011088033144115,
"learning_rate": 2.654938365377847e-05,
"loss": 0.3816,
"step": 23060
},
{
"epoch": 2.689482276119403,
"grad_norm": 0.4048436964264881,
"learning_rate": 2.6540715382370407e-05,
"loss": 0.4114,
"step": 23065
},
{
"epoch": 2.690065298507463,
"grad_norm": 0.45858943482750675,
"learning_rate": 2.6532047253598776e-05,
"loss": 0.4112,
"step": 23070
},
{
"epoch": 2.6906483208955225,
"grad_norm": 0.4093823948188578,
"learning_rate": 2.6523379268752448e-05,
"loss": 0.3849,
"step": 23075
},
{
"epoch": 2.691231343283582,
"grad_norm": 0.4727014551006822,
"learning_rate": 2.651471142912026e-05,
"loss": 0.4203,
"step": 23080
},
{
"epoch": 2.691814365671642,
"grad_norm": 0.37511059443170225,
"learning_rate": 2.6506043735991042e-05,
"loss": 0.4003,
"step": 23085
},
{
"epoch": 2.6923973880597014,
"grad_norm": 0.40664811860268496,
"learning_rate": 2.6497376190653607e-05,
"loss": 0.4036,
"step": 23090
},
{
"epoch": 2.692980410447761,
"grad_norm": 0.43437066417470704,
"learning_rate": 2.6488708794396712e-05,
"loss": 0.4038,
"step": 23095
},
{
"epoch": 2.6935634328358207,
"grad_norm": 0.41141835216852596,
"learning_rate": 2.6480041548509137e-05,
"loss": 0.3972,
"step": 23100
},
{
"epoch": 2.6941464552238807,
"grad_norm": 0.4041710313748638,
"learning_rate": 2.6471374454279617e-05,
"loss": 0.4119,
"step": 23105
},
{
"epoch": 2.6947294776119404,
"grad_norm": 0.41295264329339787,
"learning_rate": 2.6462707512996847e-05,
"loss": 0.3743,
"step": 23110
},
{
"epoch": 2.6953125,
"grad_norm": 0.3754314035467755,
"learning_rate": 2.645404072594954e-05,
"loss": 0.3946,
"step": 23115
},
{
"epoch": 2.6958955223880596,
"grad_norm": 0.39122158739748497,
"learning_rate": 2.644537409442635e-05,
"loss": 0.3883,
"step": 23120
},
{
"epoch": 2.6964785447761193,
"grad_norm": 0.3733786295671506,
"learning_rate": 2.6436707619715935e-05,
"loss": 0.3917,
"step": 23125
},
{
"epoch": 2.6970615671641793,
"grad_norm": 0.41979800453545263,
"learning_rate": 2.642804130310691e-05,
"loss": 0.4082,
"step": 23130
},
{
"epoch": 2.697644589552239,
"grad_norm": 0.46204630505241356,
"learning_rate": 2.641937514588786e-05,
"loss": 0.4238,
"step": 23135
},
{
"epoch": 2.6982276119402986,
"grad_norm": 0.4057527433244061,
"learning_rate": 2.6410709149347385e-05,
"loss": 0.3896,
"step": 23140
},
{
"epoch": 2.698810634328358,
"grad_norm": 0.39929869805403395,
"learning_rate": 2.640204331477401e-05,
"loss": 0.3862,
"step": 23145
},
{
"epoch": 2.699393656716418,
"grad_norm": 0.43679517893116127,
"learning_rate": 2.6393377643456284e-05,
"loss": 0.4329,
"step": 23150
},
{
"epoch": 2.6999766791044775,
"grad_norm": 0.4134303595321092,
"learning_rate": 2.6384712136682706e-05,
"loss": 0.3848,
"step": 23155
},
{
"epoch": 2.700559701492537,
"grad_norm": 0.40540442514472547,
"learning_rate": 2.6376046795741733e-05,
"loss": 0.4139,
"step": 23160
},
{
"epoch": 2.701142723880597,
"grad_norm": 0.40554525658034907,
"learning_rate": 2.6367381621921845e-05,
"loss": 0.4031,
"step": 23165
},
{
"epoch": 2.701725746268657,
"grad_norm": 0.435115298585795,
"learning_rate": 2.6358716616511446e-05,
"loss": 0.4148,
"step": 23170
},
{
"epoch": 2.7023087686567164,
"grad_norm": 0.3862926045381509,
"learning_rate": 2.635005178079896e-05,
"loss": 0.372,
"step": 23175
},
{
"epoch": 2.702891791044776,
"grad_norm": 0.40096781880358884,
"learning_rate": 2.6341387116072763e-05,
"loss": 0.3937,
"step": 23180
},
{
"epoch": 2.7034748134328357,
"grad_norm": 0.40341537040171827,
"learning_rate": 2.633272262362118e-05,
"loss": 0.3907,
"step": 23185
},
{
"epoch": 2.7040578358208958,
"grad_norm": 0.4253080292369856,
"learning_rate": 2.6324058304732574e-05,
"loss": 0.3989,
"step": 23190
},
{
"epoch": 2.7046408582089554,
"grad_norm": 0.4261722316781176,
"learning_rate": 2.6315394160695227e-05,
"loss": 0.4108,
"step": 23195
},
{
"epoch": 2.705223880597015,
"grad_norm": 0.40714567920789657,
"learning_rate": 2.630673019279742e-05,
"loss": 0.3914,
"step": 23200
},
{
"epoch": 2.7058069029850746,
"grad_norm": 0.388495867404363,
"learning_rate": 2.6298066402327404e-05,
"loss": 0.4058,
"step": 23205
},
{
"epoch": 2.7063899253731343,
"grad_norm": 0.4337222302072616,
"learning_rate": 2.6289402790573392e-05,
"loss": 0.4004,
"step": 23210
},
{
"epoch": 2.706972947761194,
"grad_norm": 0.44018535449347046,
"learning_rate": 2.6280739358823588e-05,
"loss": 0.4089,
"step": 23215
},
{
"epoch": 2.7075559701492535,
"grad_norm": 0.39877524769895384,
"learning_rate": 2.6272076108366163e-05,
"loss": 0.3916,
"step": 23220
},
{
"epoch": 2.7081389925373136,
"grad_norm": 0.45642928017337175,
"learning_rate": 2.6263413040489248e-05,
"loss": 0.3958,
"step": 23225
},
{
"epoch": 2.7087220149253732,
"grad_norm": 0.47021556383537044,
"learning_rate": 2.6254750156480973e-05,
"loss": 0.4572,
"step": 23230
},
{
"epoch": 2.709305037313433,
"grad_norm": 0.416273335439482,
"learning_rate": 2.6246087457629408e-05,
"loss": 0.391,
"step": 23235
},
{
"epoch": 2.7098880597014925,
"grad_norm": 0.3923420184469295,
"learning_rate": 2.623742494522264e-05,
"loss": 0.4073,
"step": 23240
},
{
"epoch": 2.710471082089552,
"grad_norm": 0.3877627467831005,
"learning_rate": 2.6228762620548675e-05,
"loss": 0.4021,
"step": 23245
},
{
"epoch": 2.711054104477612,
"grad_norm": 0.42093254685600145,
"learning_rate": 2.6220100484895527e-05,
"loss": 0.402,
"step": 23250
},
{
"epoch": 2.7116371268656714,
"grad_norm": 0.40869295789702276,
"learning_rate": 2.6211438539551187e-05,
"loss": 0.4239,
"step": 23255
},
{
"epoch": 2.7122201492537314,
"grad_norm": 0.43352026639842933,
"learning_rate": 2.620277678580358e-05,
"loss": 0.4344,
"step": 23260
},
{
"epoch": 2.712803171641791,
"grad_norm": 0.4345051480468157,
"learning_rate": 2.619411522494063e-05,
"loss": 0.4127,
"step": 23265
},
{
"epoch": 2.7133861940298507,
"grad_norm": 0.4123537753096545,
"learning_rate": 2.6185453858250242e-05,
"loss": 0.4251,
"step": 23270
},
{
"epoch": 2.7139692164179103,
"grad_norm": 0.4250618106607531,
"learning_rate": 2.6176792687020273e-05,
"loss": 0.369,
"step": 23275
},
{
"epoch": 2.71455223880597,
"grad_norm": 0.4151329426607874,
"learning_rate": 2.616813171253855e-05,
"loss": 0.4199,
"step": 23280
},
{
"epoch": 2.71513526119403,
"grad_norm": 0.38972508331106953,
"learning_rate": 2.615947093609288e-05,
"loss": 0.3615,
"step": 23285
},
{
"epoch": 2.7157182835820897,
"grad_norm": 0.45250019937154007,
"learning_rate": 2.615081035897104e-05,
"loss": 0.4137,
"step": 23290
},
{
"epoch": 2.7163013059701493,
"grad_norm": 0.4011154676934871,
"learning_rate": 2.614214998246077e-05,
"loss": 0.3812,
"step": 23295
},
{
"epoch": 2.716884328358209,
"grad_norm": 0.4040557255747825,
"learning_rate": 2.6133489807849786e-05,
"loss": 0.3944,
"step": 23300
},
{
"epoch": 2.7174673507462686,
"grad_norm": 0.39984922383296745,
"learning_rate": 2.6124829836425784e-05,
"loss": 0.3781,
"step": 23305
},
{
"epoch": 2.7180503731343286,
"grad_norm": 0.4112675512275878,
"learning_rate": 2.6116170069476397e-05,
"loss": 0.3903,
"step": 23310
},
{
"epoch": 2.718633395522388,
"grad_norm": 0.43349766500278497,
"learning_rate": 2.6107510508289274e-05,
"loss": 0.4204,
"step": 23315
},
{
"epoch": 2.719216417910448,
"grad_norm": 0.3985340664955324,
"learning_rate": 2.609885115415198e-05,
"loss": 0.3938,
"step": 23320
},
{
"epoch": 2.7197994402985075,
"grad_norm": 0.4119274080405407,
"learning_rate": 2.6090192008352115e-05,
"loss": 0.4101,
"step": 23325
},
{
"epoch": 2.720382462686567,
"grad_norm": 0.4231135539136476,
"learning_rate": 2.6081533072177183e-05,
"loss": 0.428,
"step": 23330
},
{
"epoch": 2.7209654850746268,
"grad_norm": 0.3865666551000909,
"learning_rate": 2.6072874346914688e-05,
"loss": 0.3987,
"step": 23335
},
{
"epoch": 2.7215485074626864,
"grad_norm": 0.43601065595854366,
"learning_rate": 2.6064215833852113e-05,
"loss": 0.3994,
"step": 23340
},
{
"epoch": 2.7221315298507465,
"grad_norm": 0.4505806506956824,
"learning_rate": 2.605555753427687e-05,
"loss": 0.4118,
"step": 23345
},
{
"epoch": 2.722714552238806,
"grad_norm": 0.4082580930394578,
"learning_rate": 2.6046899449476397e-05,
"loss": 0.4095,
"step": 23350
},
{
"epoch": 2.7232975746268657,
"grad_norm": 0.38769732223260783,
"learning_rate": 2.6038241580738053e-05,
"loss": 0.3926,
"step": 23355
},
{
"epoch": 2.7238805970149254,
"grad_norm": 0.4068613521392573,
"learning_rate": 2.602958392934917e-05,
"loss": 0.3833,
"step": 23360
},
{
"epoch": 2.724463619402985,
"grad_norm": 0.5974696661302847,
"learning_rate": 2.6020926496597082e-05,
"loss": 0.4046,
"step": 23365
},
{
"epoch": 2.7250466417910446,
"grad_norm": 0.4807652994716567,
"learning_rate": 2.601226928376904e-05,
"loss": 0.4328,
"step": 23370
},
{
"epoch": 2.7256296641791042,
"grad_norm": 0.40072634460980483,
"learning_rate": 2.600361229215231e-05,
"loss": 0.3638,
"step": 23375
},
{
"epoch": 2.7262126865671643,
"grad_norm": 0.4211773845340952,
"learning_rate": 2.5994955523034098e-05,
"loss": 0.4046,
"step": 23380
},
{
"epoch": 2.726795708955224,
"grad_norm": 0.39596086036641953,
"learning_rate": 2.598629897770157e-05,
"loss": 0.418,
"step": 23385
},
{
"epoch": 2.7273787313432836,
"grad_norm": 0.5036133136705389,
"learning_rate": 2.5977642657441893e-05,
"loss": 0.4182,
"step": 23390
},
{
"epoch": 2.727961753731343,
"grad_norm": 0.45720145788244165,
"learning_rate": 2.5968986563542153e-05,
"loss": 0.4097,
"step": 23395
},
{
"epoch": 2.728544776119403,
"grad_norm": 0.4589595028804862,
"learning_rate": 2.5960330697289447e-05,
"loss": 0.4094,
"step": 23400
},
{
"epoch": 2.729127798507463,
"grad_norm": 0.4067884402041407,
"learning_rate": 2.595167505997082e-05,
"loss": 0.3952,
"step": 23405
},
{
"epoch": 2.7297108208955225,
"grad_norm": 0.40815615139786815,
"learning_rate": 2.5943019652873267e-05,
"loss": 0.4175,
"step": 23410
},
{
"epoch": 2.730293843283582,
"grad_norm": 0.37160135826095847,
"learning_rate": 2.5934364477283786e-05,
"loss": 0.3883,
"step": 23415
},
{
"epoch": 2.730876865671642,
"grad_norm": 0.3894050185239279,
"learning_rate": 2.5925709534489295e-05,
"loss": 0.3664,
"step": 23420
},
{
"epoch": 2.7314598880597014,
"grad_norm": 0.4120147513178943,
"learning_rate": 2.5917054825776717e-05,
"loss": 0.3872,
"step": 23425
},
{
"epoch": 2.732042910447761,
"grad_norm": 0.4335842493381205,
"learning_rate": 2.5908400352432927e-05,
"loss": 0.4175,
"step": 23430
},
{
"epoch": 2.7326259328358207,
"grad_norm": 0.4138931421552768,
"learning_rate": 2.5899746115744743e-05,
"loss": 0.4105,
"step": 23435
},
{
"epoch": 2.7332089552238807,
"grad_norm": 0.740558249887021,
"learning_rate": 2.589109211699899e-05,
"loss": 0.4032,
"step": 23440
},
{
"epoch": 2.7337919776119404,
"grad_norm": 0.40855631044561286,
"learning_rate": 2.5882438357482414e-05,
"loss": 0.4267,
"step": 23445
},
{
"epoch": 2.734375,
"grad_norm": 0.4489663532548628,
"learning_rate": 2.5873784838481762e-05,
"loss": 0.3851,
"step": 23450
},
{
"epoch": 2.7349580223880596,
"grad_norm": 0.39658794798836106,
"learning_rate": 2.5865131561283718e-05,
"loss": 0.3966,
"step": 23455
},
{
"epoch": 2.7355410447761193,
"grad_norm": 0.4150587942209322,
"learning_rate": 2.5856478527174955e-05,
"loss": 0.3985,
"step": 23460
},
{
"epoch": 2.7361240671641793,
"grad_norm": 0.41526330251693905,
"learning_rate": 2.5847825737442088e-05,
"loss": 0.3993,
"step": 23465
},
{
"epoch": 2.736707089552239,
"grad_norm": 0.44656130500655006,
"learning_rate": 2.5839173193371697e-05,
"loss": 0.4259,
"step": 23470
},
{
"epoch": 2.7372901119402986,
"grad_norm": 0.40721729631087433,
"learning_rate": 2.583052089625035e-05,
"loss": 0.4037,
"step": 23475
},
{
"epoch": 2.737873134328358,
"grad_norm": 0.4165401357075778,
"learning_rate": 2.5821868847364534e-05,
"loss": 0.4097,
"step": 23480
},
{
"epoch": 2.738456156716418,
"grad_norm": 0.40950560526214086,
"learning_rate": 2.5813217048000753e-05,
"loss": 0.3985,
"step": 23485
},
{
"epoch": 2.7390391791044775,
"grad_norm": 0.43524417905378904,
"learning_rate": 2.5804565499445437e-05,
"loss": 0.4198,
"step": 23490
},
{
"epoch": 2.739622201492537,
"grad_norm": 0.4239421544275635,
"learning_rate": 2.5795914202984978e-05,
"loss": 0.3871,
"step": 23495
},
{
"epoch": 2.740205223880597,
"grad_norm": 0.4118146009144225,
"learning_rate": 2.578726315990576e-05,
"loss": 0.4263,
"step": 23500
},
{
"epoch": 2.740788246268657,
"grad_norm": 0.4187728226608671,
"learning_rate": 2.577861237149409e-05,
"loss": 0.4042,
"step": 23505
},
{
"epoch": 2.7413712686567164,
"grad_norm": 0.4064348141925177,
"learning_rate": 2.5769961839036277e-05,
"loss": 0.3832,
"step": 23510
},
{
"epoch": 2.741954291044776,
"grad_norm": 0.44349461881866076,
"learning_rate": 2.5761311563818564e-05,
"loss": 0.4232,
"step": 23515
},
{
"epoch": 2.7425373134328357,
"grad_norm": 0.40294997704414226,
"learning_rate": 2.575266154712715e-05,
"loss": 0.3995,
"step": 23520
},
{
"epoch": 2.7431203358208958,
"grad_norm": 0.4570692373043179,
"learning_rate": 2.574401179024823e-05,
"loss": 0.4493,
"step": 23525
},
{
"epoch": 2.7437033582089554,
"grad_norm": 0.39395181895347137,
"learning_rate": 2.5735362294467928e-05,
"loss": 0.4364,
"step": 23530
},
{
"epoch": 2.744286380597015,
"grad_norm": 0.38849911166313594,
"learning_rate": 2.5726713061072354e-05,
"loss": 0.404,
"step": 23535
},
{
"epoch": 2.7448694029850746,
"grad_norm": 0.4125750457549172,
"learning_rate": 2.571806409134756e-05,
"loss": 0.4082,
"step": 23540
},
{
"epoch": 2.7454524253731343,
"grad_norm": 0.3888922840518757,
"learning_rate": 2.5709415386579548e-05,
"loss": 0.4092,
"step": 23545
},
{
"epoch": 2.746035447761194,
"grad_norm": 0.4023705829034372,
"learning_rate": 2.570076694805432e-05,
"loss": 0.4065,
"step": 23550
},
{
"epoch": 2.7466184701492535,
"grad_norm": 0.5430678955248344,
"learning_rate": 2.5692118777057804e-05,
"loss": 0.4228,
"step": 23555
},
{
"epoch": 2.7472014925373136,
"grad_norm": 0.42836701838939134,
"learning_rate": 2.5683470874875913e-05,
"loss": 0.4181,
"step": 23560
},
{
"epoch": 2.7477845149253732,
"grad_norm": 0.39499707975033255,
"learning_rate": 2.56748232427945e-05,
"loss": 0.4027,
"step": 23565
},
{
"epoch": 2.748367537313433,
"grad_norm": 0.4048341595713592,
"learning_rate": 2.566617588209937e-05,
"loss": 0.3795,
"step": 23570
},
{
"epoch": 2.7489505597014925,
"grad_norm": 0.384570534778654,
"learning_rate": 2.5657528794076325e-05,
"loss": 0.3799,
"step": 23575
},
{
"epoch": 2.749533582089552,
"grad_norm": 0.4182858760300387,
"learning_rate": 2.564888198001109e-05,
"loss": 0.4312,
"step": 23580
},
{
"epoch": 2.750116604477612,
"grad_norm": 0.41137050486749877,
"learning_rate": 2.5640235441189374e-05,
"loss": 0.4163,
"step": 23585
},
{
"epoch": 2.7506996268656714,
"grad_norm": 0.3925721547324965,
"learning_rate": 2.563158917889683e-05,
"loss": 0.4183,
"step": 23590
},
{
"epoch": 2.7512826492537314,
"grad_norm": 0.4179279447881019,
"learning_rate": 2.562294319441907e-05,
"loss": 0.4082,
"step": 23595
},
{
"epoch": 2.751865671641791,
"grad_norm": 0.4357732605721109,
"learning_rate": 2.5614297489041673e-05,
"loss": 0.4213,
"step": 23600
},
{
"epoch": 2.7524486940298507,
"grad_norm": 0.4263211744746736,
"learning_rate": 2.5605652064050168e-05,
"loss": 0.4061,
"step": 23605
},
{
"epoch": 2.7530317164179103,
"grad_norm": 0.4279222660525636,
"learning_rate": 2.559700692073006e-05,
"loss": 0.3761,
"step": 23610
},
{
"epoch": 2.75361473880597,
"grad_norm": 0.417219076894595,
"learning_rate": 2.5588362060366788e-05,
"loss": 0.3771,
"step": 23615
},
{
"epoch": 2.75419776119403,
"grad_norm": 0.402566514565224,
"learning_rate": 2.5579717484245756e-05,
"loss": 0.3855,
"step": 23620
},
{
"epoch": 2.7547807835820897,
"grad_norm": 0.3921980022841419,
"learning_rate": 2.5571073193652344e-05,
"loss": 0.4033,
"step": 23625
},
{
"epoch": 2.7553638059701493,
"grad_norm": 0.4264633432451599,
"learning_rate": 2.556242918987185e-05,
"loss": 0.4137,
"step": 23630
},
{
"epoch": 2.755946828358209,
"grad_norm": 0.4329588491768397,
"learning_rate": 2.555378547418959e-05,
"loss": 0.4162,
"step": 23635
},
{
"epoch": 2.7565298507462686,
"grad_norm": 0.41367723409575946,
"learning_rate": 2.554514204789078e-05,
"loss": 0.3996,
"step": 23640
},
{
"epoch": 2.7571128731343286,
"grad_norm": 0.4052815832329333,
"learning_rate": 2.553649891226061e-05,
"loss": 0.4136,
"step": 23645
},
{
"epoch": 2.757695895522388,
"grad_norm": 0.39155926007138503,
"learning_rate": 2.5527856068584244e-05,
"loss": 0.3746,
"step": 23650
},
{
"epoch": 2.758278917910448,
"grad_norm": 0.4476494032793297,
"learning_rate": 2.5519213518146778e-05,
"loss": 0.4151,
"step": 23655
},
{
"epoch": 2.7588619402985075,
"grad_norm": 0.42318808533098395,
"learning_rate": 2.551057126223329e-05,
"loss": 0.4079,
"step": 23660
},
{
"epoch": 2.759444962686567,
"grad_norm": 0.419430489222338,
"learning_rate": 2.5501929302128795e-05,
"loss": 0.3974,
"step": 23665
},
{
"epoch": 2.7600279850746268,
"grad_norm": 0.43282775381006433,
"learning_rate": 2.5493287639118265e-05,
"loss": 0.4262,
"step": 23670
},
{
"epoch": 2.7606110074626864,
"grad_norm": 0.4323264282523569,
"learning_rate": 2.548464627448664e-05,
"loss": 0.4272,
"step": 23675
},
{
"epoch": 2.7611940298507465,
"grad_norm": 0.4353983566542406,
"learning_rate": 2.54760052095188e-05,
"loss": 0.3751,
"step": 23680
},
{
"epoch": 2.761777052238806,
"grad_norm": 0.39523963202263834,
"learning_rate": 2.54673644454996e-05,
"loss": 0.3989,
"step": 23685
},
{
"epoch": 2.7623600746268657,
"grad_norm": 0.4330013473539442,
"learning_rate": 2.545872398371383e-05,
"loss": 0.4311,
"step": 23690
},
{
"epoch": 2.7629430970149254,
"grad_norm": 0.41197784556490724,
"learning_rate": 2.5450083825446237e-05,
"loss": 0.3924,
"step": 23695
},
{
"epoch": 2.763526119402985,
"grad_norm": 0.4038801679236969,
"learning_rate": 2.544144397198155e-05,
"loss": 0.3977,
"step": 23700
},
{
"epoch": 2.7641091417910446,
"grad_norm": 0.45941878876760606,
"learning_rate": 2.5432804424604412e-05,
"loss": 0.4289,
"step": 23705
},
{
"epoch": 2.7646921641791042,
"grad_norm": 0.3773822659259436,
"learning_rate": 2.5424165184599457e-05,
"loss": 0.3872,
"step": 23710
},
{
"epoch": 2.7652751865671643,
"grad_norm": 0.4318969237293315,
"learning_rate": 2.541552625325125e-05,
"loss": 0.412,
"step": 23715
},
{
"epoch": 2.765858208955224,
"grad_norm": 0.40242158228990554,
"learning_rate": 2.5406887631844312e-05,
"loss": 0.4161,
"step": 23720
},
{
"epoch": 2.7664412313432836,
"grad_norm": 0.41685747516691346,
"learning_rate": 2.539824932166313e-05,
"loss": 0.3989,
"step": 23725
},
{
"epoch": 2.767024253731343,
"grad_norm": 0.5283443603392355,
"learning_rate": 2.5389611323992134e-05,
"loss": 0.4217,
"step": 23730
},
{
"epoch": 2.767607276119403,
"grad_norm": 0.4431415757580844,
"learning_rate": 2.538097364011572e-05,
"loss": 0.3959,
"step": 23735
},
{
"epoch": 2.768190298507463,
"grad_norm": 0.4708376365779287,
"learning_rate": 2.5372336271318225e-05,
"loss": 0.4275,
"step": 23740
},
{
"epoch": 2.7687733208955225,
"grad_norm": 0.3879746198686029,
"learning_rate": 2.5363699218883928e-05,
"loss": 0.4,
"step": 23745
},
{
"epoch": 2.769356343283582,
"grad_norm": 0.40532301443873264,
"learning_rate": 2.5355062484097103e-05,
"loss": 0.4123,
"step": 23750
},
{
"epoch": 2.769939365671642,
"grad_norm": 0.3868527746931474,
"learning_rate": 2.534642606824193e-05,
"loss": 0.4059,
"step": 23755
},
{
"epoch": 2.7705223880597014,
"grad_norm": 0.39234138283164033,
"learning_rate": 2.5337789972602566e-05,
"loss": 0.3997,
"step": 23760
},
{
"epoch": 2.771105410447761,
"grad_norm": 0.40105674009452147,
"learning_rate": 2.5329154198463122e-05,
"loss": 0.4219,
"step": 23765
},
{
"epoch": 2.7716884328358207,
"grad_norm": 0.4119086203387013,
"learning_rate": 2.5320518747107646e-05,
"loss": 0.4056,
"step": 23770
},
{
"epoch": 2.7722714552238807,
"grad_norm": 0.4351952534923432,
"learning_rate": 2.5311883619820148e-05,
"loss": 0.4275,
"step": 23775
},
{
"epoch": 2.7728544776119404,
"grad_norm": 0.4391144020073444,
"learning_rate": 2.530324881788459e-05,
"loss": 0.4205,
"step": 23780
},
{
"epoch": 2.7734375,
"grad_norm": 0.42911783395826564,
"learning_rate": 2.529461434258489e-05,
"loss": 0.3701,
"step": 23785
},
{
"epoch": 2.7740205223880596,
"grad_norm": 0.38659198334080536,
"learning_rate": 2.5285980195204906e-05,
"loss": 0.4171,
"step": 23790
},
{
"epoch": 2.7746035447761193,
"grad_norm": 0.4259465737601676,
"learning_rate": 2.5277346377028448e-05,
"loss": 0.4051,
"step": 23795
},
{
"epoch": 2.7751865671641793,
"grad_norm": 0.3839863991363711,
"learning_rate": 2.5268712889339296e-05,
"loss": 0.3869,
"step": 23800
},
{
"epoch": 2.775769589552239,
"grad_norm": 0.4374632446399871,
"learning_rate": 2.5260079733421155e-05,
"loss": 0.4058,
"step": 23805
},
{
"epoch": 2.7763526119402986,
"grad_norm": 0.43809648237627147,
"learning_rate": 2.5251446910557704e-05,
"loss": 0.4092,
"step": 23810
},
{
"epoch": 2.776935634328358,
"grad_norm": 0.43408596055973,
"learning_rate": 2.524281442203254e-05,
"loss": 0.4072,
"step": 23815
},
{
"epoch": 2.777518656716418,
"grad_norm": 0.4070905572092175,
"learning_rate": 2.5234182269129253e-05,
"loss": 0.3964,
"step": 23820
},
{
"epoch": 2.7781016791044775,
"grad_norm": 0.42091320895090806,
"learning_rate": 2.522555045313136e-05,
"loss": 0.3994,
"step": 23825
},
{
"epoch": 2.778684701492537,
"grad_norm": 0.40957912989869033,
"learning_rate": 2.5216918975322303e-05,
"loss": 0.4005,
"step": 23830
},
{
"epoch": 2.779267723880597,
"grad_norm": 0.42193901325138783,
"learning_rate": 2.520828783698554e-05,
"loss": 0.3921,
"step": 23835
},
{
"epoch": 2.779850746268657,
"grad_norm": 0.43234762135101035,
"learning_rate": 2.519965703940441e-05,
"loss": 0.3834,
"step": 23840
},
{
"epoch": 2.7804337686567164,
"grad_norm": 0.42060154814410344,
"learning_rate": 2.5191026583862237e-05,
"loss": 0.4232,
"step": 23845
},
{
"epoch": 2.781016791044776,
"grad_norm": 0.7083708125721425,
"learning_rate": 2.5182396471642287e-05,
"loss": 0.4051,
"step": 23850
},
{
"epoch": 2.7815998134328357,
"grad_norm": 0.387630736355572,
"learning_rate": 2.5173766704027775e-05,
"loss": 0.4121,
"step": 23855
},
{
"epoch": 2.7821828358208958,
"grad_norm": 0.4102084264934228,
"learning_rate": 2.5165137282301877e-05,
"loss": 0.4042,
"step": 23860
},
{
"epoch": 2.7827658582089554,
"grad_norm": 0.45040483006911913,
"learning_rate": 2.5156508207747686e-05,
"loss": 0.3834,
"step": 23865
},
{
"epoch": 2.783348880597015,
"grad_norm": 0.44049350009547306,
"learning_rate": 2.5147879481648266e-05,
"loss": 0.4058,
"step": 23870
},
{
"epoch": 2.7839319029850746,
"grad_norm": 0.41386777302135547,
"learning_rate": 2.5139251105286644e-05,
"loss": 0.3861,
"step": 23875
},
{
"epoch": 2.7845149253731343,
"grad_norm": 0.39641096536614573,
"learning_rate": 2.5130623079945754e-05,
"loss": 0.4048,
"step": 23880
},
{
"epoch": 2.785097947761194,
"grad_norm": 0.427528561224656,
"learning_rate": 2.5121995406908516e-05,
"loss": 0.3854,
"step": 23885
},
{
"epoch": 2.7856809701492535,
"grad_norm": 0.38980591755760313,
"learning_rate": 2.511336808745778e-05,
"loss": 0.3876,
"step": 23890
},
{
"epoch": 2.7862639925373136,
"grad_norm": 0.417593608155887,
"learning_rate": 2.510474112287633e-05,
"loss": 0.4048,
"step": 23895
},
{
"epoch": 2.7868470149253732,
"grad_norm": 0.3897415184789091,
"learning_rate": 2.5096114514446934e-05,
"loss": 0.3997,
"step": 23900
},
{
"epoch": 2.787430037313433,
"grad_norm": 0.4135504576927545,
"learning_rate": 2.508748826345227e-05,
"loss": 0.4142,
"step": 23905
},
{
"epoch": 2.7880130597014925,
"grad_norm": 0.4454685363310767,
"learning_rate": 2.5078862371175e-05,
"loss": 0.4176,
"step": 23910
},
{
"epoch": 2.788596082089552,
"grad_norm": 0.408765100671447,
"learning_rate": 2.507023683889769e-05,
"loss": 0.404,
"step": 23915
},
{
"epoch": 2.789179104477612,
"grad_norm": 2.566801092984802,
"learning_rate": 2.5061611667902878e-05,
"loss": 0.4013,
"step": 23920
},
{
"epoch": 2.7897621268656714,
"grad_norm": 0.6071793929067169,
"learning_rate": 2.505298685947306e-05,
"loss": 0.3831,
"step": 23925
},
{
"epoch": 2.7903451492537314,
"grad_norm": 0.42578873893408165,
"learning_rate": 2.504436241489064e-05,
"loss": 0.3892,
"step": 23930
},
{
"epoch": 2.790928171641791,
"grad_norm": 0.39103629170693305,
"learning_rate": 2.503573833543801e-05,
"loss": 0.424,
"step": 23935
},
{
"epoch": 2.7915111940298507,
"grad_norm": 0.47204331901509455,
"learning_rate": 2.5027114622397473e-05,
"loss": 0.4448,
"step": 23940
},
{
"epoch": 2.7920942164179103,
"grad_norm": 0.39664647146837567,
"learning_rate": 2.5018491277051297e-05,
"loss": 0.4265,
"step": 23945
},
{
"epoch": 2.79267723880597,
"grad_norm": 0.3965111167317444,
"learning_rate": 2.50098683006817e-05,
"loss": 0.4209,
"step": 23950
},
{
"epoch": 2.79326026119403,
"grad_norm": 0.42966543838020704,
"learning_rate": 2.500124569457082e-05,
"loss": 0.4326,
"step": 23955
},
{
"epoch": 2.7938432835820897,
"grad_norm": 0.39810968385462714,
"learning_rate": 2.4992623460000763e-05,
"loss": 0.3838,
"step": 23960
},
{
"epoch": 2.7944263059701493,
"grad_norm": 0.45218513820816264,
"learning_rate": 2.4984001598253586e-05,
"loss": 0.405,
"step": 23965
},
{
"epoch": 2.795009328358209,
"grad_norm": 0.4656333542507009,
"learning_rate": 2.497538011061125e-05,
"loss": 0.4084,
"step": 23970
},
{
"epoch": 2.7955923507462686,
"grad_norm": 0.4112521610946821,
"learning_rate": 2.4966758998355717e-05,
"loss": 0.4116,
"step": 23975
},
{
"epoch": 2.7961753731343286,
"grad_norm": 0.4288806020954192,
"learning_rate": 2.495813826276884e-05,
"loss": 0.403,
"step": 23980
},
{
"epoch": 2.796758395522388,
"grad_norm": 0.4224093193422178,
"learning_rate": 2.4949517905132454e-05,
"loss": 0.4129,
"step": 23985
},
{
"epoch": 2.797341417910448,
"grad_norm": 0.4205978319383233,
"learning_rate": 2.4940897926728314e-05,
"loss": 0.3884,
"step": 23990
},
{
"epoch": 2.7979244402985075,
"grad_norm": 0.3986308142627854,
"learning_rate": 2.4932278328838138e-05,
"loss": 0.3889,
"step": 23995
},
{
"epoch": 2.798507462686567,
"grad_norm": 0.4322059827673749,
"learning_rate": 2.4923659112743576e-05,
"loss": 0.3988,
"step": 24000
},
{
"epoch": 2.7990904850746268,
"grad_norm": 0.4144370652573071,
"learning_rate": 2.4915040279726204e-05,
"loss": 0.4069,
"step": 24005
},
{
"epoch": 2.7996735074626864,
"grad_norm": 0.3952033697853189,
"learning_rate": 2.490642183106759e-05,
"loss": 0.3922,
"step": 24010
},
{
"epoch": 2.8002565298507465,
"grad_norm": 0.45683165418522576,
"learning_rate": 2.48978037680492e-05,
"loss": 0.4473,
"step": 24015
},
{
"epoch": 2.800839552238806,
"grad_norm": 0.44948343362373777,
"learning_rate": 2.4889186091952444e-05,
"loss": 0.4055,
"step": 24020
},
{
"epoch": 2.8014225746268657,
"grad_norm": 0.47008610412249724,
"learning_rate": 2.4880568804058713e-05,
"loss": 0.381,
"step": 24025
},
{
"epoch": 2.8020055970149254,
"grad_norm": 0.4719458016582111,
"learning_rate": 2.48719519056493e-05,
"loss": 0.4198,
"step": 24030
},
{
"epoch": 2.802588619402985,
"grad_norm": 0.41391021805188194,
"learning_rate": 2.486333539800546e-05,
"loss": 0.4273,
"step": 24035
},
{
"epoch": 2.8031716417910446,
"grad_norm": 0.44497227363620745,
"learning_rate": 2.485471928240839e-05,
"loss": 0.3848,
"step": 24040
},
{
"epoch": 2.8037546641791042,
"grad_norm": 0.43542455319928247,
"learning_rate": 2.4846103560139206e-05,
"loss": 0.4007,
"step": 24045
},
{
"epoch": 2.8043376865671643,
"grad_norm": 0.382972498062499,
"learning_rate": 2.4837488232479005e-05,
"loss": 0.364,
"step": 24050
},
{
"epoch": 2.804920708955224,
"grad_norm": 0.41513659207359027,
"learning_rate": 2.4828873300708783e-05,
"loss": 0.4198,
"step": 24055
},
{
"epoch": 2.8055037313432836,
"grad_norm": 0.4152909338801933,
"learning_rate": 2.4820258766109515e-05,
"loss": 0.394,
"step": 24060
},
{
"epoch": 2.806086753731343,
"grad_norm": 0.41993421455879915,
"learning_rate": 2.48116446299621e-05,
"loss": 0.4269,
"step": 24065
},
{
"epoch": 2.806669776119403,
"grad_norm": 0.41329144124473244,
"learning_rate": 2.4803030893547357e-05,
"loss": 0.3952,
"step": 24070
},
{
"epoch": 2.807252798507463,
"grad_norm": 0.4189307879832587,
"learning_rate": 2.4794417558146093e-05,
"loss": 0.3887,
"step": 24075
},
{
"epoch": 2.8078358208955225,
"grad_norm": 0.4596873833252724,
"learning_rate": 2.4785804625039005e-05,
"loss": 0.387,
"step": 24080
},
{
"epoch": 2.808418843283582,
"grad_norm": 0.44059313885271306,
"learning_rate": 2.4777192095506775e-05,
"loss": 0.4037,
"step": 24085
},
{
"epoch": 2.809001865671642,
"grad_norm": 0.4489989021415904,
"learning_rate": 2.4768579970829985e-05,
"loss": 0.4079,
"step": 24090
},
{
"epoch": 2.8095848880597014,
"grad_norm": 0.46829103663309113,
"learning_rate": 2.4759968252289183e-05,
"loss": 0.4293,
"step": 24095
},
{
"epoch": 2.810167910447761,
"grad_norm": 0.4263120061814068,
"learning_rate": 2.4751356941164855e-05,
"loss": 0.3909,
"step": 24100
},
{
"epoch": 2.8107509328358207,
"grad_norm": 0.38124904574498125,
"learning_rate": 2.4742746038737404e-05,
"loss": 0.3833,
"step": 24105
},
{
"epoch": 2.8113339552238807,
"grad_norm": 0.4117583097365348,
"learning_rate": 2.4734135546287208e-05,
"loss": 0.4136,
"step": 24110
},
{
"epoch": 2.8119169776119404,
"grad_norm": 0.3947387742870654,
"learning_rate": 2.4725525465094557e-05,
"loss": 0.3972,
"step": 24115
},
{
"epoch": 2.8125,
"grad_norm": 0.4668608715868671,
"learning_rate": 2.4716915796439678e-05,
"loss": 0.4328,
"step": 24120
},
{
"epoch": 2.8130830223880596,
"grad_norm": 0.4278804795115331,
"learning_rate": 2.4708306541602765e-05,
"loss": 0.3988,
"step": 24125
},
{
"epoch": 2.8136660447761193,
"grad_norm": 0.44823585767839486,
"learning_rate": 2.4699697701863916e-05,
"loss": 0.4414,
"step": 24130
},
{
"epoch": 2.8142490671641793,
"grad_norm": 0.3792493693971136,
"learning_rate": 2.4691089278503188e-05,
"loss": 0.3826,
"step": 24135
},
{
"epoch": 2.814832089552239,
"grad_norm": 0.4250893569584375,
"learning_rate": 2.4682481272800572e-05,
"loss": 0.3922,
"step": 24140
},
{
"epoch": 2.8154151119402986,
"grad_norm": 0.529672446563032,
"learning_rate": 2.4673873686035996e-05,
"loss": 0.3969,
"step": 24145
},
{
"epoch": 2.815998134328358,
"grad_norm": 0.44283279861816877,
"learning_rate": 2.4665266519489328e-05,
"loss": 0.4131,
"step": 24150
},
{
"epoch": 2.816581156716418,
"grad_norm": 0.3953088226301666,
"learning_rate": 2.4656659774440365e-05,
"loss": 0.3914,
"step": 24155
},
{
"epoch": 2.8171641791044775,
"grad_norm": 0.42262510875290904,
"learning_rate": 2.4648053452168857e-05,
"loss": 0.3999,
"step": 24160
},
{
"epoch": 2.817747201492537,
"grad_norm": 0.39812095649226686,
"learning_rate": 2.4639447553954464e-05,
"loss": 0.3922,
"step": 24165
},
{
"epoch": 2.818330223880597,
"grad_norm": 0.4020501363321913,
"learning_rate": 2.463084208107682e-05,
"loss": 0.4158,
"step": 24170
},
{
"epoch": 2.818913246268657,
"grad_norm": 0.3872723225939495,
"learning_rate": 2.462223703481547e-05,
"loss": 0.3971,
"step": 24175
},
{
"epoch": 2.8194962686567164,
"grad_norm": 0.4555501116525867,
"learning_rate": 2.4613632416449893e-05,
"loss": 0.394,
"step": 24180
},
{
"epoch": 2.820079291044776,
"grad_norm": 0.4259640134600756,
"learning_rate": 2.4605028227259535e-05,
"loss": 0.4212,
"step": 24185
},
{
"epoch": 2.8206623134328357,
"grad_norm": 0.420385036695274,
"learning_rate": 2.4596424468523728e-05,
"loss": 0.3927,
"step": 24190
},
{
"epoch": 2.8212453358208958,
"grad_norm": 0.3909871637600437,
"learning_rate": 2.458782114152179e-05,
"loss": 0.4026,
"step": 24195
},
{
"epoch": 2.8218283582089554,
"grad_norm": 0.40224347395428345,
"learning_rate": 2.4579218247532947e-05,
"loss": 0.3792,
"step": 24200
},
{
"epoch": 2.822411380597015,
"grad_norm": 0.40495405228442655,
"learning_rate": 2.4570615787836358e-05,
"loss": 0.4073,
"step": 24205
},
{
"epoch": 2.8229944029850746,
"grad_norm": 0.42990772331215554,
"learning_rate": 2.4562013763711145e-05,
"loss": 0.4101,
"step": 24210
},
{
"epoch": 2.8235774253731343,
"grad_norm": 0.38516181306189795,
"learning_rate": 2.4553412176436323e-05,
"loss": 0.419,
"step": 24215
},
{
"epoch": 2.824160447761194,
"grad_norm": 0.4147705650040051,
"learning_rate": 2.4544811027290893e-05,
"loss": 0.3911,
"step": 24220
},
{
"epoch": 2.8247434701492535,
"grad_norm": 0.3942782293350187,
"learning_rate": 2.4536210317553744e-05,
"loss": 0.4037,
"step": 24225
},
{
"epoch": 2.8253264925373136,
"grad_norm": 0.4407757655623333,
"learning_rate": 2.452761004850371e-05,
"loss": 0.4152,
"step": 24230
},
{
"epoch": 2.8259095149253732,
"grad_norm": 0.4092463833129098,
"learning_rate": 2.4519010221419604e-05,
"loss": 0.3924,
"step": 24235
},
{
"epoch": 2.826492537313433,
"grad_norm": 0.4417188417776021,
"learning_rate": 2.4510410837580106e-05,
"loss": 0.4024,
"step": 24240
},
{
"epoch": 2.8270755597014925,
"grad_norm": 0.3988022692782764,
"learning_rate": 2.4501811898263878e-05,
"loss": 0.3807,
"step": 24245
},
{
"epoch": 2.827658582089552,
"grad_norm": 0.41403541909483943,
"learning_rate": 2.4493213404749493e-05,
"loss": 0.4077,
"step": 24250
},
{
"epoch": 2.828241604477612,
"grad_norm": 0.4526465180974963,
"learning_rate": 2.4484615358315456e-05,
"loss": 0.3814,
"step": 24255
},
{
"epoch": 2.8288246268656714,
"grad_norm": 0.4090896726844662,
"learning_rate": 2.447601776024024e-05,
"loss": 0.3579,
"step": 24260
},
{
"epoch": 2.8294076492537314,
"grad_norm": 0.389366581958165,
"learning_rate": 2.4467420611802194e-05,
"loss": 0.4005,
"step": 24265
},
{
"epoch": 2.829990671641791,
"grad_norm": 0.4193735141283348,
"learning_rate": 2.4458823914279662e-05,
"loss": 0.4058,
"step": 24270
},
{
"epoch": 2.8305736940298507,
"grad_norm": 0.4072016054515705,
"learning_rate": 2.445022766895087e-05,
"loss": 0.3911,
"step": 24275
},
{
"epoch": 2.8311567164179103,
"grad_norm": 0.3726787597832839,
"learning_rate": 2.4441631877093995e-05,
"loss": 0.3953,
"step": 24280
},
{
"epoch": 2.83173973880597,
"grad_norm": 0.4384868084146645,
"learning_rate": 2.443303653998717e-05,
"loss": 0.4294,
"step": 24285
},
{
"epoch": 2.83232276119403,
"grad_norm": 0.40754963352572265,
"learning_rate": 2.442444165890842e-05,
"loss": 0.4082,
"step": 24290
},
{
"epoch": 2.8329057835820897,
"grad_norm": 0.3817317415158813,
"learning_rate": 2.441584723513573e-05,
"loss": 0.397,
"step": 24295
},
{
"epoch": 2.8334888059701493,
"grad_norm": 0.4685953873915643,
"learning_rate": 2.4407253269947006e-05,
"loss": 0.4315,
"step": 24300
},
{
"epoch": 2.834071828358209,
"grad_norm": 0.43119704571582806,
"learning_rate": 2.4398659764620086e-05,
"loss": 0.4131,
"step": 24305
},
{
"epoch": 2.8346548507462686,
"grad_norm": 0.44133039518900996,
"learning_rate": 2.4390066720432746e-05,
"loss": 0.4377,
"step": 24310
},
{
"epoch": 2.8352378731343286,
"grad_norm": 0.40069317868975407,
"learning_rate": 2.438147413866269e-05,
"loss": 0.4012,
"step": 24315
},
{
"epoch": 2.835820895522388,
"grad_norm": 0.4130630853182529,
"learning_rate": 2.437288202058755e-05,
"loss": 0.4127,
"step": 24320
},
{
"epoch": 2.836403917910448,
"grad_norm": 0.40646135376030335,
"learning_rate": 2.4364290367484888e-05,
"loss": 0.4012,
"step": 24325
},
{
"epoch": 2.8369869402985075,
"grad_norm": 0.4158453066326914,
"learning_rate": 2.4355699180632207e-05,
"loss": 0.4292,
"step": 24330
},
{
"epoch": 2.837569962686567,
"grad_norm": 0.39902524368616477,
"learning_rate": 2.4347108461306935e-05,
"loss": 0.414,
"step": 24335
},
{
"epoch": 2.8381529850746268,
"grad_norm": 0.42744606003488217,
"learning_rate": 2.4338518210786416e-05,
"loss": 0.4204,
"step": 24340
},
{
"epoch": 2.8387360074626864,
"grad_norm": 0.41169833535803,
"learning_rate": 2.432992843034796e-05,
"loss": 0.431,
"step": 24345
},
{
"epoch": 2.8393190298507465,
"grad_norm": 0.43257474420696607,
"learning_rate": 2.4321339121268766e-05,
"loss": 0.3942,
"step": 24350
},
{
"epoch": 2.839902052238806,
"grad_norm": 0.419108016090665,
"learning_rate": 2.431275028482599e-05,
"loss": 0.4003,
"step": 24355
},
{
"epoch": 2.8404850746268657,
"grad_norm": 0.3856661560241362,
"learning_rate": 2.430416192229672e-05,
"loss": 0.4097,
"step": 24360
},
{
"epoch": 2.8410680970149254,
"grad_norm": 0.4326362279524336,
"learning_rate": 2.4295574034957935e-05,
"loss": 0.4102,
"step": 24365
},
{
"epoch": 2.841651119402985,
"grad_norm": 0.3791202902718328,
"learning_rate": 2.42869866240866e-05,
"loss": 0.3939,
"step": 24370
},
{
"epoch": 2.8422341417910446,
"grad_norm": 0.42997859107080455,
"learning_rate": 2.427839969095957e-05,
"loss": 0.3957,
"step": 24375
},
{
"epoch": 2.8428171641791042,
"grad_norm": 0.4363154249413559,
"learning_rate": 2.4269813236853632e-05,
"loss": 0.4104,
"step": 24380
},
{
"epoch": 2.8434001865671643,
"grad_norm": 0.43630810723657876,
"learning_rate": 2.4261227263045527e-05,
"loss": 0.3868,
"step": 24385
},
{
"epoch": 2.843983208955224,
"grad_norm": 0.4242797707850999,
"learning_rate": 2.4252641770811886e-05,
"loss": 0.3888,
"step": 24390
},
{
"epoch": 2.8445662313432836,
"grad_norm": 0.3786323125165545,
"learning_rate": 2.4244056761429307e-05,
"loss": 0.4153,
"step": 24395
},
{
"epoch": 2.845149253731343,
"grad_norm": 0.392344661545243,
"learning_rate": 2.423547223617429e-05,
"loss": 0.4219,
"step": 24400
},
{
"epoch": 2.845732276119403,
"grad_norm": 0.39341726507792363,
"learning_rate": 2.4226888196323268e-05,
"loss": 0.3955,
"step": 24405
},
{
"epoch": 2.846315298507463,
"grad_norm": 0.4365540625801256,
"learning_rate": 2.4218304643152617e-05,
"loss": 0.3962,
"step": 24410
},
{
"epoch": 2.8468983208955225,
"grad_norm": 0.41052541016137806,
"learning_rate": 2.4209721577938617e-05,
"loss": 0.394,
"step": 24415
},
{
"epoch": 2.847481343283582,
"grad_norm": 0.41405925493167905,
"learning_rate": 2.42011390019575e-05,
"loss": 0.4275,
"step": 24420
},
{
"epoch": 2.848064365671642,
"grad_norm": 0.3900653204452662,
"learning_rate": 2.419255691648541e-05,
"loss": 0.3977,
"step": 24425
},
{
"epoch": 2.8486473880597014,
"grad_norm": 0.4053471709622438,
"learning_rate": 2.4183975322798407e-05,
"loss": 0.3752,
"step": 24430
},
{
"epoch": 2.849230410447761,
"grad_norm": 0.39641963474272346,
"learning_rate": 2.417539422217251e-05,
"loss": 0.3933,
"step": 24435
},
{
"epoch": 2.8498134328358207,
"grad_norm": 0.4072103759711589,
"learning_rate": 2.4166813615883625e-05,
"loss": 0.4002,
"step": 24440
},
{
"epoch": 2.8503964552238807,
"grad_norm": 0.3890747829198251,
"learning_rate": 2.415823350520764e-05,
"loss": 0.4037,
"step": 24445
},
{
"epoch": 2.8509794776119404,
"grad_norm": 0.4028973809674288,
"learning_rate": 2.4149653891420304e-05,
"loss": 0.4016,
"step": 24450
},
{
"epoch": 2.8515625,
"grad_norm": 0.4037724399117923,
"learning_rate": 2.4141074775797335e-05,
"loss": 0.4039,
"step": 24455
},
{
"epoch": 2.8521455223880596,
"grad_norm": 0.5343163758156394,
"learning_rate": 2.4132496159614366e-05,
"loss": 0.4258,
"step": 24460
},
{
"epoch": 2.8527285447761193,
"grad_norm": 0.4234157935262023,
"learning_rate": 2.412391804414694e-05,
"loss": 0.3924,
"step": 24465
},
{
"epoch": 2.8533115671641793,
"grad_norm": 0.3733049551298622,
"learning_rate": 2.4115340430670574e-05,
"loss": 0.3863,
"step": 24470
},
{
"epoch": 2.853894589552239,
"grad_norm": 0.39046157315620744,
"learning_rate": 2.410676332046065e-05,
"loss": 0.4081,
"step": 24475
},
{
"epoch": 2.8544776119402986,
"grad_norm": 0.4394507404345183,
"learning_rate": 2.4098186714792504e-05,
"loss": 0.4132,
"step": 24480
},
{
"epoch": 2.855060634328358,
"grad_norm": 0.43282596564227105,
"learning_rate": 2.408961061494141e-05,
"loss": 0.4294,
"step": 24485
},
{
"epoch": 2.855643656716418,
"grad_norm": 0.4368857860325901,
"learning_rate": 2.408103502218253e-05,
"loss": 0.4134,
"step": 24490
},
{
"epoch": 2.8562266791044775,
"grad_norm": 0.40514376788318834,
"learning_rate": 2.407245993779099e-05,
"loss": 0.4126,
"step": 24495
},
{
"epoch": 2.856809701492537,
"grad_norm": 0.4689409168661143,
"learning_rate": 2.4063885363041822e-05,
"loss": 0.4228,
"step": 24500
},
{
"epoch": 2.857392723880597,
"grad_norm": 0.42742228212990213,
"learning_rate": 2.405531129920997e-05,
"loss": 0.4172,
"step": 24505
},
{
"epoch": 2.857975746268657,
"grad_norm": 0.41398247557989326,
"learning_rate": 2.4046737747570326e-05,
"loss": 0.4053,
"step": 24510
},
{
"epoch": 2.8585587686567164,
"grad_norm": 0.3850251313792589,
"learning_rate": 2.4038164709397694e-05,
"loss": 0.4174,
"step": 24515
},
{
"epoch": 2.859141791044776,
"grad_norm": 0.4101770377436773,
"learning_rate": 2.4029592185966804e-05,
"loss": 0.4024,
"step": 24520
},
{
"epoch": 2.8597248134328357,
"grad_norm": 0.389933969483989,
"learning_rate": 2.40210201785523e-05,
"loss": 0.4039,
"step": 24525
},
{
"epoch": 2.8603078358208958,
"grad_norm": 0.3864105506674258,
"learning_rate": 2.4012448688428768e-05,
"loss": 0.3806,
"step": 24530
},
{
"epoch": 2.8608908582089554,
"grad_norm": 0.41064619983657913,
"learning_rate": 2.4003877716870698e-05,
"loss": 0.3943,
"step": 24535
},
{
"epoch": 2.861473880597015,
"grad_norm": 0.41526288263186806,
"learning_rate": 2.399530726515251e-05,
"loss": 0.413,
"step": 24540
},
{
"epoch": 2.8620569029850746,
"grad_norm": 0.37817741893593837,
"learning_rate": 2.3986737334548564e-05,
"loss": 0.3732,
"step": 24545
},
{
"epoch": 2.8626399253731343,
"grad_norm": 0.4381206894440303,
"learning_rate": 2.397816792633311e-05,
"loss": 0.4261,
"step": 24550
},
{
"epoch": 2.863222947761194,
"grad_norm": 0.40738716108025813,
"learning_rate": 2.396959904178034e-05,
"loss": 0.4174,
"step": 24555
},
{
"epoch": 2.8638059701492535,
"grad_norm": 0.390537642181435,
"learning_rate": 2.396103068216437e-05,
"loss": 0.3995,
"step": 24560
},
{
"epoch": 2.8643889925373136,
"grad_norm": 0.4159336953346919,
"learning_rate": 2.395246284875922e-05,
"loss": 0.4041,
"step": 24565
},
{
"epoch": 2.8649720149253732,
"grad_norm": 0.3936853002100043,
"learning_rate": 2.3943895542838868e-05,
"loss": 0.3947,
"step": 24570
},
{
"epoch": 2.865555037313433,
"grad_norm": 0.40261557417640986,
"learning_rate": 2.3935328765677172e-05,
"loss": 0.3773,
"step": 24575
},
{
"epoch": 2.8661380597014925,
"grad_norm": 0.42449360027466265,
"learning_rate": 2.3926762518547928e-05,
"loss": 0.3903,
"step": 24580
},
{
"epoch": 2.866721082089552,
"grad_norm": 0.41778284206752914,
"learning_rate": 2.3918196802724874e-05,
"loss": 0.4046,
"step": 24585
},
{
"epoch": 2.867304104477612,
"grad_norm": 0.43191991233987104,
"learning_rate": 2.3909631619481626e-05,
"loss": 0.4109,
"step": 24590
},
{
"epoch": 2.8678871268656714,
"grad_norm": 0.4133181416603124,
"learning_rate": 2.390106697009176e-05,
"loss": 0.4089,
"step": 24595
},
{
"epoch": 2.8684701492537314,
"grad_norm": 0.44767587322082275,
"learning_rate": 2.3892502855828762e-05,
"loss": 0.3982,
"step": 24600
},
{
"epoch": 2.869053171641791,
"grad_norm": 0.4109104703305783,
"learning_rate": 2.388393927796601e-05,
"loss": 0.4083,
"step": 24605
},
{
"epoch": 2.8696361940298507,
"grad_norm": 0.4054090044863133,
"learning_rate": 2.387537623777686e-05,
"loss": 0.414,
"step": 24610
},
{
"epoch": 2.8702192164179103,
"grad_norm": 0.5099691428529806,
"learning_rate": 2.3866813736534517e-05,
"loss": 0.4245,
"step": 24615
},
{
"epoch": 2.87080223880597,
"grad_norm": 0.47498753884610195,
"learning_rate": 2.3858251775512176e-05,
"loss": 0.3947,
"step": 24620
},
{
"epoch": 2.87138526119403,
"grad_norm": 0.41510822650922674,
"learning_rate": 2.3849690355982895e-05,
"loss": 0.4114,
"step": 24625
},
{
"epoch": 2.8719682835820897,
"grad_norm": 0.3892083976859538,
"learning_rate": 2.384112947921968e-05,
"loss": 0.4255,
"step": 24630
},
{
"epoch": 2.8725513059701493,
"grad_norm": 0.38156201149519997,
"learning_rate": 2.3832569146495464e-05,
"loss": 0.4061,
"step": 24635
},
{
"epoch": 2.873134328358209,
"grad_norm": 0.3981291909865196,
"learning_rate": 2.3824009359083073e-05,
"loss": 0.38,
"step": 24640
},
{
"epoch": 2.8737173507462686,
"grad_norm": 0.4616752471859324,
"learning_rate": 2.3815450118255272e-05,
"loss": 0.4084,
"step": 24645
},
{
"epoch": 2.8743003731343286,
"grad_norm": 0.4272833105107833,
"learning_rate": 2.380689142528474e-05,
"loss": 0.4165,
"step": 24650
},
{
"epoch": 2.874883395522388,
"grad_norm": 0.4121014836198729,
"learning_rate": 2.3798333281444057e-05,
"loss": 0.4342,
"step": 24655
},
{
"epoch": 2.875466417910448,
"grad_norm": 0.42304423381586476,
"learning_rate": 2.378977568800576e-05,
"loss": 0.3939,
"step": 24660
},
{
"epoch": 2.8760494402985075,
"grad_norm": 0.405604211463261,
"learning_rate": 2.3781218646242264e-05,
"loss": 0.4179,
"step": 24665
},
{
"epoch": 2.876632462686567,
"grad_norm": 0.410351260508775,
"learning_rate": 2.3772662157425925e-05,
"loss": 0.427,
"step": 24670
},
{
"epoch": 2.8772154850746268,
"grad_norm": 0.4444117347524044,
"learning_rate": 2.376410622282902e-05,
"loss": 0.4125,
"step": 24675
},
{
"epoch": 2.8777985074626864,
"grad_norm": 0.4016325681634763,
"learning_rate": 2.375555084372371e-05,
"loss": 0.3851,
"step": 24680
},
{
"epoch": 2.8783815298507465,
"grad_norm": 0.3821765509428547,
"learning_rate": 2.3746996021382124e-05,
"loss": 0.4251,
"step": 24685
},
{
"epoch": 2.878964552238806,
"grad_norm": 0.4076947203602498,
"learning_rate": 2.3738441757076268e-05,
"loss": 0.3982,
"step": 24690
},
{
"epoch": 2.8795475746268657,
"grad_norm": 0.4005527493041794,
"learning_rate": 2.3729888052078088e-05,
"loss": 0.394,
"step": 24695
},
{
"epoch": 2.8801305970149254,
"grad_norm": 0.4060689225266477,
"learning_rate": 2.3721334907659424e-05,
"loss": 0.4106,
"step": 24700
},
{
"epoch": 2.880713619402985,
"grad_norm": 0.42718017392739294,
"learning_rate": 2.371278232509206e-05,
"loss": 0.3948,
"step": 24705
},
{
"epoch": 2.8812966417910446,
"grad_norm": 0.41383665802467334,
"learning_rate": 2.370423030564768e-05,
"loss": 0.387,
"step": 24710
},
{
"epoch": 2.8818796641791042,
"grad_norm": 0.4451802175506528,
"learning_rate": 2.3695678850597876e-05,
"loss": 0.3882,
"step": 24715
},
{
"epoch": 2.8824626865671643,
"grad_norm": 0.4055142167603417,
"learning_rate": 2.368712796121419e-05,
"loss": 0.3815,
"step": 24720
},
{
"epoch": 2.883045708955224,
"grad_norm": 0.43359950461642005,
"learning_rate": 2.3678577638768035e-05,
"loss": 0.3981,
"step": 24725
},
{
"epoch": 2.8836287313432836,
"grad_norm": 0.39948141435362394,
"learning_rate": 2.367002788453077e-05,
"loss": 0.4103,
"step": 24730
},
{
"epoch": 2.884211753731343,
"grad_norm": 0.4229619925725585,
"learning_rate": 2.366147869977367e-05,
"loss": 0.4014,
"step": 24735
},
{
"epoch": 2.884794776119403,
"grad_norm": 0.4028658884119076,
"learning_rate": 2.3652930085767904e-05,
"loss": 0.4064,
"step": 24740
},
{
"epoch": 2.885377798507463,
"grad_norm": 0.5643632036476247,
"learning_rate": 2.364438204378458e-05,
"loss": 0.3807,
"step": 24745
},
{
"epoch": 2.8859608208955225,
"grad_norm": 0.4195003817703374,
"learning_rate": 2.3635834575094705e-05,
"loss": 0.4188,
"step": 24750
},
{
"epoch": 2.886543843283582,
"grad_norm": 0.41802101909621364,
"learning_rate": 2.3627287680969195e-05,
"loss": 0.4077,
"step": 24755
},
{
"epoch": 2.887126865671642,
"grad_norm": 0.4061500737991145,
"learning_rate": 2.3618741362678915e-05,
"loss": 0.3806,
"step": 24760
},
{
"epoch": 2.8877098880597014,
"grad_norm": 0.40416655755195474,
"learning_rate": 2.3610195621494595e-05,
"loss": 0.386,
"step": 24765
},
{
"epoch": 2.888292910447761,
"grad_norm": 0.4368691942878869,
"learning_rate": 2.360165045868693e-05,
"loss": 0.3968,
"step": 24770
},
{
"epoch": 2.8888759328358207,
"grad_norm": 0.37835566532333276,
"learning_rate": 2.3593105875526488e-05,
"loss": 0.3757,
"step": 24775
},
{
"epoch": 2.8894589552238807,
"grad_norm": 0.4364902532815583,
"learning_rate": 2.358456187328376e-05,
"loss": 0.4083,
"step": 24780
},
{
"epoch": 2.8900419776119404,
"grad_norm": 0.3782607959674674,
"learning_rate": 2.3576018453229182e-05,
"loss": 0.3691,
"step": 24785
},
{
"epoch": 2.890625,
"grad_norm": 0.3797036800433863,
"learning_rate": 2.3567475616633046e-05,
"loss": 0.3827,
"step": 24790
},
{
"epoch": 2.8912080223880596,
"grad_norm": 0.40776317385220623,
"learning_rate": 2.3558933364765617e-05,
"loss": 0.4085,
"step": 24795
},
{
"epoch": 2.8917910447761193,
"grad_norm": 0.43262439495951677,
"learning_rate": 2.355039169889704e-05,
"loss": 0.395,
"step": 24800
},
{
"epoch": 2.8923740671641793,
"grad_norm": 0.4180800612198426,
"learning_rate": 2.3541850620297368e-05,
"loss": 0.392,
"step": 24805
},
{
"epoch": 2.892957089552239,
"grad_norm": 0.48634111522693574,
"learning_rate": 2.3533310130236592e-05,
"loss": 0.4063,
"step": 24810
},
{
"epoch": 2.8935401119402986,
"grad_norm": 0.41627741997295004,
"learning_rate": 2.3524770229984587e-05,
"loss": 0.4063,
"step": 24815
},
{
"epoch": 2.894123134328358,
"grad_norm": 0.41709604498467673,
"learning_rate": 2.3516230920811166e-05,
"loss": 0.3991,
"step": 24820
},
{
"epoch": 2.894706156716418,
"grad_norm": 0.40823029921163884,
"learning_rate": 2.3507692203986042e-05,
"loss": 0.3722,
"step": 24825
},
{
"epoch": 2.8952891791044775,
"grad_norm": 0.42155178738534094,
"learning_rate": 2.3499154080778823e-05,
"loss": 0.4107,
"step": 24830
},
{
"epoch": 2.895872201492537,
"grad_norm": 0.4166192474304213,
"learning_rate": 2.349061655245907e-05,
"loss": 0.4029,
"step": 24835
},
{
"epoch": 2.896455223880597,
"grad_norm": 0.3954396494556229,
"learning_rate": 2.3482079620296223e-05,
"loss": 0.3856,
"step": 24840
},
{
"epoch": 2.897038246268657,
"grad_norm": 0.3977847598870714,
"learning_rate": 2.347354328555963e-05,
"loss": 0.3888,
"step": 24845
},
{
"epoch": 2.8976212686567164,
"grad_norm": 0.4046284318378812,
"learning_rate": 2.3465007549518576e-05,
"loss": 0.3964,
"step": 24850
},
{
"epoch": 2.898204291044776,
"grad_norm": 0.433611208865864,
"learning_rate": 2.3456472413442248e-05,
"loss": 0.4367,
"step": 24855
},
{
"epoch": 2.8987873134328357,
"grad_norm": 0.3976024854002812,
"learning_rate": 2.3447937878599725e-05,
"loss": 0.4025,
"step": 24860
},
{
"epoch": 2.8993703358208958,
"grad_norm": 0.3989868868191118,
"learning_rate": 2.3439403946260007e-05,
"loss": 0.4096,
"step": 24865
},
{
"epoch": 2.8999533582089554,
"grad_norm": 0.42939386585697015,
"learning_rate": 2.343087061769203e-05,
"loss": 0.3946,
"step": 24870
},
{
"epoch": 2.900536380597015,
"grad_norm": 0.44495417024776424,
"learning_rate": 2.3422337894164594e-05,
"loss": 0.4117,
"step": 24875
},
{
"epoch": 2.9011194029850746,
"grad_norm": 0.41446939070729183,
"learning_rate": 2.3413805776946453e-05,
"loss": 0.4124,
"step": 24880
},
{
"epoch": 2.9017024253731343,
"grad_norm": 0.4057898275011049,
"learning_rate": 2.3405274267306244e-05,
"loss": 0.4058,
"step": 24885
},
{
"epoch": 2.902285447761194,
"grad_norm": 0.4097613234138069,
"learning_rate": 2.3396743366512508e-05,
"loss": 0.4219,
"step": 24890
},
{
"epoch": 2.9028684701492535,
"grad_norm": 0.4509067786161942,
"learning_rate": 2.338821307583373e-05,
"loss": 0.3958,
"step": 24895
},
{
"epoch": 2.9034514925373136,
"grad_norm": 0.4277230430884208,
"learning_rate": 2.337968339653826e-05,
"loss": 0.405,
"step": 24900
},
{
"epoch": 2.9040345149253732,
"grad_norm": 0.4297228967842881,
"learning_rate": 2.33711543298944e-05,
"loss": 0.423,
"step": 24905
},
{
"epoch": 2.904617537313433,
"grad_norm": 0.4076491715263942,
"learning_rate": 2.3362625877170336e-05,
"loss": 0.3952,
"step": 24910
},
{
"epoch": 2.9052005597014925,
"grad_norm": 0.42301387777350424,
"learning_rate": 2.3354098039634154e-05,
"loss": 0.4155,
"step": 24915
},
{
"epoch": 2.905783582089552,
"grad_norm": 0.4015962877581815,
"learning_rate": 2.3345570818553874e-05,
"loss": 0.4018,
"step": 24920
},
{
"epoch": 2.906366604477612,
"grad_norm": 0.41774728592458704,
"learning_rate": 2.3337044215197408e-05,
"loss": 0.3948,
"step": 24925
},
{
"epoch": 2.9069496268656714,
"grad_norm": 0.44990262245478574,
"learning_rate": 2.3328518230832587e-05,
"loss": 0.424,
"step": 24930
},
{
"epoch": 2.9075326492537314,
"grad_norm": 0.4413178522099354,
"learning_rate": 2.3319992866727136e-05,
"loss": 0.3854,
"step": 24935
},
{
"epoch": 2.908115671641791,
"grad_norm": 0.4161768623084127,
"learning_rate": 2.331146812414869e-05,
"loss": 0.3973,
"step": 24940
},
{
"epoch": 2.9086986940298507,
"grad_norm": 0.4568466173744008,
"learning_rate": 2.330294400436481e-05,
"loss": 0.4282,
"step": 24945
},
{
"epoch": 2.9092817164179103,
"grad_norm": 0.43582365445869903,
"learning_rate": 2.329442050864293e-05,
"loss": 0.3934,
"step": 24950
},
{
"epoch": 2.90986473880597,
"grad_norm": 0.41216198065629644,
"learning_rate": 2.3285897638250437e-05,
"loss": 0.3868,
"step": 24955
},
{
"epoch": 2.91044776119403,
"grad_norm": 0.4521166492531563,
"learning_rate": 2.3277375394454594e-05,
"loss": 0.4322,
"step": 24960
},
{
"epoch": 2.9110307835820897,
"grad_norm": 0.37478266654025316,
"learning_rate": 2.326885377852256e-05,
"loss": 0.3909,
"step": 24965
},
{
"epoch": 2.9116138059701493,
"grad_norm": 0.39825062099728903,
"learning_rate": 2.326033279172144e-05,
"loss": 0.4127,
"step": 24970
},
{
"epoch": 2.912196828358209,
"grad_norm": 0.4310244343023203,
"learning_rate": 2.32518124353182e-05,
"loss": 0.411,
"step": 24975
},
{
"epoch": 2.9127798507462686,
"grad_norm": 0.40561810191760284,
"learning_rate": 2.324329271057976e-05,
"loss": 0.3954,
"step": 24980
},
{
"epoch": 2.9133628731343286,
"grad_norm": 0.3886447880786942,
"learning_rate": 2.3234773618772916e-05,
"loss": 0.4035,
"step": 24985
},
{
"epoch": 2.913945895522388,
"grad_norm": 0.39291361858950735,
"learning_rate": 2.322625516116435e-05,
"loss": 0.3977,
"step": 24990
},
{
"epoch": 2.914528917910448,
"grad_norm": 0.4289910244574198,
"learning_rate": 2.3217737339020706e-05,
"loss": 0.4142,
"step": 24995
},
{
"epoch": 2.9151119402985075,
"grad_norm": 0.41804607838764657,
"learning_rate": 2.3209220153608486e-05,
"loss": 0.4253,
"step": 25000
},
{
"epoch": 2.915694962686567,
"grad_norm": 0.380525476332318,
"learning_rate": 2.3200703606194126e-05,
"loss": 0.3988,
"step": 25005
},
{
"epoch": 2.9162779850746268,
"grad_norm": 0.4146357362301396,
"learning_rate": 2.3192187698043944e-05,
"loss": 0.4056,
"step": 25010
},
{
"epoch": 2.9168610074626864,
"grad_norm": 0.4419473513927475,
"learning_rate": 2.3183672430424163e-05,
"loss": 0.4335,
"step": 25015
},
{
"epoch": 2.9174440298507465,
"grad_norm": 0.4497441028504768,
"learning_rate": 2.3175157804600954e-05,
"loss": 0.4157,
"step": 25020
},
{
"epoch": 2.918027052238806,
"grad_norm": 0.4231823433948403,
"learning_rate": 2.3166643821840338e-05,
"loss": 0.3964,
"step": 25025
},
{
"epoch": 2.9186100746268657,
"grad_norm": 0.40692064793035276,
"learning_rate": 2.3158130483408262e-05,
"loss": 0.384,
"step": 25030
},
{
"epoch": 2.9191930970149254,
"grad_norm": 0.43646246071138917,
"learning_rate": 2.3149617790570587e-05,
"loss": 0.4171,
"step": 25035
},
{
"epoch": 2.919776119402985,
"grad_norm": 0.41868527060090205,
"learning_rate": 2.3141105744593065e-05,
"loss": 0.3638,
"step": 25040
},
{
"epoch": 2.9203591417910446,
"grad_norm": 0.3950288733939245,
"learning_rate": 2.3132594346741353e-05,
"loss": 0.3996,
"step": 25045
},
{
"epoch": 2.9209421641791042,
"grad_norm": 0.3900419470979898,
"learning_rate": 2.3124083598281022e-05,
"loss": 0.3701,
"step": 25050
},
{
"epoch": 2.9215251865671643,
"grad_norm": 0.3757075298206466,
"learning_rate": 2.3115573500477534e-05,
"loss": 0.3826,
"step": 25055
},
{
"epoch": 2.922108208955224,
"grad_norm": 0.43993979242127856,
"learning_rate": 2.310706405459625e-05,
"loss": 0.3769,
"step": 25060
},
{
"epoch": 2.9226912313432836,
"grad_norm": 0.42997594377083986,
"learning_rate": 2.3098555261902455e-05,
"loss": 0.4117,
"step": 25065
},
{
"epoch": 2.923274253731343,
"grad_norm": 0.3630604906423459,
"learning_rate": 2.3090047123661324e-05,
"loss": 0.3755,
"step": 25070
},
{
"epoch": 2.923857276119403,
"grad_norm": 0.3984197416436645,
"learning_rate": 2.308153964113793e-05,
"loss": 0.4096,
"step": 25075
},
{
"epoch": 2.924440298507463,
"grad_norm": 0.4005189029459354,
"learning_rate": 2.3073032815597263e-05,
"loss": 0.3732,
"step": 25080
},
{
"epoch": 2.9250233208955225,
"grad_norm": 0.42754499865938367,
"learning_rate": 2.3064526648304195e-05,
"loss": 0.4026,
"step": 25085
},
{
"epoch": 2.925606343283582,
"grad_norm": 0.39659037663555,
"learning_rate": 2.3056021140523516e-05,
"loss": 0.4012,
"step": 25090
},
{
"epoch": 2.926189365671642,
"grad_norm": 0.3857244207492492,
"learning_rate": 2.304751629351992e-05,
"loss": 0.376,
"step": 25095
},
{
"epoch": 2.9267723880597014,
"grad_norm": 0.4091159243869105,
"learning_rate": 2.3039012108557982e-05,
"loss": 0.3889,
"step": 25100
},
{
"epoch": 2.927355410447761,
"grad_norm": 0.4027380642874116,
"learning_rate": 2.3030508586902215e-05,
"loss": 0.413,
"step": 25105
},
{
"epoch": 2.9279384328358207,
"grad_norm": 0.38937355535511614,
"learning_rate": 2.3022005729817e-05,
"loss": 0.3832,
"step": 25110
},
{
"epoch": 2.9285214552238807,
"grad_norm": 0.41596054645460284,
"learning_rate": 2.301350353856661e-05,
"loss": 0.4048,
"step": 25115
},
{
"epoch": 2.9291044776119404,
"grad_norm": 0.4161171037388307,
"learning_rate": 2.3005002014415274e-05,
"loss": 0.4167,
"step": 25120
},
{
"epoch": 2.9296875,
"grad_norm": 0.3833302448550085,
"learning_rate": 2.2996501158627054e-05,
"loss": 0.3982,
"step": 25125
},
{
"epoch": 2.9302705223880596,
"grad_norm": 0.4438540143346973,
"learning_rate": 2.2988000972465978e-05,
"loss": 0.4306,
"step": 25130
},
{
"epoch": 2.9308535447761193,
"grad_norm": 0.43239374779471784,
"learning_rate": 2.297950145719593e-05,
"loss": 0.4431,
"step": 25135
},
{
"epoch": 2.9314365671641793,
"grad_norm": 0.47247131842109463,
"learning_rate": 2.297100261408069e-05,
"loss": 0.3979,
"step": 25140
},
{
"epoch": 2.932019589552239,
"grad_norm": 0.40091967591663,
"learning_rate": 2.2962504444383974e-05,
"loss": 0.3908,
"step": 25145
},
{
"epoch": 2.9326026119402986,
"grad_norm": 0.3794271000855443,
"learning_rate": 2.295400694936937e-05,
"loss": 0.3978,
"step": 25150
},
{
"epoch": 2.933185634328358,
"grad_norm": 0.4095246121458622,
"learning_rate": 2.2945510130300386e-05,
"loss": 0.4215,
"step": 25155
},
{
"epoch": 2.933768656716418,
"grad_norm": 0.4064170878907522,
"learning_rate": 2.2937013988440405e-05,
"loss": 0.4178,
"step": 25160
},
{
"epoch": 2.9343516791044775,
"grad_norm": 0.4118465548109363,
"learning_rate": 2.2928518525052717e-05,
"loss": 0.4167,
"step": 25165
},
{
"epoch": 2.934934701492537,
"grad_norm": 0.41806234804780734,
"learning_rate": 2.2920023741400533e-05,
"loss": 0.4042,
"step": 25170
},
{
"epoch": 2.935517723880597,
"grad_norm": 0.42840618861931157,
"learning_rate": 2.2911529638746927e-05,
"loss": 0.3817,
"step": 25175
},
{
"epoch": 2.936100746268657,
"grad_norm": 0.41502354390655183,
"learning_rate": 2.2903036218354912e-05,
"loss": 0.3994,
"step": 25180
},
{
"epoch": 2.9366837686567164,
"grad_norm": 0.42591051491474585,
"learning_rate": 2.289454348148736e-05,
"loss": 0.4108,
"step": 25185
},
{
"epoch": 2.937266791044776,
"grad_norm": 0.41491598081280956,
"learning_rate": 2.288605142940707e-05,
"loss": 0.3917,
"step": 25190
},
{
"epoch": 2.9378498134328357,
"grad_norm": 0.3745700162861839,
"learning_rate": 2.2877560063376723e-05,
"loss": 0.4062,
"step": 25195
},
{
"epoch": 2.9384328358208958,
"grad_norm": 0.41726980475028025,
"learning_rate": 2.2869069384658908e-05,
"loss": 0.4291,
"step": 25200
},
{
"epoch": 2.9390158582089554,
"grad_norm": 0.5516879898324801,
"learning_rate": 2.2860579394516107e-05,
"loss": 0.419,
"step": 25205
},
{
"epoch": 2.939598880597015,
"grad_norm": 0.3848375565823875,
"learning_rate": 2.2852090094210698e-05,
"loss": 0.4064,
"step": 25210
},
{
"epoch": 2.9401819029850746,
"grad_norm": 0.4109948064761437,
"learning_rate": 2.2843601485004955e-05,
"loss": 0.3992,
"step": 25215
},
{
"epoch": 2.9407649253731343,
"grad_norm": 0.3927265043294491,
"learning_rate": 2.283511356816106e-05,
"loss": 0.4042,
"step": 25220
},
{
"epoch": 2.941347947761194,
"grad_norm": 0.48121892872118954,
"learning_rate": 2.2826626344941083e-05,
"loss": 0.3892,
"step": 25225
},
{
"epoch": 2.9419309701492535,
"grad_norm": 0.37256851975056027,
"learning_rate": 2.2818139816607e-05,
"loss": 0.3747,
"step": 25230
},
{
"epoch": 2.9425139925373136,
"grad_norm": 0.3921876104660543,
"learning_rate": 2.2809653984420652e-05,
"loss": 0.3831,
"step": 25235
},
{
"epoch": 2.9430970149253732,
"grad_norm": 0.4266465103397456,
"learning_rate": 2.280116884964383e-05,
"loss": 0.4239,
"step": 25240
},
{
"epoch": 2.943680037313433,
"grad_norm": 0.3947150176195217,
"learning_rate": 2.279268441353818e-05,
"loss": 0.3994,
"step": 25245
},
{
"epoch": 2.9442630597014925,
"grad_norm": 0.43091467973147646,
"learning_rate": 2.2784200677365242e-05,
"loss": 0.4288,
"step": 25250
},
{
"epoch": 2.944846082089552,
"grad_norm": 0.4142103610308874,
"learning_rate": 2.2775717642386497e-05,
"loss": 0.4131,
"step": 25255
},
{
"epoch": 2.945429104477612,
"grad_norm": 0.4215738737978956,
"learning_rate": 2.276723530986327e-05,
"loss": 0.413,
"step": 25260
},
{
"epoch": 2.9460121268656714,
"grad_norm": 0.4043071366559255,
"learning_rate": 2.27587536810568e-05,
"loss": 0.4046,
"step": 25265
},
{
"epoch": 2.9465951492537314,
"grad_norm": 0.39963288308005285,
"learning_rate": 2.2750272757228235e-05,
"loss": 0.3928,
"step": 25270
},
{
"epoch": 2.947178171641791,
"grad_norm": 0.4260079204293212,
"learning_rate": 2.27417925396386e-05,
"loss": 0.3976,
"step": 25275
},
{
"epoch": 2.9477611940298507,
"grad_norm": 0.4026489185676219,
"learning_rate": 2.273331302954883e-05,
"loss": 0.418,
"step": 25280
},
{
"epoch": 2.9483442164179103,
"grad_norm": 0.4131040866705125,
"learning_rate": 2.2724834228219742e-05,
"loss": 0.3959,
"step": 25285
},
{
"epoch": 2.94892723880597,
"grad_norm": 0.4178283482729982,
"learning_rate": 2.271635613691205e-05,
"loss": 0.3839,
"step": 25290
},
{
"epoch": 2.94951026119403,
"grad_norm": 0.419560877318633,
"learning_rate": 2.2707878756886368e-05,
"loss": 0.4201,
"step": 25295
},
{
"epoch": 2.9500932835820897,
"grad_norm": 0.42295331082686,
"learning_rate": 2.26994020894032e-05,
"loss": 0.4135,
"step": 25300
},
{
"epoch": 2.9506763059701493,
"grad_norm": 0.4007879495508421,
"learning_rate": 2.2690926135722946e-05,
"loss": 0.4063,
"step": 25305
},
{
"epoch": 2.951259328358209,
"grad_norm": 0.40328354566363045,
"learning_rate": 2.2682450897105905e-05,
"loss": 0.3961,
"step": 25310
},
{
"epoch": 2.9518423507462686,
"grad_norm": 0.38707822579641404,
"learning_rate": 2.2673976374812246e-05,
"loss": 0.3886,
"step": 25315
},
{
"epoch": 2.9524253731343286,
"grad_norm": 0.4193700698413242,
"learning_rate": 2.266550257010207e-05,
"loss": 0.4045,
"step": 25320
},
{
"epoch": 2.953008395522388,
"grad_norm": 0.3828309533087151,
"learning_rate": 2.265702948423534e-05,
"loss": 0.3874,
"step": 25325
},
{
"epoch": 2.953591417910448,
"grad_norm": 0.4138945989886876,
"learning_rate": 2.2648557118471918e-05,
"loss": 0.4097,
"step": 25330
},
{
"epoch": 2.9541744402985075,
"grad_norm": 0.39832165005252257,
"learning_rate": 2.264008547407158e-05,
"loss": 0.3731,
"step": 25335
},
{
"epoch": 2.954757462686567,
"grad_norm": 0.4221079463034623,
"learning_rate": 2.2631614552293963e-05,
"loss": 0.4295,
"step": 25340
},
{
"epoch": 2.9553404850746268,
"grad_norm": 0.4399266627497746,
"learning_rate": 2.2623144354398623e-05,
"loss": 0.4045,
"step": 25345
},
{
"epoch": 2.9559235074626864,
"grad_norm": 0.4028460067458374,
"learning_rate": 2.2614674881644974e-05,
"loss": 0.4032,
"step": 25350
},
{
"epoch": 2.9565065298507465,
"grad_norm": 0.40086976663568546,
"learning_rate": 2.2606206135292385e-05,
"loss": 0.3883,
"step": 25355
},
{
"epoch": 2.957089552238806,
"grad_norm": 0.40263627128283286,
"learning_rate": 2.2597738116600048e-05,
"loss": 0.4199,
"step": 25360
},
{
"epoch": 2.9576725746268657,
"grad_norm": 0.46214462516979704,
"learning_rate": 2.2589270826827073e-05,
"loss": 0.4086,
"step": 25365
},
{
"epoch": 2.9582555970149254,
"grad_norm": 0.39466493675004233,
"learning_rate": 2.2580804267232484e-05,
"loss": 0.3769,
"step": 25370
},
{
"epoch": 2.958838619402985,
"grad_norm": 0.4021769345625866,
"learning_rate": 2.257233843907517e-05,
"loss": 0.3894,
"step": 25375
},
{
"epoch": 2.9594216417910446,
"grad_norm": 0.38507883277133476,
"learning_rate": 2.2563873343613916e-05,
"loss": 0.4235,
"step": 25380
},
{
"epoch": 2.9600046641791042,
"grad_norm": 0.41325075109978926,
"learning_rate": 2.2555408982107407e-05,
"loss": 0.413,
"step": 25385
},
{
"epoch": 2.9605876865671643,
"grad_norm": 0.44675494871627636,
"learning_rate": 2.2546945355814196e-05,
"loss": 0.3839,
"step": 25390
},
{
"epoch": 2.961170708955224,
"grad_norm": 0.39067882987873165,
"learning_rate": 2.2538482465992762e-05,
"loss": 0.4286,
"step": 25395
},
{
"epoch": 2.9617537313432836,
"grad_norm": 0.43093610928308723,
"learning_rate": 2.2530020313901446e-05,
"loss": 0.3864,
"step": 25400
},
{
"epoch": 2.962336753731343,
"grad_norm": 0.39795231184485813,
"learning_rate": 2.2521558900798495e-05,
"loss": 0.3943,
"step": 25405
},
{
"epoch": 2.962919776119403,
"grad_norm": 0.38531472883509843,
"learning_rate": 2.2513098227942032e-05,
"loss": 0.3795,
"step": 25410
},
{
"epoch": 2.963502798507463,
"grad_norm": 0.42197054661839406,
"learning_rate": 2.250463829659008e-05,
"loss": 0.3997,
"step": 25415
},
{
"epoch": 2.9640858208955225,
"grad_norm": 0.42359026424902746,
"learning_rate": 2.249617910800056e-05,
"loss": 0.3913,
"step": 25420
},
{
"epoch": 2.964668843283582,
"grad_norm": 0.41927514123524673,
"learning_rate": 2.248772066343125e-05,
"loss": 0.4009,
"step": 25425
},
{
"epoch": 2.965251865671642,
"grad_norm": 0.44238902546771147,
"learning_rate": 2.2479262964139863e-05,
"loss": 0.4091,
"step": 25430
},
{
"epoch": 2.9658348880597014,
"grad_norm": 0.4055168047041742,
"learning_rate": 2.2470806011383972e-05,
"loss": 0.3854,
"step": 25435
},
{
"epoch": 2.966417910447761,
"grad_norm": 0.44548897158782064,
"learning_rate": 2.2462349806421035e-05,
"loss": 0.4153,
"step": 25440
},
{
"epoch": 2.9670009328358207,
"grad_norm": 0.40984566677196516,
"learning_rate": 2.245389435050842e-05,
"loss": 0.4085,
"step": 25445
},
{
"epoch": 2.9675839552238807,
"grad_norm": 0.37770958816231737,
"learning_rate": 2.244543964490336e-05,
"loss": 0.3793,
"step": 25450
},
{
"epoch": 2.9681669776119404,
"grad_norm": 0.40014186988741113,
"learning_rate": 2.2436985690863004e-05,
"loss": 0.4069,
"step": 25455
},
{
"epoch": 2.96875,
"grad_norm": 0.4271740503864899,
"learning_rate": 2.2428532489644368e-05,
"loss": 0.4004,
"step": 25460
},
{
"epoch": 2.9693330223880596,
"grad_norm": 0.4287412853262523,
"learning_rate": 2.2420080042504348e-05,
"loss": 0.4088,
"step": 25465
},
{
"epoch": 2.9699160447761193,
"grad_norm": 0.47085217031650956,
"learning_rate": 2.2411628350699766e-05,
"loss": 0.4047,
"step": 25470
},
{
"epoch": 2.9704990671641793,
"grad_norm": 0.4236414913472614,
"learning_rate": 2.2403177415487285e-05,
"loss": 0.3855,
"step": 25475
},
{
"epoch": 2.971082089552239,
"grad_norm": 0.44921836393514036,
"learning_rate": 2.2394727238123497e-05,
"loss": 0.4238,
"step": 25480
},
{
"epoch": 2.9716651119402986,
"grad_norm": 0.43653869539233847,
"learning_rate": 2.2386277819864853e-05,
"loss": 0.4049,
"step": 25485
},
{
"epoch": 2.972248134328358,
"grad_norm": 0.4453227108407065,
"learning_rate": 2.23778291619677e-05,
"loss": 0.4113,
"step": 25490
},
{
"epoch": 2.972831156716418,
"grad_norm": 0.5524259039595061,
"learning_rate": 2.2369381265688277e-05,
"loss": 0.4009,
"step": 25495
},
{
"epoch": 2.9734141791044775,
"grad_norm": 0.4118860361096215,
"learning_rate": 2.236093413228269e-05,
"loss": 0.4271,
"step": 25500
},
{
"epoch": 2.973997201492537,
"grad_norm": 0.40652814888997635,
"learning_rate": 2.2352487763006975e-05,
"loss": 0.4235,
"step": 25505
},
{
"epoch": 2.974580223880597,
"grad_norm": 0.4374563967976496,
"learning_rate": 2.2344042159117006e-05,
"loss": 0.4284,
"step": 25510
},
{
"epoch": 2.975163246268657,
"grad_norm": 0.42160551742662283,
"learning_rate": 2.2335597321868568e-05,
"loss": 0.4074,
"step": 25515
},
{
"epoch": 2.9757462686567164,
"grad_norm": 0.38660997279551207,
"learning_rate": 2.2327153252517323e-05,
"loss": 0.3876,
"step": 25520
},
{
"epoch": 2.976329291044776,
"grad_norm": 0.4126166538415622,
"learning_rate": 2.2318709952318822e-05,
"loss": 0.4142,
"step": 25525
},
{
"epoch": 2.9769123134328357,
"grad_norm": 0.3872941861425115,
"learning_rate": 2.2310267422528523e-05,
"loss": 0.3934,
"step": 25530
},
{
"epoch": 2.9774953358208958,
"grad_norm": 0.40355541577838977,
"learning_rate": 2.2301825664401733e-05,
"loss": 0.4182,
"step": 25535
},
{
"epoch": 2.9780783582089554,
"grad_norm": 0.483726374003381,
"learning_rate": 2.2293384679193645e-05,
"loss": 0.4213,
"step": 25540
},
{
"epoch": 2.978661380597015,
"grad_norm": 0.43331721004443324,
"learning_rate": 2.228494446815939e-05,
"loss": 0.4137,
"step": 25545
},
{
"epoch": 2.9792444029850746,
"grad_norm": 0.4004588618503616,
"learning_rate": 2.2276505032553912e-05,
"loss": 0.3982,
"step": 25550
},
{
"epoch": 2.9798274253731343,
"grad_norm": 0.4222765181742688,
"learning_rate": 2.2268066373632096e-05,
"loss": 0.4272,
"step": 25555
},
{
"epoch": 2.980410447761194,
"grad_norm": 0.45113868243041527,
"learning_rate": 2.2259628492648676e-05,
"loss": 0.4128,
"step": 25560
},
{
"epoch": 2.9809934701492535,
"grad_norm": 0.4342499839855839,
"learning_rate": 2.2251191390858295e-05,
"loss": 0.3826,
"step": 25565
},
{
"epoch": 2.9815764925373136,
"grad_norm": 0.4266366236984472,
"learning_rate": 2.224275506951547e-05,
"loss": 0.4174,
"step": 25570
},
{
"epoch": 2.9821595149253732,
"grad_norm": 0.407381288290989,
"learning_rate": 2.2234319529874586e-05,
"loss": 0.3967,
"step": 25575
},
{
"epoch": 2.982742537313433,
"grad_norm": 0.3899195848917077,
"learning_rate": 2.2225884773189936e-05,
"loss": 0.4168,
"step": 25580
},
{
"epoch": 2.9833255597014925,
"grad_norm": 0.3990786929492579,
"learning_rate": 2.221745080071569e-05,
"loss": 0.3774,
"step": 25585
},
{
"epoch": 2.983908582089552,
"grad_norm": 0.4012067435433367,
"learning_rate": 2.2209017613705908e-05,
"loss": 0.4093,
"step": 25590
},
{
"epoch": 2.984491604477612,
"grad_norm": 0.4829709734822829,
"learning_rate": 2.22005852134145e-05,
"loss": 0.4028,
"step": 25595
},
{
"epoch": 2.9850746268656714,
"grad_norm": 0.42554208792579185,
"learning_rate": 2.2192153601095293e-05,
"loss": 0.4084,
"step": 25600
},
{
"epoch": 2.9856576492537314,
"grad_norm": 0.3953817743750672,
"learning_rate": 2.2183722778002004e-05,
"loss": 0.377,
"step": 25605
},
{
"epoch": 2.986240671641791,
"grad_norm": 0.4166501312880979,
"learning_rate": 2.2175292745388186e-05,
"loss": 0.3996,
"step": 25610
},
{
"epoch": 2.9868236940298507,
"grad_norm": 0.4538563083315286,
"learning_rate": 2.2166863504507336e-05,
"loss": 0.4162,
"step": 25615
},
{
"epoch": 2.9874067164179103,
"grad_norm": 0.42389743691607706,
"learning_rate": 2.2158435056612775e-05,
"loss": 0.4036,
"step": 25620
},
{
"epoch": 2.98798973880597,
"grad_norm": 0.4116072944395864,
"learning_rate": 2.215000740295774e-05,
"loss": 0.4144,
"step": 25625
},
{
"epoch": 2.98857276119403,
"grad_norm": 0.40491099460562413,
"learning_rate": 2.2141580544795353e-05,
"loss": 0.3677,
"step": 25630
},
{
"epoch": 2.9891557835820897,
"grad_norm": 0.41021211807311764,
"learning_rate": 2.2133154483378587e-05,
"loss": 0.3931,
"step": 25635
},
{
"epoch": 2.9897388059701493,
"grad_norm": 0.4142741077006574,
"learning_rate": 2.2124729219960343e-05,
"loss": 0.4042,
"step": 25640
},
{
"epoch": 2.990321828358209,
"grad_norm": 0.5130815958463901,
"learning_rate": 2.211630475579336e-05,
"loss": 0.4139,
"step": 25645
},
{
"epoch": 2.9909048507462686,
"grad_norm": 0.41636737963451764,
"learning_rate": 2.2107881092130266e-05,
"loss": 0.4167,
"step": 25650
},
{
"epoch": 2.9914878731343286,
"grad_norm": 0.37704740496619205,
"learning_rate": 2.209945823022361e-05,
"loss": 0.4094,
"step": 25655
},
{
"epoch": 2.992070895522388,
"grad_norm": 0.44399220632690417,
"learning_rate": 2.2091036171325754e-05,
"loss": 0.4024,
"step": 25660
},
{
"epoch": 2.992653917910448,
"grad_norm": 0.40764856190725607,
"learning_rate": 2.2082614916689002e-05,
"loss": 0.3627,
"step": 25665
},
{
"epoch": 2.9932369402985075,
"grad_norm": 0.4239153676933328,
"learning_rate": 2.2074194467565514e-05,
"loss": 0.3837,
"step": 25670
},
{
"epoch": 2.993819962686567,
"grad_norm": 0.40367519234787475,
"learning_rate": 2.2065774825207304e-05,
"loss": 0.4064,
"step": 25675
},
{
"epoch": 2.9944029850746268,
"grad_norm": 0.41618454761265616,
"learning_rate": 2.2057355990866328e-05,
"loss": 0.3786,
"step": 25680
},
{
"epoch": 2.9949860074626864,
"grad_norm": 0.4684558839151656,
"learning_rate": 2.204893796579436e-05,
"loss": 0.4113,
"step": 25685
},
{
"epoch": 2.9955690298507465,
"grad_norm": 0.39257373267619683,
"learning_rate": 2.2040520751243094e-05,
"loss": 0.4094,
"step": 25690
},
{
"epoch": 2.996152052238806,
"grad_norm": 0.421877977446944,
"learning_rate": 2.2032104348464082e-05,
"loss": 0.4149,
"step": 25695
},
{
"epoch": 2.9967350746268657,
"grad_norm": 0.4045253191912498,
"learning_rate": 2.2023688758708767e-05,
"loss": 0.4285,
"step": 25700
},
{
"epoch": 2.9973180970149254,
"grad_norm": 0.4073067509537844,
"learning_rate": 2.201527398322846e-05,
"loss": 0.3997,
"step": 25705
},
{
"epoch": 2.997901119402985,
"grad_norm": 0.4153333223415684,
"learning_rate": 2.2006860023274363e-05,
"loss": 0.412,
"step": 25710
},
{
"epoch": 2.9984841417910446,
"grad_norm": 0.38451643205960345,
"learning_rate": 2.199844688009755e-05,
"loss": 0.4017,
"step": 25715
},
{
"epoch": 2.9990671641791042,
"grad_norm": 0.3894626368127005,
"learning_rate": 2.199003455494898e-05,
"loss": 0.431,
"step": 25720
},
{
"epoch": 2.9996501865671643,
"grad_norm": 0.413333139108324,
"learning_rate": 2.198162304907947e-05,
"loss": 0.3862,
"step": 25725
},
{
"epoch": 3.000233208955224,
"grad_norm": 0.3709667014045933,
"learning_rate": 2.1973212363739747e-05,
"loss": 0.3349,
"step": 25730
},
{
"epoch": 3.0008162313432836,
"grad_norm": 0.37324013937538697,
"learning_rate": 2.1964802500180388e-05,
"loss": 0.3258,
"step": 25735
},
{
"epoch": 3.001399253731343,
"grad_norm": 0.4321794169270504,
"learning_rate": 2.1956393459651864e-05,
"loss": 0.306,
"step": 25740
},
{
"epoch": 3.001982276119403,
"grad_norm": 0.43584264007268964,
"learning_rate": 2.1947985243404522e-05,
"loss": 0.3349,
"step": 25745
},
{
"epoch": 3.002565298507463,
"grad_norm": 0.4946303778086085,
"learning_rate": 2.1939577852688576e-05,
"loss": 0.3399,
"step": 25750
},
{
"epoch": 3.0031483208955225,
"grad_norm": 0.478556817161445,
"learning_rate": 2.1931171288754133e-05,
"loss": 0.3145,
"step": 25755
},
{
"epoch": 3.003731343283582,
"grad_norm": 0.43115696590550234,
"learning_rate": 2.1922765552851155e-05,
"loss": 0.324,
"step": 25760
},
{
"epoch": 3.004314365671642,
"grad_norm": 0.41381544876817744,
"learning_rate": 2.1914360646229508e-05,
"loss": 0.3308,
"step": 25765
},
{
"epoch": 3.0048973880597014,
"grad_norm": 0.45448327625698814,
"learning_rate": 2.190595657013892e-05,
"loss": 0.3501,
"step": 25770
},
{
"epoch": 3.005480410447761,
"grad_norm": 0.44213183445629095,
"learning_rate": 2.1897553325828984e-05,
"loss": 0.3219,
"step": 25775
},
{
"epoch": 3.006063432835821,
"grad_norm": 0.4284389496074295,
"learning_rate": 2.1889150914549195e-05,
"loss": 0.3102,
"step": 25780
},
{
"epoch": 3.0066464552238807,
"grad_norm": 0.4281305962350065,
"learning_rate": 2.18807493375489e-05,
"loss": 0.3164,
"step": 25785
},
{
"epoch": 3.0072294776119404,
"grad_norm": 0.46760793885757,
"learning_rate": 2.1872348596077348e-05,
"loss": 0.3303,
"step": 25790
},
{
"epoch": 3.0078125,
"grad_norm": 0.48810245439796657,
"learning_rate": 2.186394869138364e-05,
"loss": 0.3626,
"step": 25795
},
{
"epoch": 3.0083955223880596,
"grad_norm": 0.4188558881957621,
"learning_rate": 2.1855549624716755e-05,
"loss": 0.3079,
"step": 25800
},
{
"epoch": 3.0089785447761193,
"grad_norm": 0.6868156571860319,
"learning_rate": 2.1847151397325567e-05,
"loss": 0.3506,
"step": 25805
},
{
"epoch": 3.009561567164179,
"grad_norm": 0.4166324683203662,
"learning_rate": 2.1838754010458796e-05,
"loss": 0.3395,
"step": 25810
},
{
"epoch": 3.010144589552239,
"grad_norm": 0.439798060859837,
"learning_rate": 2.183035746536507e-05,
"loss": 0.3185,
"step": 25815
},
{
"epoch": 3.0107276119402986,
"grad_norm": 0.4368121556123018,
"learning_rate": 2.182196176329287e-05,
"loss": 0.3215,
"step": 25820
},
{
"epoch": 3.011310634328358,
"grad_norm": 0.5375571776187688,
"learning_rate": 2.181356690549054e-05,
"loss": 0.3328,
"step": 25825
},
{
"epoch": 3.011893656716418,
"grad_norm": 0.5128792384264712,
"learning_rate": 2.1805172893206342e-05,
"loss": 0.3191,
"step": 25830
},
{
"epoch": 3.0124766791044775,
"grad_norm": 0.5022495614254291,
"learning_rate": 2.179677972768836e-05,
"loss": 0.3202,
"step": 25835
},
{
"epoch": 3.013059701492537,
"grad_norm": 0.44919366412330874,
"learning_rate": 2.1788387410184603e-05,
"loss": 0.3083,
"step": 25840
},
{
"epoch": 3.013642723880597,
"grad_norm": 0.44961762537675826,
"learning_rate": 2.177999594194291e-05,
"loss": 0.3243,
"step": 25845
},
{
"epoch": 3.014225746268657,
"grad_norm": 0.4497450788881165,
"learning_rate": 2.177160532421101e-05,
"loss": 0.3275,
"step": 25850
},
{
"epoch": 3.0148087686567164,
"grad_norm": 0.5022477741310479,
"learning_rate": 2.1763215558236515e-05,
"loss": 0.3167,
"step": 25855
},
{
"epoch": 3.015391791044776,
"grad_norm": 0.4308664530537648,
"learning_rate": 2.1754826645266895e-05,
"loss": 0.328,
"step": 25860
},
{
"epoch": 3.0159748134328357,
"grad_norm": 0.5031641219848102,
"learning_rate": 2.1746438586549516e-05,
"loss": 0.3348,
"step": 25865
},
{
"epoch": 3.0165578358208953,
"grad_norm": 0.4583637418903681,
"learning_rate": 2.1738051383331598e-05,
"loss": 0.3192,
"step": 25870
},
{
"epoch": 3.0171408582089554,
"grad_norm": 0.47224843054081067,
"learning_rate": 2.1729665036860225e-05,
"loss": 0.3544,
"step": 25875
},
{
"epoch": 3.017723880597015,
"grad_norm": 0.4515220245909902,
"learning_rate": 2.172127954838238e-05,
"loss": 0.3094,
"step": 25880
},
{
"epoch": 3.0183069029850746,
"grad_norm": 0.45586070303977266,
"learning_rate": 2.1712894919144888e-05,
"loss": 0.3384,
"step": 25885
},
{
"epoch": 3.0188899253731343,
"grad_norm": 0.5169637575283627,
"learning_rate": 2.1704511150394486e-05,
"loss": 0.3156,
"step": 25890
},
{
"epoch": 3.019472947761194,
"grad_norm": 0.4461111433917812,
"learning_rate": 2.1696128243377743e-05,
"loss": 0.3282,
"step": 25895
},
{
"epoch": 3.0200559701492535,
"grad_norm": 0.437128585129177,
"learning_rate": 2.1687746199341118e-05,
"loss": 0.3028,
"step": 25900
},
{
"epoch": 3.0206389925373136,
"grad_norm": 0.4353962297447659,
"learning_rate": 2.1679365019530956e-05,
"loss": 0.3234,
"step": 25905
},
{
"epoch": 3.0212220149253732,
"grad_norm": 0.4144946526033545,
"learning_rate": 2.167098470519344e-05,
"loss": 0.3134,
"step": 25910
},
{
"epoch": 3.021805037313433,
"grad_norm": 0.45415991580330156,
"learning_rate": 2.1662605257574647e-05,
"loss": 0.3226,
"step": 25915
},
{
"epoch": 3.0223880597014925,
"grad_norm": 0.4494434819632714,
"learning_rate": 2.165422667792053e-05,
"loss": 0.3232,
"step": 25920
},
{
"epoch": 3.022971082089552,
"grad_norm": 0.4633580086275821,
"learning_rate": 2.1645848967476895e-05,
"loss": 0.3484,
"step": 25925
},
{
"epoch": 3.0235541044776117,
"grad_norm": 0.41772851509616826,
"learning_rate": 2.1637472127489427e-05,
"loss": 0.3049,
"step": 25930
},
{
"epoch": 3.024137126865672,
"grad_norm": 0.5298250438349478,
"learning_rate": 2.1629096159203683e-05,
"loss": 0.3369,
"step": 25935
},
{
"epoch": 3.0247201492537314,
"grad_norm": 0.47392793455278825,
"learning_rate": 2.162072106386509e-05,
"loss": 0.3282,
"step": 25940
},
{
"epoch": 3.025303171641791,
"grad_norm": 0.4667852627077425,
"learning_rate": 2.161234684271895e-05,
"loss": 0.3259,
"step": 25945
},
{
"epoch": 3.0258861940298507,
"grad_norm": 0.46375690535376113,
"learning_rate": 2.1603973497010417e-05,
"loss": 0.3221,
"step": 25950
},
{
"epoch": 3.0264692164179103,
"grad_norm": 0.4814207884770054,
"learning_rate": 2.1595601027984535e-05,
"loss": 0.3336,
"step": 25955
},
{
"epoch": 3.02705223880597,
"grad_norm": 0.4285397278536263,
"learning_rate": 2.158722943688621e-05,
"loss": 0.325,
"step": 25960
},
{
"epoch": 3.02763526119403,
"grad_norm": 0.5222797070512876,
"learning_rate": 2.1578858724960215e-05,
"loss": 0.3369,
"step": 25965
},
{
"epoch": 3.0282182835820897,
"grad_norm": 0.46558662752257896,
"learning_rate": 2.1570488893451203e-05,
"loss": 0.3177,
"step": 25970
},
{
"epoch": 3.0288013059701493,
"grad_norm": 0.4740422033882584,
"learning_rate": 2.1562119943603672e-05,
"loss": 0.3298,
"step": 25975
},
{
"epoch": 3.029384328358209,
"grad_norm": 0.48152240232416327,
"learning_rate": 2.1553751876662014e-05,
"loss": 0.3435,
"step": 25980
},
{
"epoch": 3.0299673507462686,
"grad_norm": 0.49326475205634374,
"learning_rate": 2.154538469387048e-05,
"loss": 0.3295,
"step": 25985
},
{
"epoch": 3.030550373134328,
"grad_norm": 0.4587260614381078,
"learning_rate": 2.1537018396473195e-05,
"loss": 0.3103,
"step": 25990
},
{
"epoch": 3.0311333955223883,
"grad_norm": 0.4509155814863767,
"learning_rate": 2.1528652985714143e-05,
"loss": 0.3269,
"step": 25995
},
{
"epoch": 3.031716417910448,
"grad_norm": 0.4828891286812978,
"learning_rate": 2.1520288462837175e-05,
"loss": 0.3316,
"step": 26000
},
{
"epoch": 3.0322994402985075,
"grad_norm": 0.48231842060370533,
"learning_rate": 2.1511924829086015e-05,
"loss": 0.3108,
"step": 26005
},
{
"epoch": 3.032882462686567,
"grad_norm": 0.5775698144252633,
"learning_rate": 2.1503562085704265e-05,
"loss": 0.3236,
"step": 26010
},
{
"epoch": 3.0334654850746268,
"grad_norm": 0.46151414345237757,
"learning_rate": 2.149520023393538e-05,
"loss": 0.3317,
"step": 26015
},
{
"epoch": 3.0340485074626864,
"grad_norm": 0.46242092784899796,
"learning_rate": 2.148683927502269e-05,
"loss": 0.3306,
"step": 26020
},
{
"epoch": 3.0346315298507465,
"grad_norm": 0.4189566355856064,
"learning_rate": 2.1478479210209383e-05,
"loss": 0.3335,
"step": 26025
},
{
"epoch": 3.035214552238806,
"grad_norm": 0.442510967396775,
"learning_rate": 2.147012004073853e-05,
"loss": 0.3416,
"step": 26030
},
{
"epoch": 3.0357975746268657,
"grad_norm": 0.48666760042409635,
"learning_rate": 2.1461761767853038e-05,
"loss": 0.3099,
"step": 26035
},
{
"epoch": 3.0363805970149254,
"grad_norm": 0.4111072274366574,
"learning_rate": 2.1453404392795735e-05,
"loss": 0.3184,
"step": 26040
},
{
"epoch": 3.036963619402985,
"grad_norm": 0.4640345669066334,
"learning_rate": 2.1445047916809262e-05,
"loss": 0.3181,
"step": 26045
},
{
"epoch": 3.0375466417910446,
"grad_norm": 0.46677437904450414,
"learning_rate": 2.143669234113614e-05,
"loss": 0.3174,
"step": 26050
},
{
"epoch": 3.0381296641791047,
"grad_norm": 0.4626266546083552,
"learning_rate": 2.1428337667018782e-05,
"loss": 0.3388,
"step": 26055
},
{
"epoch": 3.0387126865671643,
"grad_norm": 0.4956156984419573,
"learning_rate": 2.1419983895699437e-05,
"loss": 0.3387,
"step": 26060
},
{
"epoch": 3.039295708955224,
"grad_norm": 0.441904913513983,
"learning_rate": 2.141163102842023e-05,
"loss": 0.3292,
"step": 26065
},
{
"epoch": 3.0398787313432836,
"grad_norm": 0.4340883696843336,
"learning_rate": 2.1403279066423166e-05,
"loss": 0.3259,
"step": 26070
},
{
"epoch": 3.040461753731343,
"grad_norm": 0.44951348637735855,
"learning_rate": 2.1394928010950077e-05,
"loss": 0.3407,
"step": 26075
},
{
"epoch": 3.041044776119403,
"grad_norm": 0.48862955978877737,
"learning_rate": 2.1386577863242708e-05,
"loss": 0.3338,
"step": 26080
},
{
"epoch": 3.041627798507463,
"grad_norm": 0.44175615140140967,
"learning_rate": 2.1378228624542628e-05,
"loss": 0.3361,
"step": 26085
},
{
"epoch": 3.0422108208955225,
"grad_norm": 0.4990818637282398,
"learning_rate": 2.136988029609131e-05,
"loss": 0.3255,
"step": 26090
},
{
"epoch": 3.042793843283582,
"grad_norm": 0.4916890579858683,
"learning_rate": 2.1361532879130058e-05,
"loss": 0.3269,
"step": 26095
},
{
"epoch": 3.043376865671642,
"grad_norm": 0.44104377142597967,
"learning_rate": 2.135318637490004e-05,
"loss": 0.3116,
"step": 26100
},
{
"epoch": 3.0439598880597014,
"grad_norm": 0.46567665822910337,
"learning_rate": 2.1344840784642322e-05,
"loss": 0.3194,
"step": 26105
},
{
"epoch": 3.044542910447761,
"grad_norm": 0.46562424960840615,
"learning_rate": 2.1336496109597804e-05,
"loss": 0.3118,
"step": 26110
},
{
"epoch": 3.0451259328358207,
"grad_norm": 0.4892840179391156,
"learning_rate": 2.1328152351007264e-05,
"loss": 0.3103,
"step": 26115
},
{
"epoch": 3.0457089552238807,
"grad_norm": 0.43899662845298115,
"learning_rate": 2.131980951011134e-05,
"loss": 0.3369,
"step": 26120
},
{
"epoch": 3.0462919776119404,
"grad_norm": 0.513675059625149,
"learning_rate": 2.1311467588150518e-05,
"loss": 0.3341,
"step": 26125
},
{
"epoch": 3.046875,
"grad_norm": 0.47734407906100274,
"learning_rate": 2.1303126586365175e-05,
"loss": 0.3417,
"step": 26130
},
{
"epoch": 3.0474580223880596,
"grad_norm": 0.43312626180627223,
"learning_rate": 2.1294786505995534e-05,
"loss": 0.3144,
"step": 26135
},
{
"epoch": 3.0480410447761193,
"grad_norm": 0.45351095507528777,
"learning_rate": 2.1286447348281695e-05,
"loss": 0.3246,
"step": 26140
},
{
"epoch": 3.0486240671641793,
"grad_norm": 0.550977527627621,
"learning_rate": 2.1278109114463594e-05,
"loss": 0.343,
"step": 26145
},
{
"epoch": 3.049207089552239,
"grad_norm": 0.4570110420316436,
"learning_rate": 2.126977180578106e-05,
"loss": 0.3456,
"step": 26150
},
{
"epoch": 3.0497901119402986,
"grad_norm": 0.4726144359521414,
"learning_rate": 2.1261435423473765e-05,
"loss": 0.3282,
"step": 26155
},
{
"epoch": 3.050373134328358,
"grad_norm": 0.4383123748438573,
"learning_rate": 2.1253099968781237e-05,
"loss": 0.3162,
"step": 26160
},
{
"epoch": 3.050956156716418,
"grad_norm": 0.4733717130768152,
"learning_rate": 2.1244765442942904e-05,
"loss": 0.3178,
"step": 26165
},
{
"epoch": 3.0515391791044775,
"grad_norm": 0.46542673525491274,
"learning_rate": 2.1236431847198017e-05,
"loss": 0.3281,
"step": 26170
},
{
"epoch": 3.052122201492537,
"grad_norm": 0.4735626309677211,
"learning_rate": 2.1228099182785693e-05,
"loss": 0.3389,
"step": 26175
},
{
"epoch": 3.052705223880597,
"grad_norm": 0.4818373430655894,
"learning_rate": 2.1219767450944938e-05,
"loss": 0.3179,
"step": 26180
},
{
"epoch": 3.053288246268657,
"grad_norm": 0.4681946119389254,
"learning_rate": 2.1211436652914585e-05,
"loss": 0.3105,
"step": 26185
},
{
"epoch": 3.0538712686567164,
"grad_norm": 0.49628334189273177,
"learning_rate": 2.1203106789933352e-05,
"loss": 0.3468,
"step": 26190
},
{
"epoch": 3.054454291044776,
"grad_norm": 0.4760500825592149,
"learning_rate": 2.119477786323981e-05,
"loss": 0.3201,
"step": 26195
},
{
"epoch": 3.0550373134328357,
"grad_norm": 0.45477546124887575,
"learning_rate": 2.1186449874072385e-05,
"loss": 0.341,
"step": 26200
},
{
"epoch": 3.0556203358208953,
"grad_norm": 0.5026800137037384,
"learning_rate": 2.1178122823669373e-05,
"loss": 0.3393,
"step": 26205
},
{
"epoch": 3.0562033582089554,
"grad_norm": 0.5065218787007753,
"learning_rate": 2.116979671326892e-05,
"loss": 0.3267,
"step": 26210
},
{
"epoch": 3.056786380597015,
"grad_norm": 0.41895410179357156,
"learning_rate": 2.1161471544109057e-05,
"loss": 0.3151,
"step": 26215
},
{
"epoch": 3.0573694029850746,
"grad_norm": 0.40538522270532434,
"learning_rate": 2.115314731742764e-05,
"loss": 0.3071,
"step": 26220
},
{
"epoch": 3.0579524253731343,
"grad_norm": 0.48454399106534213,
"learning_rate": 2.1144824034462403e-05,
"loss": 0.337,
"step": 26225
},
{
"epoch": 3.058535447761194,
"grad_norm": 0.4370705887097963,
"learning_rate": 2.1136501696450943e-05,
"loss": 0.3302,
"step": 26230
},
{
"epoch": 3.0591184701492535,
"grad_norm": 0.4445676743234194,
"learning_rate": 2.112818030463071e-05,
"loss": 0.3185,
"step": 26235
},
{
"epoch": 3.0597014925373136,
"grad_norm": 0.4991765898821802,
"learning_rate": 2.1119859860239023e-05,
"loss": 0.3275,
"step": 26240
},
{
"epoch": 3.0602845149253732,
"grad_norm": 0.46111349895725334,
"learning_rate": 2.1111540364513045e-05,
"loss": 0.3186,
"step": 26245
},
{
"epoch": 3.060867537313433,
"grad_norm": 0.4484815999642295,
"learning_rate": 2.1103221818689794e-05,
"loss": 0.3072,
"step": 26250
},
{
"epoch": 3.0614505597014925,
"grad_norm": 0.43613201402576735,
"learning_rate": 2.1094904224006185e-05,
"loss": 0.3189,
"step": 26255
},
{
"epoch": 3.062033582089552,
"grad_norm": 0.4368901513564297,
"learning_rate": 2.108658758169893e-05,
"loss": 0.3168,
"step": 26260
},
{
"epoch": 3.0626166044776117,
"grad_norm": 0.46999070952488564,
"learning_rate": 2.107827189300467e-05,
"loss": 0.3177,
"step": 26265
},
{
"epoch": 3.063199626865672,
"grad_norm": 0.47318691393207823,
"learning_rate": 2.1069957159159848e-05,
"loss": 0.3332,
"step": 26270
},
{
"epoch": 3.0637826492537314,
"grad_norm": 0.46268788338947603,
"learning_rate": 2.1061643381400785e-05,
"loss": 0.3084,
"step": 26275
},
{
"epoch": 3.064365671641791,
"grad_norm": 0.44748836691943034,
"learning_rate": 2.105333056096367e-05,
"loss": 0.3388,
"step": 26280
},
{
"epoch": 3.0649486940298507,
"grad_norm": 0.41553750338952344,
"learning_rate": 2.104501869908453e-05,
"loss": 0.312,
"step": 26285
},
{
"epoch": 3.0655317164179103,
"grad_norm": 0.48119437755123573,
"learning_rate": 2.1036707796999267e-05,
"loss": 0.315,
"step": 26290
},
{
"epoch": 3.06611473880597,
"grad_norm": 0.5485148429526312,
"learning_rate": 2.102839785594362e-05,
"loss": 0.3217,
"step": 26295
},
{
"epoch": 3.06669776119403,
"grad_norm": 0.4813421692880979,
"learning_rate": 2.1020088877153215e-05,
"loss": 0.3154,
"step": 26300
},
{
"epoch": 3.0672807835820897,
"grad_norm": 0.45170470024821985,
"learning_rate": 2.1011780861863504e-05,
"loss": 0.3131,
"step": 26305
},
{
"epoch": 3.0678638059701493,
"grad_norm": 0.45835875961148714,
"learning_rate": 2.100347381130982e-05,
"loss": 0.3368,
"step": 26310
},
{
"epoch": 3.068446828358209,
"grad_norm": 0.510026992930468,
"learning_rate": 2.099516772672733e-05,
"loss": 0.3332,
"step": 26315
},
{
"epoch": 3.0690298507462686,
"grad_norm": 0.5120793089999462,
"learning_rate": 2.0986862609351077e-05,
"loss": 0.3464,
"step": 26320
},
{
"epoch": 3.069612873134328,
"grad_norm": 0.46710572074543905,
"learning_rate": 2.0978558460415954e-05,
"loss": 0.3372,
"step": 26325
},
{
"epoch": 3.0701958955223883,
"grad_norm": 0.6917733491212216,
"learning_rate": 2.09702552811567e-05,
"loss": 0.3305,
"step": 26330
},
{
"epoch": 3.070778917910448,
"grad_norm": 0.44714880732994,
"learning_rate": 2.096195307280792e-05,
"loss": 0.3166,
"step": 26335
},
{
"epoch": 3.0713619402985075,
"grad_norm": 0.4460138250360666,
"learning_rate": 2.0953651836604083e-05,
"loss": 0.3315,
"step": 26340
},
{
"epoch": 3.071944962686567,
"grad_norm": 0.43659882259326416,
"learning_rate": 2.094535157377949e-05,
"loss": 0.3325,
"step": 26345
},
{
"epoch": 3.0725279850746268,
"grad_norm": 0.4523932005657126,
"learning_rate": 2.093705228556832e-05,
"loss": 0.325,
"step": 26350
},
{
"epoch": 3.0731110074626864,
"grad_norm": 0.4360071900869937,
"learning_rate": 2.0928753973204597e-05,
"loss": 0.3187,
"step": 26355
},
{
"epoch": 3.0736940298507465,
"grad_norm": 0.515786247549207,
"learning_rate": 2.0920456637922194e-05,
"loss": 0.3373,
"step": 26360
},
{
"epoch": 3.074277052238806,
"grad_norm": 0.4664235027987304,
"learning_rate": 2.0912160280954852e-05,
"loss": 0.3293,
"step": 26365
},
{
"epoch": 3.0748600746268657,
"grad_norm": 0.5105316255986599,
"learning_rate": 2.0903864903536147e-05,
"loss": 0.332,
"step": 26370
},
{
"epoch": 3.0754430970149254,
"grad_norm": 0.465501450938767,
"learning_rate": 2.0895570506899544e-05,
"loss": 0.3122,
"step": 26375
},
{
"epoch": 3.076026119402985,
"grad_norm": 0.4784328153797881,
"learning_rate": 2.088727709227833e-05,
"loss": 0.3366,
"step": 26380
},
{
"epoch": 3.0766091417910446,
"grad_norm": 0.4562543660615144,
"learning_rate": 2.0878984660905644e-05,
"loss": 0.3149,
"step": 26385
},
{
"epoch": 3.0771921641791047,
"grad_norm": 0.4496520926125867,
"learning_rate": 2.087069321401451e-05,
"loss": 0.3207,
"step": 26390
},
{
"epoch": 3.0777751865671643,
"grad_norm": 0.4712252095464664,
"learning_rate": 2.0862402752837768e-05,
"loss": 0.3332,
"step": 26395
},
{
"epoch": 3.078358208955224,
"grad_norm": 0.4569205082117521,
"learning_rate": 2.085411327860815e-05,
"loss": 0.3405,
"step": 26400
},
{
"epoch": 3.0789412313432836,
"grad_norm": 0.4602314078830094,
"learning_rate": 2.084582479255821e-05,
"loss": 0.3318,
"step": 26405
},
{
"epoch": 3.079524253731343,
"grad_norm": 0.4696749255322814,
"learning_rate": 2.083753729592037e-05,
"loss": 0.321,
"step": 26410
},
{
"epoch": 3.080107276119403,
"grad_norm": 0.4352764869390288,
"learning_rate": 2.0829250789926898e-05,
"loss": 0.3098,
"step": 26415
},
{
"epoch": 3.080690298507463,
"grad_norm": 0.48320746852302104,
"learning_rate": 2.0820965275809913e-05,
"loss": 0.3328,
"step": 26420
},
{
"epoch": 3.0812733208955225,
"grad_norm": 0.46097197214179425,
"learning_rate": 2.081268075480141e-05,
"loss": 0.3121,
"step": 26425
},
{
"epoch": 3.081856343283582,
"grad_norm": 0.4922339007608091,
"learning_rate": 2.0804397228133205e-05,
"loss": 0.3418,
"step": 26430
},
{
"epoch": 3.082439365671642,
"grad_norm": 0.46339810671505294,
"learning_rate": 2.0796114697036972e-05,
"loss": 0.3308,
"step": 26435
},
{
"epoch": 3.0830223880597014,
"grad_norm": 0.574397678354726,
"learning_rate": 2.0787833162744257e-05,
"loss": 0.3119,
"step": 26440
},
{
"epoch": 3.083605410447761,
"grad_norm": 0.4282297159701541,
"learning_rate": 2.0779552626486444e-05,
"loss": 0.347,
"step": 26445
},
{
"epoch": 3.0841884328358207,
"grad_norm": 0.5042237481556545,
"learning_rate": 2.077127308949476e-05,
"loss": 0.3299,
"step": 26450
},
{
"epoch": 3.0847714552238807,
"grad_norm": 0.4351639604910172,
"learning_rate": 2.0762994553000304e-05,
"loss": 0.3214,
"step": 26455
},
{
"epoch": 3.0853544776119404,
"grad_norm": 0.4558374981572576,
"learning_rate": 2.0754717018234003e-05,
"loss": 0.3151,
"step": 26460
},
{
"epoch": 3.0859375,
"grad_norm": 0.44193020864322696,
"learning_rate": 2.074644048642666e-05,
"loss": 0.3301,
"step": 26465
},
{
"epoch": 3.0865205223880596,
"grad_norm": 0.4409441882722866,
"learning_rate": 2.0738164958808905e-05,
"loss": 0.3045,
"step": 26470
},
{
"epoch": 3.0871035447761193,
"grad_norm": 0.46245906743626414,
"learning_rate": 2.072989043661124e-05,
"loss": 0.3305,
"step": 26475
},
{
"epoch": 3.0876865671641793,
"grad_norm": 0.5313669338688435,
"learning_rate": 2.072161692106399e-05,
"loss": 0.3753,
"step": 26480
},
{
"epoch": 3.088269589552239,
"grad_norm": 0.43890620433513766,
"learning_rate": 2.0713344413397368e-05,
"loss": 0.3048,
"step": 26485
},
{
"epoch": 3.0888526119402986,
"grad_norm": 0.45550645734538264,
"learning_rate": 2.0705072914841407e-05,
"loss": 0.3254,
"step": 26490
},
{
"epoch": 3.089435634328358,
"grad_norm": 0.45127483412657704,
"learning_rate": 2.0696802426625993e-05,
"loss": 0.314,
"step": 26495
},
{
"epoch": 3.090018656716418,
"grad_norm": 0.47559466833847036,
"learning_rate": 2.0688532949980882e-05,
"loss": 0.3491,
"step": 26500
},
{
"epoch": 3.0906016791044775,
"grad_norm": 0.4560452850152337,
"learning_rate": 2.0680264486135665e-05,
"loss": 0.3025,
"step": 26505
},
{
"epoch": 3.091184701492537,
"grad_norm": 0.44289312672337844,
"learning_rate": 2.0671997036319763e-05,
"loss": 0.3096,
"step": 26510
},
{
"epoch": 3.091767723880597,
"grad_norm": 0.44977244207477446,
"learning_rate": 2.0663730601762494e-05,
"loss": 0.3249,
"step": 26515
},
{
"epoch": 3.092350746268657,
"grad_norm": 0.4944824796926875,
"learning_rate": 2.0655465183692972e-05,
"loss": 0.3373,
"step": 26520
},
{
"epoch": 3.0929337686567164,
"grad_norm": 0.4602852658140172,
"learning_rate": 2.0647200783340214e-05,
"loss": 0.3265,
"step": 26525
},
{
"epoch": 3.093516791044776,
"grad_norm": 0.43619491761256923,
"learning_rate": 2.063893740193304e-05,
"loss": 0.3409,
"step": 26530
},
{
"epoch": 3.0940998134328357,
"grad_norm": 0.49781783714355804,
"learning_rate": 2.063067504070012e-05,
"loss": 0.35,
"step": 26535
},
{
"epoch": 3.0946828358208953,
"grad_norm": 0.49332653496439094,
"learning_rate": 2.0622413700870026e-05,
"loss": 0.3266,
"step": 26540
},
{
"epoch": 3.0952658582089554,
"grad_norm": 0.45841570511793567,
"learning_rate": 2.0614153383671103e-05,
"loss": 0.3234,
"step": 26545
},
{
"epoch": 3.095848880597015,
"grad_norm": 0.4661202506373183,
"learning_rate": 2.0605894090331607e-05,
"loss": 0.3386,
"step": 26550
},
{
"epoch": 3.0964319029850746,
"grad_norm": 0.43819664647364176,
"learning_rate": 2.0597635822079607e-05,
"loss": 0.3205,
"step": 26555
},
{
"epoch": 3.0970149253731343,
"grad_norm": 0.4437153488494203,
"learning_rate": 2.0589378580143016e-05,
"loss": 0.3369,
"step": 26560
},
{
"epoch": 3.097597947761194,
"grad_norm": 0.48069583190874654,
"learning_rate": 2.058112236574963e-05,
"loss": 0.3218,
"step": 26565
},
{
"epoch": 3.0981809701492535,
"grad_norm": 0.47697809357117477,
"learning_rate": 2.057286718012705e-05,
"loss": 0.3231,
"step": 26570
},
{
"epoch": 3.0987639925373136,
"grad_norm": 0.4840688119797522,
"learning_rate": 2.0564613024502754e-05,
"loss": 0.3395,
"step": 26575
},
{
"epoch": 3.0993470149253732,
"grad_norm": 0.603029117989374,
"learning_rate": 2.0556359900104054e-05,
"loss": 0.3311,
"step": 26580
},
{
"epoch": 3.099930037313433,
"grad_norm": 0.47697535251778106,
"learning_rate": 2.0548107808158102e-05,
"loss": 0.3093,
"step": 26585
},
{
"epoch": 3.1005130597014925,
"grad_norm": 0.4882263624088247,
"learning_rate": 2.0539856749891918e-05,
"loss": 0.3437,
"step": 26590
},
{
"epoch": 3.101096082089552,
"grad_norm": 0.451119064870134,
"learning_rate": 2.0531606726532344e-05,
"loss": 0.3493,
"step": 26595
},
{
"epoch": 3.1016791044776117,
"grad_norm": 0.4338425342728425,
"learning_rate": 2.0523357739306087e-05,
"loss": 0.3246,
"step": 26600
},
{
"epoch": 3.102262126865672,
"grad_norm": 0.4448999392854656,
"learning_rate": 2.0515109789439695e-05,
"loss": 0.3395,
"step": 26605
},
{
"epoch": 3.1028451492537314,
"grad_norm": 0.4976464626815555,
"learning_rate": 2.050686287815954e-05,
"loss": 0.3382,
"step": 26610
},
{
"epoch": 3.103428171641791,
"grad_norm": 0.4663663206133034,
"learning_rate": 2.049861700669189e-05,
"loss": 0.3335,
"step": 26615
},
{
"epoch": 3.1040111940298507,
"grad_norm": 0.4542148386148399,
"learning_rate": 2.049037217626279e-05,
"loss": 0.3412,
"step": 26620
},
{
"epoch": 3.1045942164179103,
"grad_norm": 0.46942413203721617,
"learning_rate": 2.0482128388098202e-05,
"loss": 0.3191,
"step": 26625
},
{
"epoch": 3.10517723880597,
"grad_norm": 0.4688295158411415,
"learning_rate": 2.0473885643423885e-05,
"loss": 0.3352,
"step": 26630
},
{
"epoch": 3.10576026119403,
"grad_norm": 0.47340585860574863,
"learning_rate": 2.046564394346544e-05,
"loss": 0.3296,
"step": 26635
},
{
"epoch": 3.1063432835820897,
"grad_norm": 0.43339045008618204,
"learning_rate": 2.0457403289448353e-05,
"loss": 0.3305,
"step": 26640
},
{
"epoch": 3.1069263059701493,
"grad_norm": 0.4527274948867366,
"learning_rate": 2.0449163682597915e-05,
"loss": 0.3239,
"step": 26645
},
{
"epoch": 3.107509328358209,
"grad_norm": 0.46326112077044784,
"learning_rate": 2.0440925124139286e-05,
"loss": 0.3337,
"step": 26650
},
{
"epoch": 3.1080923507462686,
"grad_norm": 0.4754515700991158,
"learning_rate": 2.0432687615297458e-05,
"loss": 0.3164,
"step": 26655
},
{
"epoch": 3.108675373134328,
"grad_norm": 0.45125925500992764,
"learning_rate": 2.0424451157297264e-05,
"loss": 0.3166,
"step": 26660
},
{
"epoch": 3.1092583955223883,
"grad_norm": 0.4711447156286339,
"learning_rate": 2.0416215751363392e-05,
"loss": 0.3157,
"step": 26665
},
{
"epoch": 3.109841417910448,
"grad_norm": 0.4641385040928347,
"learning_rate": 2.040798139872037e-05,
"loss": 0.3308,
"step": 26670
},
{
"epoch": 3.1104244402985075,
"grad_norm": 0.5248646209056408,
"learning_rate": 2.0399748100592564e-05,
"loss": 0.3325,
"step": 26675
},
{
"epoch": 3.111007462686567,
"grad_norm": 0.4591315961710023,
"learning_rate": 2.0391515858204184e-05,
"loss": 0.3104,
"step": 26680
},
{
"epoch": 3.1115904850746268,
"grad_norm": 0.4807515110675291,
"learning_rate": 2.038328467277929e-05,
"loss": 0.3196,
"step": 26685
},
{
"epoch": 3.1121735074626864,
"grad_norm": 0.457473043091165,
"learning_rate": 2.0375054545541776e-05,
"loss": 0.3348,
"step": 26690
},
{
"epoch": 3.1127565298507465,
"grad_norm": 0.488233815914533,
"learning_rate": 2.0366825477715386e-05,
"loss": 0.338,
"step": 26695
},
{
"epoch": 3.113339552238806,
"grad_norm": 1.2441433195164318,
"learning_rate": 2.0358597470523706e-05,
"loss": 0.3272,
"step": 26700
},
{
"epoch": 3.1139225746268657,
"grad_norm": 0.5547517656519116,
"learning_rate": 2.035037052519016e-05,
"loss": 0.3199,
"step": 26705
},
{
"epoch": 3.1145055970149254,
"grad_norm": 0.5006793922568553,
"learning_rate": 2.034214464293801e-05,
"loss": 0.326,
"step": 26710
},
{
"epoch": 3.115088619402985,
"grad_norm": 0.46256078339483764,
"learning_rate": 2.0333919824990372e-05,
"loss": 0.3262,
"step": 26715
},
{
"epoch": 3.1156716417910446,
"grad_norm": 0.4248843856616436,
"learning_rate": 2.0325696072570195e-05,
"loss": 0.3116,
"step": 26720
},
{
"epoch": 3.1162546641791047,
"grad_norm": 0.46781738586616756,
"learning_rate": 2.0317473386900275e-05,
"loss": 0.321,
"step": 26725
},
{
"epoch": 3.1168376865671643,
"grad_norm": 0.48694496352209315,
"learning_rate": 2.0309251769203252e-05,
"loss": 0.3293,
"step": 26730
},
{
"epoch": 3.117420708955224,
"grad_norm": 0.4822676988381848,
"learning_rate": 2.0301031220701582e-05,
"loss": 0.3224,
"step": 26735
},
{
"epoch": 3.1180037313432836,
"grad_norm": 0.4377358685137589,
"learning_rate": 2.0292811742617607e-05,
"loss": 0.3327,
"step": 26740
},
{
"epoch": 3.118586753731343,
"grad_norm": 0.4404678461258958,
"learning_rate": 2.028459333617346e-05,
"loss": 0.3119,
"step": 26745
},
{
"epoch": 3.119169776119403,
"grad_norm": 0.494397776515022,
"learning_rate": 2.0276376002591164e-05,
"loss": 0.351,
"step": 26750
},
{
"epoch": 3.119752798507463,
"grad_norm": 0.44359941570618844,
"learning_rate": 2.0268159743092546e-05,
"loss": 0.3227,
"step": 26755
},
{
"epoch": 3.1203358208955225,
"grad_norm": 0.4705878230779239,
"learning_rate": 2.0259944558899274e-05,
"loss": 0.3375,
"step": 26760
},
{
"epoch": 3.120918843283582,
"grad_norm": 0.4397419378532711,
"learning_rate": 2.0251730451232886e-05,
"loss": 0.3172,
"step": 26765
},
{
"epoch": 3.121501865671642,
"grad_norm": 0.4672998399885591,
"learning_rate": 2.0243517421314727e-05,
"loss": 0.3377,
"step": 26770
},
{
"epoch": 3.1220848880597014,
"grad_norm": 0.4488275991075581,
"learning_rate": 2.0235305470366008e-05,
"loss": 0.3341,
"step": 26775
},
{
"epoch": 3.122667910447761,
"grad_norm": 0.45724136835508916,
"learning_rate": 2.022709459960776e-05,
"loss": 0.3263,
"step": 26780
},
{
"epoch": 3.1232509328358207,
"grad_norm": 0.4303589813302327,
"learning_rate": 2.021888481026086e-05,
"loss": 0.3258,
"step": 26785
},
{
"epoch": 3.1238339552238807,
"grad_norm": 0.44367388925675355,
"learning_rate": 2.0210676103546028e-05,
"loss": 0.3302,
"step": 26790
},
{
"epoch": 3.1244169776119404,
"grad_norm": 0.5671474336419254,
"learning_rate": 2.0202468480683812e-05,
"loss": 0.3226,
"step": 26795
},
{
"epoch": 3.125,
"grad_norm": 0.4693549663720372,
"learning_rate": 2.0194261942894628e-05,
"loss": 0.3249,
"step": 26800
},
{
"epoch": 3.1255830223880596,
"grad_norm": 0.47169687262536697,
"learning_rate": 2.0186056491398686e-05,
"loss": 0.3251,
"step": 26805
},
{
"epoch": 3.1261660447761193,
"grad_norm": 0.5245771146045501,
"learning_rate": 2.0177852127416063e-05,
"loss": 0.3275,
"step": 26810
},
{
"epoch": 3.1267490671641793,
"grad_norm": 0.44686541430709736,
"learning_rate": 2.0169648852166684e-05,
"loss": 0.3225,
"step": 26815
},
{
"epoch": 3.127332089552239,
"grad_norm": 0.4740393682290807,
"learning_rate": 2.016144666687029e-05,
"loss": 0.3379,
"step": 26820
},
{
"epoch": 3.1279151119402986,
"grad_norm": 0.41623195515592065,
"learning_rate": 2.015324557274645e-05,
"loss": 0.316,
"step": 26825
},
{
"epoch": 3.128498134328358,
"grad_norm": 0.4294606089025559,
"learning_rate": 2.0145045571014614e-05,
"loss": 0.2945,
"step": 26830
},
{
"epoch": 3.129081156716418,
"grad_norm": 0.454012613019494,
"learning_rate": 2.0136846662894028e-05,
"loss": 0.3257,
"step": 26835
},
{
"epoch": 3.1296641791044775,
"grad_norm": 0.5001273339562864,
"learning_rate": 2.0128648849603798e-05,
"loss": 0.3357,
"step": 26840
},
{
"epoch": 3.130247201492537,
"grad_norm": 0.4811349204320928,
"learning_rate": 2.0120452132362854e-05,
"loss": 0.3325,
"step": 26845
},
{
"epoch": 3.130830223880597,
"grad_norm": 0.4938092125575546,
"learning_rate": 2.0112256512389976e-05,
"loss": 0.3091,
"step": 26850
},
{
"epoch": 3.131413246268657,
"grad_norm": 0.42774595838386315,
"learning_rate": 2.0104061990903773e-05,
"loss": 0.3103,
"step": 26855
},
{
"epoch": 3.1319962686567164,
"grad_norm": 0.4660227477236933,
"learning_rate": 2.009586856912269e-05,
"loss": 0.3312,
"step": 26860
},
{
"epoch": 3.132579291044776,
"grad_norm": 0.4510791455861746,
"learning_rate": 2.008767624826501e-05,
"loss": 0.3213,
"step": 26865
},
{
"epoch": 3.1331623134328357,
"grad_norm": 0.48343884159203576,
"learning_rate": 2.0079485029548838e-05,
"loss": 0.3293,
"step": 26870
},
{
"epoch": 3.1337453358208953,
"grad_norm": 0.4567351985253224,
"learning_rate": 2.007129491419217e-05,
"loss": 0.3147,
"step": 26875
},
{
"epoch": 3.1343283582089554,
"grad_norm": 0.46898131402176935,
"learning_rate": 2.006310590341276e-05,
"loss": 0.3176,
"step": 26880
},
{
"epoch": 3.134911380597015,
"grad_norm": 0.4289934744637185,
"learning_rate": 2.005491799842824e-05,
"loss": 0.3313,
"step": 26885
},
{
"epoch": 3.1354944029850746,
"grad_norm": 0.6515331892054995,
"learning_rate": 2.0046731200456097e-05,
"loss": 0.3547,
"step": 26890
},
{
"epoch": 3.1360774253731343,
"grad_norm": 0.4230308599433454,
"learning_rate": 2.0038545510713597e-05,
"loss": 0.3303,
"step": 26895
},
{
"epoch": 3.136660447761194,
"grad_norm": 0.47907343313743017,
"learning_rate": 2.00303609304179e-05,
"loss": 0.3365,
"step": 26900
},
{
"epoch": 3.1372434701492535,
"grad_norm": 0.46736099186068,
"learning_rate": 2.0022177460785974e-05,
"loss": 0.3423,
"step": 26905
},
{
"epoch": 3.1378264925373136,
"grad_norm": 0.4512622531085032,
"learning_rate": 2.0013995103034594e-05,
"loss": 0.332,
"step": 26910
},
{
"epoch": 3.1384095149253732,
"grad_norm": 0.4825903985622654,
"learning_rate": 2.000581385838043e-05,
"loss": 0.3213,
"step": 26915
},
{
"epoch": 3.138992537313433,
"grad_norm": 0.5276528592938703,
"learning_rate": 1.9997633728039933e-05,
"loss": 0.319,
"step": 26920
},
{
"epoch": 3.1395755597014925,
"grad_norm": 0.4775936717912453,
"learning_rate": 1.9989454713229426e-05,
"loss": 0.3271,
"step": 26925
},
{
"epoch": 3.140158582089552,
"grad_norm": 0.4603167929873936,
"learning_rate": 1.9981276815165046e-05,
"loss": 0.3216,
"step": 26930
},
{
"epoch": 3.1407416044776117,
"grad_norm": 0.452610017241828,
"learning_rate": 1.9973100035062753e-05,
"loss": 0.3216,
"step": 26935
},
{
"epoch": 3.141324626865672,
"grad_norm": 0.43628172518109953,
"learning_rate": 1.996492437413838e-05,
"loss": 0.3111,
"step": 26940
},
{
"epoch": 3.1419076492537314,
"grad_norm": 0.44650003921095405,
"learning_rate": 1.9956749833607545e-05,
"loss": 0.3218,
"step": 26945
},
{
"epoch": 3.142490671641791,
"grad_norm": 0.44428101590584085,
"learning_rate": 1.994857641468575e-05,
"loss": 0.3357,
"step": 26950
},
{
"epoch": 3.1430736940298507,
"grad_norm": 0.4624194022412386,
"learning_rate": 1.994040411858829e-05,
"loss": 0.3158,
"step": 26955
},
{
"epoch": 3.1436567164179103,
"grad_norm": 0.44922257500149254,
"learning_rate": 1.99322329465303e-05,
"loss": 0.311,
"step": 26960
},
{
"epoch": 3.14423973880597,
"grad_norm": 0.4556748287330155,
"learning_rate": 1.992406289972677e-05,
"loss": 0.3252,
"step": 26965
},
{
"epoch": 3.14482276119403,
"grad_norm": 0.47534573579196415,
"learning_rate": 1.9915893979392492e-05,
"loss": 0.3524,
"step": 26970
},
{
"epoch": 3.1454057835820897,
"grad_norm": 0.4296917439311925,
"learning_rate": 1.9907726186742122e-05,
"loss": 0.327,
"step": 26975
},
{
"epoch": 3.1459888059701493,
"grad_norm": 0.4707200928426098,
"learning_rate": 1.989955952299012e-05,
"loss": 0.3179,
"step": 26980
},
{
"epoch": 3.146571828358209,
"grad_norm": 0.4610040585110528,
"learning_rate": 1.9891393989350794e-05,
"loss": 0.3198,
"step": 26985
},
{
"epoch": 3.1471548507462686,
"grad_norm": 0.5120188222955415,
"learning_rate": 1.9883229587038287e-05,
"loss": 0.3403,
"step": 26990
},
{
"epoch": 3.147737873134328,
"grad_norm": 0.44834359511844246,
"learning_rate": 1.987506631726656e-05,
"loss": 0.3194,
"step": 26995
},
{
"epoch": 3.1483208955223883,
"grad_norm": 0.4544558546793936,
"learning_rate": 1.986690418124942e-05,
"loss": 0.3417,
"step": 27000
},
{
"epoch": 3.148903917910448,
"grad_norm": 0.4207714545786385,
"learning_rate": 1.985874318020049e-05,
"loss": 0.3174,
"step": 27005
},
{
"epoch": 3.1494869402985075,
"grad_norm": 0.4823490868960121,
"learning_rate": 1.9850583315333242e-05,
"loss": 0.3287,
"step": 27010
},
{
"epoch": 3.150069962686567,
"grad_norm": 0.47785976388211493,
"learning_rate": 1.9842424587860958e-05,
"loss": 0.3246,
"step": 27015
},
{
"epoch": 3.1506529850746268,
"grad_norm": 0.4452797578776353,
"learning_rate": 1.983426699899677e-05,
"loss": 0.3156,
"step": 27020
},
{
"epoch": 3.1512360074626864,
"grad_norm": 0.457534383185208,
"learning_rate": 1.982611054995364e-05,
"loss": 0.3279,
"step": 27025
},
{
"epoch": 3.1518190298507465,
"grad_norm": 0.47946140418120975,
"learning_rate": 1.9817955241944335e-05,
"loss": 0.3366,
"step": 27030
},
{
"epoch": 3.152402052238806,
"grad_norm": 0.47569245373062846,
"learning_rate": 1.980980107618149e-05,
"loss": 0.328,
"step": 27035
},
{
"epoch": 3.1529850746268657,
"grad_norm": 0.44578237018932065,
"learning_rate": 1.9801648053877548e-05,
"loss": 0.3306,
"step": 27040
},
{
"epoch": 3.1535680970149254,
"grad_norm": 0.479070478783956,
"learning_rate": 1.979349617624477e-05,
"loss": 0.3062,
"step": 27045
},
{
"epoch": 3.154151119402985,
"grad_norm": 0.4470158079249052,
"learning_rate": 1.978534544449528e-05,
"loss": 0.306,
"step": 27050
},
{
"epoch": 3.1547341417910446,
"grad_norm": 0.44846548897816296,
"learning_rate": 1.9777195859840997e-05,
"loss": 0.3253,
"step": 27055
},
{
"epoch": 3.1553171641791047,
"grad_norm": 0.4385400951501217,
"learning_rate": 1.9769047423493707e-05,
"loss": 0.3361,
"step": 27060
},
{
"epoch": 3.1559001865671643,
"grad_norm": 0.46022618469429677,
"learning_rate": 1.9760900136664994e-05,
"loss": 0.3278,
"step": 27065
},
{
"epoch": 3.156483208955224,
"grad_norm": 0.5262601960800007,
"learning_rate": 1.975275400056627e-05,
"loss": 0.3476,
"step": 27070
},
{
"epoch": 3.1570662313432836,
"grad_norm": 0.4211487289158979,
"learning_rate": 1.974460901640881e-05,
"loss": 0.3067,
"step": 27075
},
{
"epoch": 3.157649253731343,
"grad_norm": 0.45755982586604127,
"learning_rate": 1.9736465185403675e-05,
"loss": 0.3272,
"step": 27080
},
{
"epoch": 3.158232276119403,
"grad_norm": 0.46597636071099746,
"learning_rate": 1.9728322508761794e-05,
"loss": 0.3336,
"step": 27085
},
{
"epoch": 3.158815298507463,
"grad_norm": 0.44247473549671384,
"learning_rate": 1.9720180987693888e-05,
"loss": 0.3336,
"step": 27090
},
{
"epoch": 3.1593983208955225,
"grad_norm": 0.47766538399245295,
"learning_rate": 1.9712040623410523e-05,
"loss": 0.3046,
"step": 27095
},
{
"epoch": 3.159981343283582,
"grad_norm": 0.4968219837604121,
"learning_rate": 1.9703901417122106e-05,
"loss": 0.3449,
"step": 27100
},
{
"epoch": 3.160564365671642,
"grad_norm": 0.43325286695602533,
"learning_rate": 1.9695763370038846e-05,
"loss": 0.318,
"step": 27105
},
{
"epoch": 3.1611473880597014,
"grad_norm": 0.427183151884983,
"learning_rate": 1.968762648337081e-05,
"loss": 0.3099,
"step": 27110
},
{
"epoch": 3.161730410447761,
"grad_norm": 0.43966371576803287,
"learning_rate": 1.9679490758327862e-05,
"loss": 0.3254,
"step": 27115
},
{
"epoch": 3.1623134328358207,
"grad_norm": 0.5057308075063571,
"learning_rate": 1.96713561961197e-05,
"loss": 0.3507,
"step": 27120
},
{
"epoch": 3.1628964552238807,
"grad_norm": 0.47777550297006854,
"learning_rate": 1.966322279795587e-05,
"loss": 0.3463,
"step": 27125
},
{
"epoch": 3.1634794776119404,
"grad_norm": 0.4559489386403158,
"learning_rate": 1.9655090565045718e-05,
"loss": 0.3263,
"step": 27130
},
{
"epoch": 3.1640625,
"grad_norm": 0.4474189629131407,
"learning_rate": 1.9646959498598444e-05,
"loss": 0.3275,
"step": 27135
},
{
"epoch": 3.1646455223880596,
"grad_norm": 0.4247199729692992,
"learning_rate": 1.9638829599823056e-05,
"loss": 0.3183,
"step": 27140
},
{
"epoch": 3.1652285447761193,
"grad_norm": 0.43245919663374977,
"learning_rate": 1.963070086992837e-05,
"loss": 0.3276,
"step": 27145
},
{
"epoch": 3.1658115671641793,
"grad_norm": 0.4972264534956099,
"learning_rate": 1.9622573310123082e-05,
"loss": 0.3491,
"step": 27150
},
{
"epoch": 3.166394589552239,
"grad_norm": 0.4730961918270202,
"learning_rate": 1.9614446921615654e-05,
"loss": 0.3375,
"step": 27155
},
{
"epoch": 3.1669776119402986,
"grad_norm": 0.4486450978632667,
"learning_rate": 1.9606321705614427e-05,
"loss": 0.3268,
"step": 27160
},
{
"epoch": 3.167560634328358,
"grad_norm": 0.4488841836518065,
"learning_rate": 1.9598197663327534e-05,
"loss": 0.3204,
"step": 27165
},
{
"epoch": 3.168143656716418,
"grad_norm": 0.4696329381013253,
"learning_rate": 1.9590074795962925e-05,
"loss": 0.312,
"step": 27170
},
{
"epoch": 3.1687266791044775,
"grad_norm": 0.48907045259273,
"learning_rate": 1.9581953104728422e-05,
"loss": 0.3477,
"step": 27175
},
{
"epoch": 3.169309701492537,
"grad_norm": 0.4805331764337033,
"learning_rate": 1.957383259083162e-05,
"loss": 0.3229,
"step": 27180
},
{
"epoch": 3.169892723880597,
"grad_norm": 0.4450143918692602,
"learning_rate": 1.9565713255479974e-05,
"loss": 0.3353,
"step": 27185
},
{
"epoch": 3.170475746268657,
"grad_norm": 0.4812438140518185,
"learning_rate": 1.955759509988075e-05,
"loss": 0.3281,
"step": 27190
},
{
"epoch": 3.1710587686567164,
"grad_norm": 0.44373473747948927,
"learning_rate": 1.9549478125241034e-05,
"loss": 0.3445,
"step": 27195
},
{
"epoch": 3.171641791044776,
"grad_norm": 0.4824582958979891,
"learning_rate": 1.9541362332767737e-05,
"loss": 0.3304,
"step": 27200
},
{
"epoch": 3.1722248134328357,
"grad_norm": 0.4719028439016698,
"learning_rate": 1.9533247723667613e-05,
"loss": 0.3346,
"step": 27205
},
{
"epoch": 3.1728078358208953,
"grad_norm": 0.4394771670223666,
"learning_rate": 1.952513429914723e-05,
"loss": 0.3389,
"step": 27210
},
{
"epoch": 3.1733908582089554,
"grad_norm": 0.46966246137225504,
"learning_rate": 1.9517022060412958e-05,
"loss": 0.3287,
"step": 27215
},
{
"epoch": 3.173973880597015,
"grad_norm": 0.46628037746950846,
"learning_rate": 1.950891100867102e-05,
"loss": 0.3299,
"step": 27220
},
{
"epoch": 3.1745569029850746,
"grad_norm": 0.4785703641201323,
"learning_rate": 1.950080114512746e-05,
"loss": 0.3184,
"step": 27225
},
{
"epoch": 3.1751399253731343,
"grad_norm": 0.43906165693485444,
"learning_rate": 1.9492692470988115e-05,
"loss": 0.3165,
"step": 27230
},
{
"epoch": 3.175722947761194,
"grad_norm": 0.4621376834633981,
"learning_rate": 1.9484584987458693e-05,
"loss": 0.3241,
"step": 27235
},
{
"epoch": 3.1763059701492535,
"grad_norm": 0.44326331540932834,
"learning_rate": 1.9476478695744683e-05,
"loss": 0.3209,
"step": 27240
},
{
"epoch": 3.1768889925373136,
"grad_norm": 0.46263449751192637,
"learning_rate": 1.9468373597051404e-05,
"loss": 0.3106,
"step": 27245
},
{
"epoch": 3.1774720149253732,
"grad_norm": 0.4287359278303057,
"learning_rate": 1.9460269692584034e-05,
"loss": 0.3025,
"step": 27250
},
{
"epoch": 3.178055037313433,
"grad_norm": 0.4681246345170257,
"learning_rate": 1.9452166983547516e-05,
"loss": 0.3215,
"step": 27255
},
{
"epoch": 3.1786380597014925,
"grad_norm": 0.4006218678599732,
"learning_rate": 1.944406547114667e-05,
"loss": 0.3101,
"step": 27260
},
{
"epoch": 3.179221082089552,
"grad_norm": 0.4778409445664328,
"learning_rate": 1.9435965156586105e-05,
"loss": 0.3495,
"step": 27265
},
{
"epoch": 3.1798041044776117,
"grad_norm": 0.4521839032924177,
"learning_rate": 1.9427866041070254e-05,
"loss": 0.3302,
"step": 27270
},
{
"epoch": 3.180387126865672,
"grad_norm": 0.451989568859708,
"learning_rate": 1.9419768125803382e-05,
"loss": 0.307,
"step": 27275
},
{
"epoch": 3.1809701492537314,
"grad_norm": 0.5400648266188888,
"learning_rate": 1.9411671411989568e-05,
"loss": 0.3593,
"step": 27280
},
{
"epoch": 3.181553171641791,
"grad_norm": 0.4667878863039788,
"learning_rate": 1.9403575900832726e-05,
"loss": 0.3377,
"step": 27285
},
{
"epoch": 3.1821361940298507,
"grad_norm": 0.4724285012856178,
"learning_rate": 1.9395481593536575e-05,
"loss": 0.3347,
"step": 27290
},
{
"epoch": 3.1827192164179103,
"grad_norm": 0.45862786465029165,
"learning_rate": 1.9387388491304646e-05,
"loss": 0.3074,
"step": 27295
},
{
"epoch": 3.18330223880597,
"grad_norm": 0.47227524445544855,
"learning_rate": 1.937929659534034e-05,
"loss": 0.3252,
"step": 27300
},
{
"epoch": 3.18388526119403,
"grad_norm": 0.43162785406712434,
"learning_rate": 1.9371205906846808e-05,
"loss": 0.3083,
"step": 27305
},
{
"epoch": 3.1844682835820897,
"grad_norm": 0.4876120836497574,
"learning_rate": 1.9363116427027084e-05,
"loss": 0.3162,
"step": 27310
},
{
"epoch": 3.1850513059701493,
"grad_norm": 0.46307715110791076,
"learning_rate": 1.9355028157083988e-05,
"loss": 0.3501,
"step": 27315
},
{
"epoch": 3.185634328358209,
"grad_norm": 0.46941187892021596,
"learning_rate": 1.9346941098220157e-05,
"loss": 0.3266,
"step": 27320
},
{
"epoch": 3.1862173507462686,
"grad_norm": 0.47493550264768114,
"learning_rate": 1.933885525163807e-05,
"loss": 0.3392,
"step": 27325
},
{
"epoch": 3.186800373134328,
"grad_norm": 0.47281274512403776,
"learning_rate": 1.933077061854002e-05,
"loss": 0.3309,
"step": 27330
},
{
"epoch": 3.1873833955223883,
"grad_norm": 0.4349454178005739,
"learning_rate": 1.9322687200128103e-05,
"loss": 0.3313,
"step": 27335
},
{
"epoch": 3.187966417910448,
"grad_norm": 0.44484068267761423,
"learning_rate": 1.931460499760426e-05,
"loss": 0.3116,
"step": 27340
},
{
"epoch": 3.1885494402985075,
"grad_norm": 0.4631234982395724,
"learning_rate": 1.930652401217021e-05,
"loss": 0.3332,
"step": 27345
},
{
"epoch": 3.189132462686567,
"grad_norm": 0.44363469283627516,
"learning_rate": 1.929844424502755e-05,
"loss": 0.32,
"step": 27350
},
{
"epoch": 3.1897154850746268,
"grad_norm": 0.5142219286656142,
"learning_rate": 1.929036569737765e-05,
"loss": 0.333,
"step": 27355
},
{
"epoch": 3.1902985074626864,
"grad_norm": 0.41970889457686733,
"learning_rate": 1.9282288370421708e-05,
"loss": 0.3154,
"step": 27360
},
{
"epoch": 3.1908815298507465,
"grad_norm": 0.4469941346947236,
"learning_rate": 1.9274212265360757e-05,
"loss": 0.3214,
"step": 27365
},
{
"epoch": 3.191464552238806,
"grad_norm": 0.4692755715896205,
"learning_rate": 1.9266137383395626e-05,
"loss": 0.311,
"step": 27370
},
{
"epoch": 3.1920475746268657,
"grad_norm": 0.46469323415769187,
"learning_rate": 1.925806372572697e-05,
"loss": 0.3258,
"step": 27375
},
{
"epoch": 3.1926305970149254,
"grad_norm": 0.4573763955838952,
"learning_rate": 1.9249991293555276e-05,
"loss": 0.3233,
"step": 27380
},
{
"epoch": 3.193213619402985,
"grad_norm": 0.43970584272676727,
"learning_rate": 1.9241920088080833e-05,
"loss": 0.3098,
"step": 27385
},
{
"epoch": 3.1937966417910446,
"grad_norm": 0.43208933972253044,
"learning_rate": 1.9233850110503748e-05,
"loss": 0.3066,
"step": 27390
},
{
"epoch": 3.1943796641791047,
"grad_norm": 0.47906991700293433,
"learning_rate": 1.9225781362023955e-05,
"loss": 0.3139,
"step": 27395
},
{
"epoch": 3.1949626865671643,
"grad_norm": 0.4930373154110795,
"learning_rate": 1.9217713843841195e-05,
"loss": 0.3228,
"step": 27400
},
{
"epoch": 3.195545708955224,
"grad_norm": 0.44358866677645276,
"learning_rate": 1.9209647557155025e-05,
"loss": 0.33,
"step": 27405
},
{
"epoch": 3.1961287313432836,
"grad_norm": 0.49269024353102203,
"learning_rate": 1.9201582503164845e-05,
"loss": 0.3356,
"step": 27410
},
{
"epoch": 3.196711753731343,
"grad_norm": 0.4696815736959036,
"learning_rate": 1.9193518683069833e-05,
"loss": 0.301,
"step": 27415
},
{
"epoch": 3.197294776119403,
"grad_norm": 0.4871495656179168,
"learning_rate": 1.9185456098068998e-05,
"loss": 0.3408,
"step": 27420
},
{
"epoch": 3.197877798507463,
"grad_norm": 0.4213310904332378,
"learning_rate": 1.9177394749361193e-05,
"loss": 0.3099,
"step": 27425
},
{
"epoch": 3.1984608208955225,
"grad_norm": 0.47862103029689884,
"learning_rate": 1.9169334638145037e-05,
"loss": 0.2995,
"step": 27430
},
{
"epoch": 3.199043843283582,
"grad_norm": 0.44777242909659876,
"learning_rate": 1.9161275765619007e-05,
"loss": 0.3192,
"step": 27435
},
{
"epoch": 3.199626865671642,
"grad_norm": 0.47406176834663966,
"learning_rate": 1.9153218132981375e-05,
"loss": 0.3321,
"step": 27440
},
{
"epoch": 3.2002098880597014,
"grad_norm": 0.47048401955783853,
"learning_rate": 1.9145161741430234e-05,
"loss": 0.3217,
"step": 27445
},
{
"epoch": 3.200792910447761,
"grad_norm": 0.4531181811978807,
"learning_rate": 1.9137106592163495e-05,
"loss": 0.3308,
"step": 27450
},
{
"epoch": 3.2013759328358207,
"grad_norm": 0.45942249968126514,
"learning_rate": 1.9129052686378873e-05,
"loss": 0.3182,
"step": 27455
},
{
"epoch": 3.2019589552238807,
"grad_norm": 0.42095683723958044,
"learning_rate": 1.912100002527392e-05,
"loss": 0.3243,
"step": 27460
},
{
"epoch": 3.2025419776119404,
"grad_norm": 0.45713255314341616,
"learning_rate": 1.9112948610045982e-05,
"loss": 0.3339,
"step": 27465
},
{
"epoch": 3.203125,
"grad_norm": 0.47613917907366915,
"learning_rate": 1.9104898441892222e-05,
"loss": 0.3394,
"step": 27470
},
{
"epoch": 3.2037080223880596,
"grad_norm": 0.4594895192147124,
"learning_rate": 1.909684952200964e-05,
"loss": 0.3252,
"step": 27475
},
{
"epoch": 3.2042910447761193,
"grad_norm": 0.962513676859494,
"learning_rate": 1.9088801851595008e-05,
"loss": 0.3173,
"step": 27480
},
{
"epoch": 3.2048740671641793,
"grad_norm": 0.43709419203162797,
"learning_rate": 1.908075543184496e-05,
"loss": 0.3018,
"step": 27485
},
{
"epoch": 3.205457089552239,
"grad_norm": 0.440565691190322,
"learning_rate": 1.907271026395592e-05,
"loss": 0.3129,
"step": 27490
},
{
"epoch": 3.2060401119402986,
"grad_norm": 0.5112532590069527,
"learning_rate": 1.90646663491241e-05,
"loss": 0.3485,
"step": 27495
},
{
"epoch": 3.206623134328358,
"grad_norm": 0.47127425580398696,
"learning_rate": 1.9056623688545588e-05,
"loss": 0.3242,
"step": 27500
},
{
"epoch": 3.207206156716418,
"grad_norm": 0.4464146934009306,
"learning_rate": 1.904858228341623e-05,
"loss": 0.339,
"step": 27505
},
{
"epoch": 3.2077891791044775,
"grad_norm": 0.46980098724748476,
"learning_rate": 1.9040542134931715e-05,
"loss": 0.3097,
"step": 27510
},
{
"epoch": 3.208372201492537,
"grad_norm": 0.440758758693857,
"learning_rate": 1.9032503244287537e-05,
"loss": 0.3312,
"step": 27515
},
{
"epoch": 3.208955223880597,
"grad_norm": 0.47824340765477763,
"learning_rate": 1.9024465612678993e-05,
"loss": 0.3373,
"step": 27520
},
{
"epoch": 3.209538246268657,
"grad_norm": 0.46386513781546374,
"learning_rate": 1.901642924130121e-05,
"loss": 0.3404,
"step": 27525
},
{
"epoch": 3.2101212686567164,
"grad_norm": 0.47197157493893405,
"learning_rate": 1.900839413134911e-05,
"loss": 0.3304,
"step": 27530
},
{
"epoch": 3.210704291044776,
"grad_norm": 0.43944612589420196,
"learning_rate": 1.9000360284017448e-05,
"loss": 0.3278,
"step": 27535
},
{
"epoch": 3.2112873134328357,
"grad_norm": 0.46241840789993244,
"learning_rate": 1.8992327700500772e-05,
"loss": 0.3174,
"step": 27540
},
{
"epoch": 3.2118703358208953,
"grad_norm": 0.4233215422194263,
"learning_rate": 1.8984296381993454e-05,
"loss": 0.3249,
"step": 27545
},
{
"epoch": 3.2124533582089554,
"grad_norm": 0.4165212543451875,
"learning_rate": 1.897626632968968e-05,
"loss": 0.2915,
"step": 27550
},
{
"epoch": 3.213036380597015,
"grad_norm": 0.4079197063432625,
"learning_rate": 1.8968237544783423e-05,
"loss": 0.3321,
"step": 27555
},
{
"epoch": 3.2136194029850746,
"grad_norm": 0.5014130420173903,
"learning_rate": 1.8960210028468512e-05,
"loss": 0.3392,
"step": 27560
},
{
"epoch": 3.2142024253731343,
"grad_norm": 0.47986743675377647,
"learning_rate": 1.8952183781938538e-05,
"loss": 0.3203,
"step": 27565
},
{
"epoch": 3.214785447761194,
"grad_norm": 0.4808351354951185,
"learning_rate": 1.8944158806386942e-05,
"loss": 0.3286,
"step": 27570
},
{
"epoch": 3.2153684701492535,
"grad_norm": 0.47026680395669446,
"learning_rate": 1.8936135103006957e-05,
"loss": 0.3394,
"step": 27575
},
{
"epoch": 3.2159514925373136,
"grad_norm": 0.44150994897363893,
"learning_rate": 1.8928112672991626e-05,
"loss": 0.3101,
"step": 27580
},
{
"epoch": 3.2165345149253732,
"grad_norm": 0.4828026626153867,
"learning_rate": 1.8920091517533818e-05,
"loss": 0.306,
"step": 27585
},
{
"epoch": 3.217117537313433,
"grad_norm": 0.8046588090155792,
"learning_rate": 1.8912071637826196e-05,
"loss": 0.3207,
"step": 27590
},
{
"epoch": 3.2177005597014925,
"grad_norm": 0.4941742594832407,
"learning_rate": 1.890405303506123e-05,
"loss": 0.3495,
"step": 27595
},
{
"epoch": 3.218283582089552,
"grad_norm": 0.4703362026510565,
"learning_rate": 1.8896035710431225e-05,
"loss": 0.3565,
"step": 27600
},
{
"epoch": 3.2188666044776117,
"grad_norm": 0.495189929149459,
"learning_rate": 1.888801966512827e-05,
"loss": 0.3453,
"step": 27605
},
{
"epoch": 3.219449626865672,
"grad_norm": 0.4719978449184704,
"learning_rate": 1.8880004900344283e-05,
"loss": 0.3315,
"step": 27610
},
{
"epoch": 3.2200326492537314,
"grad_norm": 0.49674022335482504,
"learning_rate": 1.8871991417270978e-05,
"loss": 0.3552,
"step": 27615
},
{
"epoch": 3.220615671641791,
"grad_norm": 0.4393130089474178,
"learning_rate": 1.8863979217099874e-05,
"loss": 0.3151,
"step": 27620
},
{
"epoch": 3.2211986940298507,
"grad_norm": 0.4523050799744314,
"learning_rate": 1.8855968301022326e-05,
"loss": 0.3213,
"step": 27625
},
{
"epoch": 3.2217817164179103,
"grad_norm": 0.6886153682176874,
"learning_rate": 1.8847958670229465e-05,
"loss": 0.3294,
"step": 27630
},
{
"epoch": 3.22236473880597,
"grad_norm": 0.44038399138162393,
"learning_rate": 1.883995032591226e-05,
"loss": 0.3286,
"step": 27635
},
{
"epoch": 3.22294776119403,
"grad_norm": 0.43311241772978903,
"learning_rate": 1.8831943269261467e-05,
"loss": 0.3283,
"step": 27640
},
{
"epoch": 3.2235307835820897,
"grad_norm": 0.4388420255717518,
"learning_rate": 1.8823937501467648e-05,
"loss": 0.3235,
"step": 27645
},
{
"epoch": 3.2241138059701493,
"grad_norm": 0.44877364529184444,
"learning_rate": 1.8815933023721206e-05,
"loss": 0.3396,
"step": 27650
},
{
"epoch": 3.224696828358209,
"grad_norm": 0.44087057548560593,
"learning_rate": 1.880792983721231e-05,
"loss": 0.3292,
"step": 27655
},
{
"epoch": 3.2252798507462686,
"grad_norm": 0.4567254636672784,
"learning_rate": 1.8799927943130986e-05,
"loss": 0.3317,
"step": 27660
},
{
"epoch": 3.225862873134328,
"grad_norm": 0.43552549652640776,
"learning_rate": 1.879192734266701e-05,
"loss": 0.3236,
"step": 27665
},
{
"epoch": 3.2264458955223883,
"grad_norm": 0.4799781666815651,
"learning_rate": 1.878392803701e-05,
"loss": 0.3337,
"step": 27670
},
{
"epoch": 3.227028917910448,
"grad_norm": 0.4468328174907116,
"learning_rate": 1.8775930027349386e-05,
"loss": 0.3259,
"step": 27675
},
{
"epoch": 3.2276119402985075,
"grad_norm": 0.470926021553355,
"learning_rate": 1.8767933314874382e-05,
"loss": 0.3499,
"step": 27680
},
{
"epoch": 3.228194962686567,
"grad_norm": 0.4926736932412245,
"learning_rate": 1.8759937900774038e-05,
"loss": 0.3498,
"step": 27685
},
{
"epoch": 3.2287779850746268,
"grad_norm": 0.4725154869096927,
"learning_rate": 1.875194378623718e-05,
"loss": 0.3272,
"step": 27690
},
{
"epoch": 3.2293610074626864,
"grad_norm": 0.47622368975208734,
"learning_rate": 1.8743950972452477e-05,
"loss": 0.3238,
"step": 27695
},
{
"epoch": 3.2299440298507465,
"grad_norm": 0.5072568601536372,
"learning_rate": 1.8735959460608364e-05,
"loss": 0.3509,
"step": 27700
},
{
"epoch": 3.230527052238806,
"grad_norm": 0.5023368884886755,
"learning_rate": 1.8727969251893107e-05,
"loss": 0.3198,
"step": 27705
},
{
"epoch": 3.2311100746268657,
"grad_norm": 0.5256277684008088,
"learning_rate": 1.871998034749478e-05,
"loss": 0.3401,
"step": 27710
},
{
"epoch": 3.2316930970149254,
"grad_norm": 0.5022381624468324,
"learning_rate": 1.8711992748601252e-05,
"loss": 0.3458,
"step": 27715
},
{
"epoch": 3.232276119402985,
"grad_norm": 0.4608319003836731,
"learning_rate": 1.8704006456400202e-05,
"loss": 0.3321,
"step": 27720
},
{
"epoch": 3.2328591417910446,
"grad_norm": 0.4706421932327294,
"learning_rate": 1.8696021472079118e-05,
"loss": 0.3253,
"step": 27725
},
{
"epoch": 3.2334421641791047,
"grad_norm": 0.49681056697961345,
"learning_rate": 1.8688037796825285e-05,
"loss": 0.3299,
"step": 27730
},
{
"epoch": 3.2340251865671643,
"grad_norm": 0.4590968039711858,
"learning_rate": 1.8680055431825804e-05,
"loss": 0.3209,
"step": 27735
},
{
"epoch": 3.234608208955224,
"grad_norm": 0.44550716999248235,
"learning_rate": 1.8672074378267573e-05,
"loss": 0.3324,
"step": 27740
},
{
"epoch": 3.2351912313432836,
"grad_norm": 0.4470425587062118,
"learning_rate": 1.8664094637337303e-05,
"loss": 0.3317,
"step": 27745
},
{
"epoch": 3.235774253731343,
"grad_norm": 0.48446303867359625,
"learning_rate": 1.8656116210221502e-05,
"loss": 0.3259,
"step": 27750
},
{
"epoch": 3.236357276119403,
"grad_norm": 0.4840083407763402,
"learning_rate": 1.8648139098106482e-05,
"loss": 0.3369,
"step": 27755
},
{
"epoch": 3.236940298507463,
"grad_norm": 0.44067379565281556,
"learning_rate": 1.8640163302178377e-05,
"loss": 0.3355,
"step": 27760
},
{
"epoch": 3.2375233208955225,
"grad_norm": 0.49928397720096795,
"learning_rate": 1.8632188823623086e-05,
"loss": 0.3429,
"step": 27765
},
{
"epoch": 3.238106343283582,
"grad_norm": 0.49338482903304315,
"learning_rate": 1.8624215663626365e-05,
"loss": 0.3184,
"step": 27770
},
{
"epoch": 3.238689365671642,
"grad_norm": 0.4649169797107502,
"learning_rate": 1.861624382337373e-05,
"loss": 0.3704,
"step": 27775
},
{
"epoch": 3.2392723880597014,
"grad_norm": 0.45303669376757155,
"learning_rate": 1.8608273304050515e-05,
"loss": 0.3168,
"step": 27780
},
{
"epoch": 3.239855410447761,
"grad_norm": 0.7655008250028477,
"learning_rate": 1.8600304106841876e-05,
"loss": 0.3337,
"step": 27785
},
{
"epoch": 3.2404384328358207,
"grad_norm": 0.4386919846379524,
"learning_rate": 1.859233623293274e-05,
"loss": 0.3179,
"step": 27790
},
{
"epoch": 3.2410214552238807,
"grad_norm": 0.4965513787903615,
"learning_rate": 1.8584369683507863e-05,
"loss": 0.3437,
"step": 27795
},
{
"epoch": 3.2416044776119404,
"grad_norm": 0.46268733115301086,
"learning_rate": 1.8576404459751796e-05,
"loss": 0.3239,
"step": 27800
},
{
"epoch": 3.2421875,
"grad_norm": 0.4545564327310699,
"learning_rate": 1.8568440562848876e-05,
"loss": 0.3211,
"step": 27805
},
{
"epoch": 3.2427705223880596,
"grad_norm": 0.4934358035272012,
"learning_rate": 1.8560477993983284e-05,
"loss": 0.343,
"step": 27810
},
{
"epoch": 3.2433535447761193,
"grad_norm": 0.422853285427749,
"learning_rate": 1.855251675433895e-05,
"loss": 0.3235,
"step": 27815
},
{
"epoch": 3.2439365671641793,
"grad_norm": 0.4751754389204498,
"learning_rate": 1.8544556845099657e-05,
"loss": 0.3305,
"step": 27820
},
{
"epoch": 3.244519589552239,
"grad_norm": 0.4647433580608973,
"learning_rate": 1.8536598267448958e-05,
"loss": 0.3384,
"step": 27825
},
{
"epoch": 3.2451026119402986,
"grad_norm": 0.4540116861372445,
"learning_rate": 1.8528641022570202e-05,
"loss": 0.3324,
"step": 27830
},
{
"epoch": 3.245685634328358,
"grad_norm": 0.521097018048201,
"learning_rate": 1.8520685111646585e-05,
"loss": 0.3368,
"step": 27835
},
{
"epoch": 3.246268656716418,
"grad_norm": 0.46427046263451505,
"learning_rate": 1.851273053586105e-05,
"loss": 0.3115,
"step": 27840
},
{
"epoch": 3.2468516791044775,
"grad_norm": 0.4662757007968462,
"learning_rate": 1.850477729639638e-05,
"loss": 0.3373,
"step": 27845
},
{
"epoch": 3.247434701492537,
"grad_norm": 0.4410616567724678,
"learning_rate": 1.8496825394435146e-05,
"loss": 0.3283,
"step": 27850
},
{
"epoch": 3.248017723880597,
"grad_norm": 0.4669657380849723,
"learning_rate": 1.8488874831159703e-05,
"loss": 0.3401,
"step": 27855
},
{
"epoch": 3.248600746268657,
"grad_norm": 0.5169708732045697,
"learning_rate": 1.8480925607752248e-05,
"loss": 0.3328,
"step": 27860
},
{
"epoch": 3.2491837686567164,
"grad_norm": 0.4820043528671602,
"learning_rate": 1.847297772539473e-05,
"loss": 0.319,
"step": 27865
},
{
"epoch": 3.249766791044776,
"grad_norm": 0.48121569675310194,
"learning_rate": 1.8465031185268943e-05,
"loss": 0.3708,
"step": 27870
},
{
"epoch": 3.2503498134328357,
"grad_norm": 0.4156510208095905,
"learning_rate": 1.845708598855645e-05,
"loss": 0.3327,
"step": 27875
},
{
"epoch": 3.2509328358208958,
"grad_norm": 0.44055453521381466,
"learning_rate": 1.8449142136438628e-05,
"loss": 0.3186,
"step": 27880
},
{
"epoch": 3.2515158582089554,
"grad_norm": 0.46750972785844375,
"learning_rate": 1.8441199630096655e-05,
"loss": 0.333,
"step": 27885
},
{
"epoch": 3.252098880597015,
"grad_norm": 0.4822002310139299,
"learning_rate": 1.84332584707115e-05,
"loss": 0.3268,
"step": 27890
},
{
"epoch": 3.2526819029850746,
"grad_norm": 0.45841592667995923,
"learning_rate": 1.8425318659463942e-05,
"loss": 0.3139,
"step": 27895
},
{
"epoch": 3.2532649253731343,
"grad_norm": 0.46469078456422636,
"learning_rate": 1.8417380197534558e-05,
"loss": 0.3392,
"step": 27900
},
{
"epoch": 3.253847947761194,
"grad_norm": 0.44008698671573293,
"learning_rate": 1.8409443086103713e-05,
"loss": 0.3288,
"step": 27905
},
{
"epoch": 3.2544309701492535,
"grad_norm": 0.42758635449004245,
"learning_rate": 1.8401507326351575e-05,
"loss": 0.3285,
"step": 27910
},
{
"epoch": 3.2550139925373136,
"grad_norm": 0.45371761745377537,
"learning_rate": 1.839357291945813e-05,
"loss": 0.3325,
"step": 27915
},
{
"epoch": 3.2555970149253732,
"grad_norm": 0.42570080917107816,
"learning_rate": 1.8385639866603144e-05,
"loss": 0.3385,
"step": 27920
},
{
"epoch": 3.256180037313433,
"grad_norm": 0.47613616679112847,
"learning_rate": 1.837770816896618e-05,
"loss": 0.3464,
"step": 27925
},
{
"epoch": 3.2567630597014925,
"grad_norm": 0.4875952718236274,
"learning_rate": 1.836977782772661e-05,
"loss": 0.3269,
"step": 27930
},
{
"epoch": 3.257346082089552,
"grad_norm": 0.4574369613739823,
"learning_rate": 1.8361848844063594e-05,
"loss": 0.3313,
"step": 27935
},
{
"epoch": 3.2579291044776117,
"grad_norm": 0.4703903863214078,
"learning_rate": 1.8353921219156102e-05,
"loss": 0.3297,
"step": 27940
},
{
"epoch": 3.258512126865672,
"grad_norm": 0.47177636995396366,
"learning_rate": 1.83459949541829e-05,
"loss": 0.3202,
"step": 27945
},
{
"epoch": 3.2590951492537314,
"grad_norm": 0.45047096290180144,
"learning_rate": 1.8338070050322544e-05,
"loss": 0.3508,
"step": 27950
},
{
"epoch": 3.259678171641791,
"grad_norm": 0.4765900942250275,
"learning_rate": 1.8330146508753377e-05,
"loss": 0.338,
"step": 27955
},
{
"epoch": 3.2602611940298507,
"grad_norm": 0.46011900679469064,
"learning_rate": 1.8322224330653576e-05,
"loss": 0.3252,
"step": 27960
},
{
"epoch": 3.2608442164179103,
"grad_norm": 0.4515748400652757,
"learning_rate": 1.8314303517201076e-05,
"loss": 0.3283,
"step": 27965
},
{
"epoch": 3.26142723880597,
"grad_norm": 0.46490340258957585,
"learning_rate": 1.830638406957364e-05,
"loss": 0.328,
"step": 27970
},
{
"epoch": 3.26201026119403,
"grad_norm": 0.43386741663300754,
"learning_rate": 1.829846598894881e-05,
"loss": 0.3362,
"step": 27975
},
{
"epoch": 3.2625932835820897,
"grad_norm": 0.45571419509069383,
"learning_rate": 1.8290549276503915e-05,
"loss": 0.3349,
"step": 27980
},
{
"epoch": 3.2631763059701493,
"grad_norm": 0.4826439613582884,
"learning_rate": 1.8282633933416115e-05,
"loss": 0.3425,
"step": 27985
},
{
"epoch": 3.263759328358209,
"grad_norm": 0.45438105183907856,
"learning_rate": 1.8274719960862325e-05,
"loss": 0.3282,
"step": 27990
},
{
"epoch": 3.2643423507462686,
"grad_norm": 0.4754247902304436,
"learning_rate": 1.82668073600193e-05,
"loss": 0.3252,
"step": 27995
},
{
"epoch": 3.264925373134328,
"grad_norm": 0.46722667906052695,
"learning_rate": 1.825889613206355e-05,
"loss": 0.335,
"step": 28000
},
{
"epoch": 3.2655083955223883,
"grad_norm": 0.46493889162506147,
"learning_rate": 1.82509862781714e-05,
"loss": 0.3212,
"step": 28005
},
{
"epoch": 3.266091417910448,
"grad_norm": 0.4612213513933666,
"learning_rate": 1.824307779951898e-05,
"loss": 0.3217,
"step": 28010
},
{
"epoch": 3.2666744402985075,
"grad_norm": 0.4643578057614236,
"learning_rate": 1.8235170697282194e-05,
"loss": 0.3108,
"step": 28015
},
{
"epoch": 3.267257462686567,
"grad_norm": 0.44275932576560134,
"learning_rate": 1.8227264972636758e-05,
"loss": 0.3018,
"step": 28020
},
{
"epoch": 3.2678404850746268,
"grad_norm": 0.46553997310047907,
"learning_rate": 1.8219360626758185e-05,
"loss": 0.3301,
"step": 28025
},
{
"epoch": 3.2684235074626864,
"grad_norm": 0.46688242085330117,
"learning_rate": 1.821145766082176e-05,
"loss": 0.3389,
"step": 28030
},
{
"epoch": 3.2690065298507465,
"grad_norm": 0.45708609894384883,
"learning_rate": 1.820355607600259e-05,
"loss": 0.3459,
"step": 28035
},
{
"epoch": 3.269589552238806,
"grad_norm": 0.4591085831877663,
"learning_rate": 1.8195655873475554e-05,
"loss": 0.3447,
"step": 28040
},
{
"epoch": 3.2701725746268657,
"grad_norm": 0.43371501436425636,
"learning_rate": 1.818775705441535e-05,
"loss": 0.3085,
"step": 28045
},
{
"epoch": 3.2707555970149254,
"grad_norm": 0.4565167376854876,
"learning_rate": 1.8179859619996448e-05,
"loss": 0.3408,
"step": 28050
},
{
"epoch": 3.271338619402985,
"grad_norm": 0.44336035486171527,
"learning_rate": 1.8171963571393112e-05,
"loss": 0.333,
"step": 28055
},
{
"epoch": 3.2719216417910446,
"grad_norm": 0.451954050568279,
"learning_rate": 1.8164068909779437e-05,
"loss": 0.3195,
"step": 28060
},
{
"epoch": 3.2725046641791042,
"grad_norm": 0.5202880557337407,
"learning_rate": 1.8156175636329252e-05,
"loss": 0.3344,
"step": 28065
},
{
"epoch": 3.2730876865671643,
"grad_norm": 0.5296368128506362,
"learning_rate": 1.814828375221623e-05,
"loss": 0.3298,
"step": 28070
},
{
"epoch": 3.273670708955224,
"grad_norm": 0.43793489149007375,
"learning_rate": 1.814039325861382e-05,
"loss": 0.3353,
"step": 28075
},
{
"epoch": 3.2742537313432836,
"grad_norm": 0.4584593923335097,
"learning_rate": 1.8132504156695245e-05,
"loss": 0.3334,
"step": 28080
},
{
"epoch": 3.274836753731343,
"grad_norm": 0.4861907953454029,
"learning_rate": 1.812461644763355e-05,
"loss": 0.3429,
"step": 28085
},
{
"epoch": 3.275419776119403,
"grad_norm": 0.46246625320099805,
"learning_rate": 1.8116730132601565e-05,
"loss": 0.3217,
"step": 28090
},
{
"epoch": 3.276002798507463,
"grad_norm": 0.47876767332398457,
"learning_rate": 1.8108845212771913e-05,
"loss": 0.3137,
"step": 28095
},
{
"epoch": 3.2765858208955225,
"grad_norm": 0.44409562799986846,
"learning_rate": 1.8100961689317003e-05,
"loss": 0.3318,
"step": 28100
},
{
"epoch": 3.277168843283582,
"grad_norm": 0.40478347277432103,
"learning_rate": 1.8093079563409017e-05,
"loss": 0.3146,
"step": 28105
},
{
"epoch": 3.277751865671642,
"grad_norm": 0.47570380198610074,
"learning_rate": 1.808519883621999e-05,
"loss": 0.3245,
"step": 28110
},
{
"epoch": 3.2783348880597014,
"grad_norm": 0.45409502237373334,
"learning_rate": 1.807731950892167e-05,
"loss": 0.3191,
"step": 28115
},
{
"epoch": 3.278917910447761,
"grad_norm": 0.47488286340939584,
"learning_rate": 1.806944158268568e-05,
"loss": 0.3346,
"step": 28120
},
{
"epoch": 3.2795009328358207,
"grad_norm": 0.4421012793522906,
"learning_rate": 1.8061565058683362e-05,
"loss": 0.3258,
"step": 28125
},
{
"epoch": 3.2800839552238807,
"grad_norm": 0.4359737526431151,
"learning_rate": 1.805368993808589e-05,
"loss": 0.3119,
"step": 28130
},
{
"epoch": 3.2806669776119404,
"grad_norm": 0.4714093924509547,
"learning_rate": 1.8045816222064212e-05,
"loss": 0.3295,
"step": 28135
},
{
"epoch": 3.28125,
"grad_norm": 0.48000776117763994,
"learning_rate": 1.803794391178908e-05,
"loss": 0.3311,
"step": 28140
},
{
"epoch": 3.2818330223880596,
"grad_norm": 0.45661806602343735,
"learning_rate": 1.8030073008431032e-05,
"loss": 0.3432,
"step": 28145
},
{
"epoch": 3.2824160447761193,
"grad_norm": 0.4398196977266842,
"learning_rate": 1.8022203513160406e-05,
"loss": 0.3021,
"step": 28150
},
{
"epoch": 3.2829990671641793,
"grad_norm": 0.48306013572476025,
"learning_rate": 1.8014335427147288e-05,
"loss": 0.3325,
"step": 28155
},
{
"epoch": 3.283582089552239,
"grad_norm": 0.44625314047636055,
"learning_rate": 1.8006468751561628e-05,
"loss": 0.3257,
"step": 28160
},
{
"epoch": 3.2841651119402986,
"grad_norm": 0.4466349940081072,
"learning_rate": 1.7998603487573095e-05,
"loss": 0.3206,
"step": 28165
},
{
"epoch": 3.284748134328358,
"grad_norm": 0.4676684095924915,
"learning_rate": 1.7990739636351188e-05,
"loss": 0.3321,
"step": 28170
},
{
"epoch": 3.285331156716418,
"grad_norm": 0.46672320227419334,
"learning_rate": 1.798287719906519e-05,
"loss": 0.307,
"step": 28175
},
{
"epoch": 3.2859141791044775,
"grad_norm": 0.44967161705789455,
"learning_rate": 1.797501617688417e-05,
"loss": 0.3159,
"step": 28180
},
{
"epoch": 3.286497201492537,
"grad_norm": 0.4531141777991533,
"learning_rate": 1.7967156570976974e-05,
"loss": 0.3355,
"step": 28185
},
{
"epoch": 3.287080223880597,
"grad_norm": 0.41605086071344555,
"learning_rate": 1.795929838251227e-05,
"loss": 0.3208,
"step": 28190
},
{
"epoch": 3.287663246268657,
"grad_norm": 0.4785387238146306,
"learning_rate": 1.7951441612658488e-05,
"loss": 0.3427,
"step": 28195
},
{
"epoch": 3.2882462686567164,
"grad_norm": 0.4416147481573904,
"learning_rate": 1.7943586262583846e-05,
"loss": 0.3247,
"step": 28200
},
{
"epoch": 3.288829291044776,
"grad_norm": 0.48887574830993924,
"learning_rate": 1.7935732333456362e-05,
"loss": 0.3474,
"step": 28205
},
{
"epoch": 3.2894123134328357,
"grad_norm": 0.4048883473534703,
"learning_rate": 1.7927879826443844e-05,
"loss": 0.3243,
"step": 28210
},
{
"epoch": 3.2899953358208958,
"grad_norm": 0.4871027223708983,
"learning_rate": 1.7920028742713885e-05,
"loss": 0.3483,
"step": 28215
},
{
"epoch": 3.2905783582089554,
"grad_norm": 0.4446307158366329,
"learning_rate": 1.791217908343386e-05,
"loss": 0.3227,
"step": 28220
},
{
"epoch": 3.291161380597015,
"grad_norm": 0.46756764098881587,
"learning_rate": 1.7904330849770957e-05,
"loss": 0.3263,
"step": 28225
},
{
"epoch": 3.2917444029850746,
"grad_norm": 0.48562959405852096,
"learning_rate": 1.78964840428921e-05,
"loss": 0.334,
"step": 28230
},
{
"epoch": 3.2923274253731343,
"grad_norm": 0.47108291908589084,
"learning_rate": 1.788863866396407e-05,
"loss": 0.3198,
"step": 28235
},
{
"epoch": 3.292910447761194,
"grad_norm": 0.45797650806212836,
"learning_rate": 1.7880794714153366e-05,
"loss": 0.3291,
"step": 28240
},
{
"epoch": 3.2934934701492535,
"grad_norm": 0.46790517134111836,
"learning_rate": 1.7872952194626337e-05,
"loss": 0.3249,
"step": 28245
},
{
"epoch": 3.2940764925373136,
"grad_norm": 0.4753403323837971,
"learning_rate": 1.786511110654907e-05,
"loss": 0.3446,
"step": 28250
},
{
"epoch": 3.2946595149253732,
"grad_norm": 0.44769364250340016,
"learning_rate": 1.785727145108747e-05,
"loss": 0.3241,
"step": 28255
},
{
"epoch": 3.295242537313433,
"grad_norm": 0.45298169045203396,
"learning_rate": 1.784943322940722e-05,
"loss": 0.3369,
"step": 28260
},
{
"epoch": 3.2958255597014925,
"grad_norm": 0.44704865585719183,
"learning_rate": 1.7841596442673785e-05,
"loss": 0.3225,
"step": 28265
},
{
"epoch": 3.296408582089552,
"grad_norm": 0.4229007651588271,
"learning_rate": 1.7833761092052415e-05,
"loss": 0.3268,
"step": 28270
},
{
"epoch": 3.2969916044776117,
"grad_norm": 0.4693133143323309,
"learning_rate": 1.782592717870817e-05,
"loss": 0.3252,
"step": 28275
},
{
"epoch": 3.297574626865672,
"grad_norm": 0.45258144180134846,
"learning_rate": 1.7818094703805837e-05,
"loss": 0.3169,
"step": 28280
},
{
"epoch": 3.2981576492537314,
"grad_norm": 0.47177011099188704,
"learning_rate": 1.7810263668510084e-05,
"loss": 0.3338,
"step": 28285
},
{
"epoch": 3.298740671641791,
"grad_norm": 0.4894847741427928,
"learning_rate": 1.780243407398527e-05,
"loss": 0.3199,
"step": 28290
},
{
"epoch": 3.2993236940298507,
"grad_norm": 0.5215752438023614,
"learning_rate": 1.7794605921395607e-05,
"loss": 0.3502,
"step": 28295
},
{
"epoch": 3.2999067164179103,
"grad_norm": 0.5104828930950608,
"learning_rate": 1.7786779211905048e-05,
"loss": 0.3469,
"step": 28300
},
{
"epoch": 3.30048973880597,
"grad_norm": 0.4475261165375108,
"learning_rate": 1.7778953946677353e-05,
"loss": 0.3341,
"step": 28305
},
{
"epoch": 3.30107276119403,
"grad_norm": 0.4698180033120021,
"learning_rate": 1.7771130126876068e-05,
"loss": 0.3309,
"step": 28310
},
{
"epoch": 3.3016557835820897,
"grad_norm": 0.45325965435071763,
"learning_rate": 1.776330775366452e-05,
"loss": 0.3092,
"step": 28315
},
{
"epoch": 3.3022388059701493,
"grad_norm": 0.465617570605433,
"learning_rate": 1.775548682820582e-05,
"loss": 0.3468,
"step": 28320
},
{
"epoch": 3.302821828358209,
"grad_norm": 0.4542573607391547,
"learning_rate": 1.774766735166287e-05,
"loss": 0.3085,
"step": 28325
},
{
"epoch": 3.3034048507462686,
"grad_norm": 0.4345564294184516,
"learning_rate": 1.7739849325198334e-05,
"loss": 0.3117,
"step": 28330
},
{
"epoch": 3.303987873134328,
"grad_norm": 0.4843823089731792,
"learning_rate": 1.7732032749974704e-05,
"loss": 0.2899,
"step": 28335
},
{
"epoch": 3.3045708955223883,
"grad_norm": 0.4660624778568792,
"learning_rate": 1.7724217627154204e-05,
"loss": 0.3313,
"step": 28340
},
{
"epoch": 3.305153917910448,
"grad_norm": 0.4837352189507676,
"learning_rate": 1.771640395789888e-05,
"loss": 0.3398,
"step": 28345
},
{
"epoch": 3.3057369402985075,
"grad_norm": 0.5750150917239619,
"learning_rate": 1.7708591743370555e-05,
"loss": 0.3346,
"step": 28350
},
{
"epoch": 3.306319962686567,
"grad_norm": 0.4942291051775827,
"learning_rate": 1.7700780984730818e-05,
"loss": 0.3194,
"step": 28355
},
{
"epoch": 3.3069029850746268,
"grad_norm": 0.44461377262811785,
"learning_rate": 1.7692971683141063e-05,
"loss": 0.3116,
"step": 28360
},
{
"epoch": 3.3074860074626864,
"grad_norm": 0.40068238418777885,
"learning_rate": 1.7685163839762457e-05,
"loss": 0.3085,
"step": 28365
},
{
"epoch": 3.3080690298507465,
"grad_norm": 0.46351437324640127,
"learning_rate": 1.7677357455755954e-05,
"loss": 0.3155,
"step": 28370
},
{
"epoch": 3.308652052238806,
"grad_norm": 0.4289306287963423,
"learning_rate": 1.7669552532282287e-05,
"loss": 0.3295,
"step": 28375
},
{
"epoch": 3.3092350746268657,
"grad_norm": 0.46317987590943466,
"learning_rate": 1.766174907050196e-05,
"loss": 0.326,
"step": 28380
},
{
"epoch": 3.3098180970149254,
"grad_norm": 0.511695342506279,
"learning_rate": 1.765394707157529e-05,
"loss": 0.3434,
"step": 28385
},
{
"epoch": 3.310401119402985,
"grad_norm": 0.4556319825438499,
"learning_rate": 1.764614653666235e-05,
"loss": 0.3212,
"step": 28390
},
{
"epoch": 3.3109841417910446,
"grad_norm": 0.439370769368817,
"learning_rate": 1.763834746692301e-05,
"loss": 0.3132,
"step": 28395
},
{
"epoch": 3.3115671641791042,
"grad_norm": 0.5276460677109263,
"learning_rate": 1.7630549863516914e-05,
"loss": 0.3447,
"step": 28400
},
{
"epoch": 3.3121501865671643,
"grad_norm": 0.44358787596302135,
"learning_rate": 1.76227537276035e-05,
"loss": 0.3127,
"step": 28405
},
{
"epoch": 3.312733208955224,
"grad_norm": 0.5168235568497626,
"learning_rate": 1.7614959060341968e-05,
"loss": 0.3381,
"step": 28410
},
{
"epoch": 3.3133162313432836,
"grad_norm": 0.46956042215881105,
"learning_rate": 1.7607165862891302e-05,
"loss": 0.3257,
"step": 28415
},
{
"epoch": 3.313899253731343,
"grad_norm": 0.473856643686133,
"learning_rate": 1.75993741364103e-05,
"loss": 0.3171,
"step": 28420
},
{
"epoch": 3.314482276119403,
"grad_norm": 0.48769321188112036,
"learning_rate": 1.7591583882057495e-05,
"loss": 0.3339,
"step": 28425
},
{
"epoch": 3.315065298507463,
"grad_norm": 0.43452129837794445,
"learning_rate": 1.7583795100991246e-05,
"loss": 0.3246,
"step": 28430
},
{
"epoch": 3.3156483208955225,
"grad_norm": 0.42564160138718315,
"learning_rate": 1.7576007794369648e-05,
"loss": 0.3266,
"step": 28435
},
{
"epoch": 3.316231343283582,
"grad_norm": 0.44775557185309367,
"learning_rate": 1.7568221963350605e-05,
"loss": 0.31,
"step": 28440
},
{
"epoch": 3.316814365671642,
"grad_norm": 0.443918874549826,
"learning_rate": 1.75604376090918e-05,
"loss": 0.3422,
"step": 28445
},
{
"epoch": 3.3173973880597014,
"grad_norm": 0.48853054823756,
"learning_rate": 1.755265473275069e-05,
"loss": 0.3398,
"step": 28450
},
{
"epoch": 3.317980410447761,
"grad_norm": 0.4453288753925691,
"learning_rate": 1.7544873335484514e-05,
"loss": 0.3315,
"step": 28455
},
{
"epoch": 3.3185634328358207,
"grad_norm": 0.4706736696432995,
"learning_rate": 1.7537093418450294e-05,
"loss": 0.3323,
"step": 28460
},
{
"epoch": 3.3191464552238807,
"grad_norm": 0.4622062786366639,
"learning_rate": 1.7529314982804817e-05,
"loss": 0.32,
"step": 28465
},
{
"epoch": 3.3197294776119404,
"grad_norm": 0.5128681127573129,
"learning_rate": 1.7521538029704682e-05,
"loss": 0.3417,
"step": 28470
},
{
"epoch": 3.3203125,
"grad_norm": 0.42812012078225126,
"learning_rate": 1.7513762560306224e-05,
"loss": 0.3266,
"step": 28475
},
{
"epoch": 3.3208955223880596,
"grad_norm": 0.4179109474240003,
"learning_rate": 1.750598857576561e-05,
"loss": 0.332,
"step": 28480
},
{
"epoch": 3.3214785447761193,
"grad_norm": 0.43273395310547047,
"learning_rate": 1.749821607723873e-05,
"loss": 0.3304,
"step": 28485
},
{
"epoch": 3.3220615671641793,
"grad_norm": 0.45259230208095724,
"learning_rate": 1.749044506588129e-05,
"loss": 0.3101,
"step": 28490
},
{
"epoch": 3.322644589552239,
"grad_norm": 0.5092566706794512,
"learning_rate": 1.748267554284877e-05,
"loss": 0.3426,
"step": 28495
},
{
"epoch": 3.3232276119402986,
"grad_norm": 0.46642444532670313,
"learning_rate": 1.7474907509296412e-05,
"loss": 0.3219,
"step": 28500
},
{
"epoch": 3.323810634328358,
"grad_norm": 0.48434216570939764,
"learning_rate": 1.746714096637926e-05,
"loss": 0.3509,
"step": 28505
},
{
"epoch": 3.324393656716418,
"grad_norm": 0.4625398070916998,
"learning_rate": 1.7459375915252123e-05,
"loss": 0.3382,
"step": 28510
},
{
"epoch": 3.3249766791044775,
"grad_norm": 0.4731801618433754,
"learning_rate": 1.745161235706958e-05,
"loss": 0.3149,
"step": 28515
},
{
"epoch": 3.325559701492537,
"grad_norm": 0.5480336339026786,
"learning_rate": 1.7443850292986007e-05,
"loss": 0.321,
"step": 28520
},
{
"epoch": 3.326142723880597,
"grad_norm": 0.4731093715929599,
"learning_rate": 1.7436089724155545e-05,
"loss": 0.3352,
"step": 28525
},
{
"epoch": 3.326725746268657,
"grad_norm": 0.482348860518997,
"learning_rate": 1.742833065173212e-05,
"loss": 0.3195,
"step": 28530
},
{
"epoch": 3.3273087686567164,
"grad_norm": 0.512404518918994,
"learning_rate": 1.7420573076869422e-05,
"loss": 0.3419,
"step": 28535
},
{
"epoch": 3.327891791044776,
"grad_norm": 0.44950430326831653,
"learning_rate": 1.7412817000720937e-05,
"loss": 0.3461,
"step": 28540
},
{
"epoch": 3.3284748134328357,
"grad_norm": 0.4675428940933235,
"learning_rate": 1.7405062424439916e-05,
"loss": 0.3174,
"step": 28545
},
{
"epoch": 3.3290578358208958,
"grad_norm": 0.46836753540610626,
"learning_rate": 1.7397309349179393e-05,
"loss": 0.311,
"step": 28550
},
{
"epoch": 3.3296408582089554,
"grad_norm": 0.4612559364998148,
"learning_rate": 1.7389557776092182e-05,
"loss": 0.3388,
"step": 28555
},
{
"epoch": 3.330223880597015,
"grad_norm": 0.4710098687914657,
"learning_rate": 1.738180770633085e-05,
"loss": 0.3294,
"step": 28560
},
{
"epoch": 3.3308069029850746,
"grad_norm": 0.48319244955653473,
"learning_rate": 1.7374059141047772e-05,
"loss": 0.3349,
"step": 28565
},
{
"epoch": 3.3313899253731343,
"grad_norm": 0.4742826927232586,
"learning_rate": 1.7366312081395075e-05,
"loss": 0.3496,
"step": 28570
},
{
"epoch": 3.331972947761194,
"grad_norm": 0.4679596663176974,
"learning_rate": 1.735856652852468e-05,
"loss": 0.3378,
"step": 28575
},
{
"epoch": 3.3325559701492535,
"grad_norm": 0.4700004437056801,
"learning_rate": 1.7350822483588277e-05,
"loss": 0.3407,
"step": 28580
},
{
"epoch": 3.3331389925373136,
"grad_norm": 0.4661625701935188,
"learning_rate": 1.734307994773734e-05,
"loss": 0.3146,
"step": 28585
},
{
"epoch": 3.3337220149253732,
"grad_norm": 0.4274293603330983,
"learning_rate": 1.7335338922123076e-05,
"loss": 0.3395,
"step": 28590
},
{
"epoch": 3.334305037313433,
"grad_norm": 0.4357412023859455,
"learning_rate": 1.732759940789655e-05,
"loss": 0.3066,
"step": 28595
},
{
"epoch": 3.3348880597014925,
"grad_norm": 0.47209808031962,
"learning_rate": 1.7319861406208504e-05,
"loss": 0.3258,
"step": 28600
},
{
"epoch": 3.335471082089552,
"grad_norm": 0.4310846757244315,
"learning_rate": 1.7312124918209548e-05,
"loss": 0.337,
"step": 28605
},
{
"epoch": 3.3360541044776117,
"grad_norm": 0.45143116184202875,
"learning_rate": 1.7304389945050004e-05,
"loss": 0.3227,
"step": 28610
},
{
"epoch": 3.336637126865672,
"grad_norm": 0.4718825945233989,
"learning_rate": 1.729665648787998e-05,
"loss": 0.335,
"step": 28615
},
{
"epoch": 3.3372201492537314,
"grad_norm": 0.45236652474602207,
"learning_rate": 1.728892454784938e-05,
"loss": 0.3228,
"step": 28620
},
{
"epoch": 3.337803171641791,
"grad_norm": 0.4756684838279807,
"learning_rate": 1.7281194126107864e-05,
"loss": 0.337,
"step": 28625
},
{
"epoch": 3.3383861940298507,
"grad_norm": 0.4832103297157127,
"learning_rate": 1.7273465223804876e-05,
"loss": 0.3235,
"step": 28630
},
{
"epoch": 3.3389692164179103,
"grad_norm": 0.4620669106805667,
"learning_rate": 1.726573784208963e-05,
"loss": 0.3352,
"step": 28635
},
{
"epoch": 3.33955223880597,
"grad_norm": 0.457379511350836,
"learning_rate": 1.7258011982111094e-05,
"loss": 0.3165,
"step": 28640
},
{
"epoch": 3.34013526119403,
"grad_norm": 0.4837806776741993,
"learning_rate": 1.7250287645018067e-05,
"loss": 0.3398,
"step": 28645
},
{
"epoch": 3.3407182835820897,
"grad_norm": 0.47797831299703214,
"learning_rate": 1.7242564831959045e-05,
"loss": 0.3164,
"step": 28650
},
{
"epoch": 3.3413013059701493,
"grad_norm": 0.45061023593329286,
"learning_rate": 1.7234843544082368e-05,
"loss": 0.3114,
"step": 28655
},
{
"epoch": 3.341884328358209,
"grad_norm": 0.4699462388404545,
"learning_rate": 1.72271237825361e-05,
"loss": 0.3261,
"step": 28660
},
{
"epoch": 3.3424673507462686,
"grad_norm": 0.4408518335167113,
"learning_rate": 1.7219405548468098e-05,
"loss": 0.3109,
"step": 28665
},
{
"epoch": 3.343050373134328,
"grad_norm": 0.47090059737791345,
"learning_rate": 1.7211688843025987e-05,
"loss": 0.3419,
"step": 28670
},
{
"epoch": 3.3436333955223883,
"grad_norm": 0.4644265693422495,
"learning_rate": 1.7203973667357177e-05,
"loss": 0.3281,
"step": 28675
},
{
"epoch": 3.344216417910448,
"grad_norm": 0.4373562328813315,
"learning_rate": 1.7196260022608828e-05,
"loss": 0.3275,
"step": 28680
},
{
"epoch": 3.3447994402985075,
"grad_norm": 0.47346008251378013,
"learning_rate": 1.7188547909927903e-05,
"loss": 0.3282,
"step": 28685
},
{
"epoch": 3.345382462686567,
"grad_norm": 1.0564611428526927,
"learning_rate": 1.7180837330461093e-05,
"loss": 0.3364,
"step": 28690
},
{
"epoch": 3.3459654850746268,
"grad_norm": 0.4913014920273401,
"learning_rate": 1.7173128285354917e-05,
"loss": 0.3347,
"step": 28695
},
{
"epoch": 3.3465485074626864,
"grad_norm": 0.4357942703481503,
"learning_rate": 1.716542077575561e-05,
"loss": 0.3373,
"step": 28700
},
{
"epoch": 3.3471315298507465,
"grad_norm": 0.47395184014634933,
"learning_rate": 1.7157714802809223e-05,
"loss": 0.3308,
"step": 28705
},
{
"epoch": 3.347714552238806,
"grad_norm": 0.4802861614279903,
"learning_rate": 1.7150010367661546e-05,
"loss": 0.3464,
"step": 28710
},
{
"epoch": 3.3482975746268657,
"grad_norm": 0.47638191405072194,
"learning_rate": 1.7142307471458165e-05,
"loss": 0.3293,
"step": 28715
},
{
"epoch": 3.3488805970149254,
"grad_norm": 0.48583329962606997,
"learning_rate": 1.7134606115344427e-05,
"loss": 0.3377,
"step": 28720
},
{
"epoch": 3.349463619402985,
"grad_norm": 0.4708102851131365,
"learning_rate": 1.7126906300465446e-05,
"loss": 0.3311,
"step": 28725
},
{
"epoch": 3.3500466417910446,
"grad_norm": 0.44741603367765165,
"learning_rate": 1.7119208027966116e-05,
"loss": 0.3022,
"step": 28730
},
{
"epoch": 3.3506296641791042,
"grad_norm": 0.4369750820887205,
"learning_rate": 1.7111511298991088e-05,
"loss": 0.315,
"step": 28735
},
{
"epoch": 3.3512126865671643,
"grad_norm": 0.4635991699528975,
"learning_rate": 1.710381611468479e-05,
"loss": 0.3441,
"step": 28740
},
{
"epoch": 3.351795708955224,
"grad_norm": 0.4393989496610577,
"learning_rate": 1.7096122476191434e-05,
"loss": 0.3375,
"step": 28745
},
{
"epoch": 3.3523787313432836,
"grad_norm": 0.4575591715744827,
"learning_rate": 1.7088430384654984e-05,
"loss": 0.3183,
"step": 28750
},
{
"epoch": 3.352961753731343,
"grad_norm": 0.4837015506161303,
"learning_rate": 1.7080739841219175e-05,
"loss": 0.3005,
"step": 28755
},
{
"epoch": 3.353544776119403,
"grad_norm": 0.41873851176301535,
"learning_rate": 1.7073050847027537e-05,
"loss": 0.3254,
"step": 28760
},
{
"epoch": 3.354127798507463,
"grad_norm": 0.4385907189743988,
"learning_rate": 1.7065363403223316e-05,
"loss": 0.3221,
"step": 28765
},
{
"epoch": 3.3547108208955225,
"grad_norm": 0.4703976876800958,
"learning_rate": 1.7057677510949598e-05,
"loss": 0.3454,
"step": 28770
},
{
"epoch": 3.355293843283582,
"grad_norm": 0.44711954650808494,
"learning_rate": 1.704999317134917e-05,
"loss": 0.3384,
"step": 28775
},
{
"epoch": 3.355876865671642,
"grad_norm": 0.45746519877296554,
"learning_rate": 1.704231038556465e-05,
"loss": 0.323,
"step": 28780
},
{
"epoch": 3.3564598880597014,
"grad_norm": 0.48997222303227145,
"learning_rate": 1.7034629154738368e-05,
"loss": 0.3455,
"step": 28785
},
{
"epoch": 3.357042910447761,
"grad_norm": 0.44365154822478664,
"learning_rate": 1.702694948001246e-05,
"loss": 0.3289,
"step": 28790
},
{
"epoch": 3.3576259328358207,
"grad_norm": 0.4535111888588438,
"learning_rate": 1.7019271362528823e-05,
"loss": 0.3118,
"step": 28795
},
{
"epoch": 3.3582089552238807,
"grad_norm": 0.47714760336776346,
"learning_rate": 1.701159480342911e-05,
"loss": 0.3335,
"step": 28800
},
{
"epoch": 3.3587919776119404,
"grad_norm": 0.48959669410442797,
"learning_rate": 1.7003919803854765e-05,
"loss": 0.3325,
"step": 28805
},
{
"epoch": 3.359375,
"grad_norm": 0.46285775475973334,
"learning_rate": 1.6996246364946985e-05,
"loss": 0.3116,
"step": 28810
},
{
"epoch": 3.3599580223880596,
"grad_norm": 0.4169969152918247,
"learning_rate": 1.698857448784672e-05,
"loss": 0.3417,
"step": 28815
},
{
"epoch": 3.3605410447761193,
"grad_norm": 0.42493875863731995,
"learning_rate": 1.6980904173694727e-05,
"loss": 0.3176,
"step": 28820
},
{
"epoch": 3.3611240671641793,
"grad_norm": 0.479417795940034,
"learning_rate": 1.6973235423631488e-05,
"loss": 0.3339,
"step": 28825
},
{
"epoch": 3.361707089552239,
"grad_norm": 0.46063227863602507,
"learning_rate": 1.69655682387973e-05,
"loss": 0.3213,
"step": 28830
},
{
"epoch": 3.3622901119402986,
"grad_norm": 0.42695906558339464,
"learning_rate": 1.6957902620332163e-05,
"loss": 0.3249,
"step": 28835
},
{
"epoch": 3.362873134328358,
"grad_norm": 0.46454051173634997,
"learning_rate": 1.695023856937591e-05,
"loss": 0.3426,
"step": 28840
},
{
"epoch": 3.363456156716418,
"grad_norm": 0.42512367792774186,
"learning_rate": 1.6942576087068096e-05,
"loss": 0.3224,
"step": 28845
},
{
"epoch": 3.3640391791044775,
"grad_norm": 0.43984928949360896,
"learning_rate": 1.6934915174548073e-05,
"loss": 0.3272,
"step": 28850
},
{
"epoch": 3.364622201492537,
"grad_norm": 0.4761012354743501,
"learning_rate": 1.6927255832954934e-05,
"loss": 0.3332,
"step": 28855
},
{
"epoch": 3.365205223880597,
"grad_norm": 0.4349741218695636,
"learning_rate": 1.691959806342756e-05,
"loss": 0.3166,
"step": 28860
},
{
"epoch": 3.365788246268657,
"grad_norm": 0.4941038573767465,
"learning_rate": 1.691194186710457e-05,
"loss": 0.3307,
"step": 28865
},
{
"epoch": 3.3663712686567164,
"grad_norm": 0.46814148887806617,
"learning_rate": 1.690428724512439e-05,
"loss": 0.3354,
"step": 28870
},
{
"epoch": 3.366954291044776,
"grad_norm": 0.4927434354158718,
"learning_rate": 1.6896634198625167e-05,
"loss": 0.3062,
"step": 28875
},
{
"epoch": 3.3675373134328357,
"grad_norm": 0.4495077343515924,
"learning_rate": 1.688898272874485e-05,
"loss": 0.3162,
"step": 28880
},
{
"epoch": 3.3681203358208958,
"grad_norm": 0.42772372072909065,
"learning_rate": 1.688133283662113e-05,
"loss": 0.3207,
"step": 28885
},
{
"epoch": 3.3687033582089554,
"grad_norm": 0.506251918914207,
"learning_rate": 1.6873684523391487e-05,
"loss": 0.3232,
"step": 28890
},
{
"epoch": 3.369286380597015,
"grad_norm": 0.44161011006478523,
"learning_rate": 1.6866037790193135e-05,
"loss": 0.3217,
"step": 28895
},
{
"epoch": 3.3698694029850746,
"grad_norm": 0.47497485510111964,
"learning_rate": 1.685839263816308e-05,
"loss": 0.3391,
"step": 28900
},
{
"epoch": 3.3704524253731343,
"grad_norm": 0.4446775217842558,
"learning_rate": 1.6850749068438092e-05,
"loss": 0.3113,
"step": 28905
},
{
"epoch": 3.371035447761194,
"grad_norm": 0.4456686993412567,
"learning_rate": 1.6843107082154675e-05,
"loss": 0.3224,
"step": 28910
},
{
"epoch": 3.3716184701492535,
"grad_norm": 0.4640905670462758,
"learning_rate": 1.6835466680449128e-05,
"loss": 0.339,
"step": 28915
},
{
"epoch": 3.3722014925373136,
"grad_norm": 0.4449103955020881,
"learning_rate": 1.68278278644575e-05,
"loss": 0.3248,
"step": 28920
},
{
"epoch": 3.3727845149253732,
"grad_norm": 0.4272115502334145,
"learning_rate": 1.6820190635315627e-05,
"loss": 0.3202,
"step": 28925
},
{
"epoch": 3.373367537313433,
"grad_norm": 0.4984313639098382,
"learning_rate": 1.6812554994159073e-05,
"loss": 0.3489,
"step": 28930
},
{
"epoch": 3.3739505597014925,
"grad_norm": 0.45636220850360476,
"learning_rate": 1.6804920942123207e-05,
"loss": 0.3609,
"step": 28935
},
{
"epoch": 3.374533582089552,
"grad_norm": 0.5119129821437987,
"learning_rate": 1.679728848034311e-05,
"loss": 0.3361,
"step": 28940
},
{
"epoch": 3.3751166044776117,
"grad_norm": 0.4832296650669876,
"learning_rate": 1.6789657609953686e-05,
"loss": 0.2973,
"step": 28945
},
{
"epoch": 3.375699626865672,
"grad_norm": 0.49182622850313595,
"learning_rate": 1.678202833208954e-05,
"loss": 0.3598,
"step": 28950
},
{
"epoch": 3.3762826492537314,
"grad_norm": 0.5078875959243815,
"learning_rate": 1.6774400647885106e-05,
"loss": 0.3187,
"step": 28955
},
{
"epoch": 3.376865671641791,
"grad_norm": 0.48009172975848813,
"learning_rate": 1.6766774558474523e-05,
"loss": 0.3383,
"step": 28960
},
{
"epoch": 3.3774486940298507,
"grad_norm": 0.45320603960838435,
"learning_rate": 1.675915006499173e-05,
"loss": 0.3169,
"step": 28965
},
{
"epoch": 3.3780317164179103,
"grad_norm": 0.4372366280181476,
"learning_rate": 1.675152716857041e-05,
"loss": 0.3334,
"step": 28970
},
{
"epoch": 3.37861473880597,
"grad_norm": 0.5026424127544947,
"learning_rate": 1.6743905870344017e-05,
"loss": 0.32,
"step": 28975
},
{
"epoch": 3.37919776119403,
"grad_norm": 0.4433482644885729,
"learning_rate": 1.6736286171445763e-05,
"loss": 0.3242,
"step": 28980
},
{
"epoch": 3.3797807835820897,
"grad_norm": 0.4607096737178669,
"learning_rate": 1.672866807300864e-05,
"loss": 0.3153,
"step": 28985
},
{
"epoch": 3.3803638059701493,
"grad_norm": 0.4534037189966212,
"learning_rate": 1.672105157616535e-05,
"loss": 0.3156,
"step": 28990
},
{
"epoch": 3.380946828358209,
"grad_norm": 0.4754535574299298,
"learning_rate": 1.6713436682048436e-05,
"loss": 0.3397,
"step": 28995
},
{
"epoch": 3.3815298507462686,
"grad_norm": 0.48982348927044955,
"learning_rate": 1.670582339179012e-05,
"loss": 0.3392,
"step": 29000
},
{
"epoch": 3.382112873134328,
"grad_norm": 0.5006803940184319,
"learning_rate": 1.669821170652246e-05,
"loss": 0.3392,
"step": 29005
},
{
"epoch": 3.3826958955223883,
"grad_norm": 0.43469884603822234,
"learning_rate": 1.669060162737722e-05,
"loss": 0.3239,
"step": 29010
},
{
"epoch": 3.383278917910448,
"grad_norm": 0.43993247285998877,
"learning_rate": 1.6682993155485956e-05,
"loss": 0.3302,
"step": 29015
},
{
"epoch": 3.3838619402985075,
"grad_norm": 0.45745638126081706,
"learning_rate": 1.667538629197996e-05,
"loss": 0.325,
"step": 29020
},
{
"epoch": 3.384444962686567,
"grad_norm": 0.4904067030829676,
"learning_rate": 1.6667781037990317e-05,
"loss": 0.3493,
"step": 29025
},
{
"epoch": 3.3850279850746268,
"grad_norm": 0.47994232510089185,
"learning_rate": 1.666017739464784e-05,
"loss": 0.3294,
"step": 29030
},
{
"epoch": 3.3856110074626864,
"grad_norm": 0.43084287031412444,
"learning_rate": 1.6652575363083145e-05,
"loss": 0.3228,
"step": 29035
},
{
"epoch": 3.3861940298507465,
"grad_norm": 0.5108201858031536,
"learning_rate": 1.664497494442654e-05,
"loss": 0.3402,
"step": 29040
},
{
"epoch": 3.386777052238806,
"grad_norm": 0.4516662511890705,
"learning_rate": 1.6637376139808174e-05,
"loss": 0.3215,
"step": 29045
},
{
"epoch": 3.3873600746268657,
"grad_norm": 0.47598794694666274,
"learning_rate": 1.6629778950357883e-05,
"loss": 0.3293,
"step": 29050
},
{
"epoch": 3.3879430970149254,
"grad_norm": 0.4634828057975539,
"learning_rate": 1.6622183377205325e-05,
"loss": 0.3522,
"step": 29055
},
{
"epoch": 3.388526119402985,
"grad_norm": 0.4800220506569302,
"learning_rate": 1.6614589421479876e-05,
"loss": 0.3033,
"step": 29060
},
{
"epoch": 3.3891091417910446,
"grad_norm": 0.45987315002339646,
"learning_rate": 1.660699708431068e-05,
"loss": 0.3451,
"step": 29065
},
{
"epoch": 3.3896921641791042,
"grad_norm": 0.4840242205468901,
"learning_rate": 1.6599406366826648e-05,
"loss": 0.3189,
"step": 29070
},
{
"epoch": 3.3902751865671643,
"grad_norm": 0.4681427252517476,
"learning_rate": 1.659181727015645e-05,
"loss": 0.3242,
"step": 29075
},
{
"epoch": 3.390858208955224,
"grad_norm": 0.5206917501264354,
"learning_rate": 1.6584229795428514e-05,
"loss": 0.364,
"step": 29080
},
{
"epoch": 3.3914412313432836,
"grad_norm": 0.45132905670678686,
"learning_rate": 1.6576643943771026e-05,
"loss": 0.3198,
"step": 29085
},
{
"epoch": 3.392024253731343,
"grad_norm": 0.42703489141135753,
"learning_rate": 1.656905971631192e-05,
"loss": 0.3282,
"step": 29090
},
{
"epoch": 3.392607276119403,
"grad_norm": 0.4956592487889643,
"learning_rate": 1.65614771141789e-05,
"loss": 0.3318,
"step": 29095
},
{
"epoch": 3.393190298507463,
"grad_norm": 0.4597032552965804,
"learning_rate": 1.655389613849943e-05,
"loss": 0.3192,
"step": 29100
},
{
"epoch": 3.3937733208955225,
"grad_norm": 0.47238754573752373,
"learning_rate": 1.6546316790400726e-05,
"loss": 0.3238,
"step": 29105
},
{
"epoch": 3.394356343283582,
"grad_norm": 0.45524961260409447,
"learning_rate": 1.653873907100977e-05,
"loss": 0.3242,
"step": 29110
},
{
"epoch": 3.394939365671642,
"grad_norm": 0.44529409247332336,
"learning_rate": 1.6531162981453286e-05,
"loss": 0.3147,
"step": 29115
},
{
"epoch": 3.3955223880597014,
"grad_norm": 0.4393766798726157,
"learning_rate": 1.6523588522857784e-05,
"loss": 0.3321,
"step": 29120
},
{
"epoch": 3.396105410447761,
"grad_norm": 0.48351579617423796,
"learning_rate": 1.6516015696349486e-05,
"loss": 0.3219,
"step": 29125
},
{
"epoch": 3.3966884328358207,
"grad_norm": 0.48109320080614015,
"learning_rate": 1.6508444503054432e-05,
"loss": 0.3297,
"step": 29130
},
{
"epoch": 3.3972714552238807,
"grad_norm": 0.43121269326400713,
"learning_rate": 1.6500874944098348e-05,
"loss": 0.3341,
"step": 29135
},
{
"epoch": 3.3978544776119404,
"grad_norm": 0.45953798551611186,
"learning_rate": 1.6493307020606796e-05,
"loss": 0.3231,
"step": 29140
},
{
"epoch": 3.3984375,
"grad_norm": 0.4968957971702939,
"learning_rate": 1.648574073370502e-05,
"loss": 0.3204,
"step": 29145
},
{
"epoch": 3.3990205223880596,
"grad_norm": 0.484987188142252,
"learning_rate": 1.647817608451807e-05,
"loss": 0.3483,
"step": 29150
},
{
"epoch": 3.3996035447761193,
"grad_norm": 0.453128891702804,
"learning_rate": 1.647061307417073e-05,
"loss": 0.3305,
"step": 29155
},
{
"epoch": 3.4001865671641793,
"grad_norm": 0.46707430914885883,
"learning_rate": 1.6463051703787557e-05,
"loss": 0.3289,
"step": 29160
},
{
"epoch": 3.400769589552239,
"grad_norm": 0.442771230547862,
"learning_rate": 1.645549197449285e-05,
"loss": 0.3317,
"step": 29165
},
{
"epoch": 3.4013526119402986,
"grad_norm": 0.4518401278396903,
"learning_rate": 1.644793388741067e-05,
"loss": 0.3192,
"step": 29170
},
{
"epoch": 3.401935634328358,
"grad_norm": 0.46346999965867547,
"learning_rate": 1.6440377443664816e-05,
"loss": 0.3284,
"step": 29175
},
{
"epoch": 3.402518656716418,
"grad_norm": 0.48560352645991145,
"learning_rate": 1.6432822644378888e-05,
"loss": 0.3321,
"step": 29180
},
{
"epoch": 3.4031016791044775,
"grad_norm": 0.49327932470834185,
"learning_rate": 1.6425269490676176e-05,
"loss": 0.3271,
"step": 29185
},
{
"epoch": 3.403684701492537,
"grad_norm": 0.455797669188555,
"learning_rate": 1.64177179836798e-05,
"loss": 0.3284,
"step": 29190
},
{
"epoch": 3.404267723880597,
"grad_norm": 0.49251452493343373,
"learning_rate": 1.641016812451257e-05,
"loss": 0.3472,
"step": 29195
},
{
"epoch": 3.404850746268657,
"grad_norm": 0.4423612059084429,
"learning_rate": 1.6402619914297087e-05,
"loss": 0.3268,
"step": 29200
},
{
"epoch": 3.4054337686567164,
"grad_norm": 0.41882583544444013,
"learning_rate": 1.6395073354155694e-05,
"loss": 0.3192,
"step": 29205
},
{
"epoch": 3.406016791044776,
"grad_norm": 0.4694412268068585,
"learning_rate": 1.6387528445210497e-05,
"loss": 0.3452,
"step": 29210
},
{
"epoch": 3.4065998134328357,
"grad_norm": 0.4370163867794354,
"learning_rate": 1.6379985188583352e-05,
"loss": 0.3067,
"step": 29215
},
{
"epoch": 3.4071828358208958,
"grad_norm": 0.6320314077603403,
"learning_rate": 1.6372443585395875e-05,
"loss": 0.3453,
"step": 29220
},
{
"epoch": 3.4077658582089554,
"grad_norm": 0.4622433438894852,
"learning_rate": 1.6364903636769403e-05,
"loss": 0.3216,
"step": 29225
},
{
"epoch": 3.408348880597015,
"grad_norm": 0.4562735122226254,
"learning_rate": 1.6357365343825088e-05,
"loss": 0.3128,
"step": 29230
},
{
"epoch": 3.4089319029850746,
"grad_norm": 0.43711650765912896,
"learning_rate": 1.634982870768378e-05,
"loss": 0.3337,
"step": 29235
},
{
"epoch": 3.4095149253731343,
"grad_norm": 0.4934253764714628,
"learning_rate": 1.634229372946611e-05,
"loss": 0.3184,
"step": 29240
},
{
"epoch": 3.410097947761194,
"grad_norm": 0.47260380063181173,
"learning_rate": 1.633476041029246e-05,
"loss": 0.3434,
"step": 29245
},
{
"epoch": 3.4106809701492535,
"grad_norm": 0.45886172596055086,
"learning_rate": 1.632722875128296e-05,
"loss": 0.3261,
"step": 29250
},
{
"epoch": 3.4112639925373136,
"grad_norm": 0.49068242514814053,
"learning_rate": 1.6319698753557498e-05,
"loss": 0.3669,
"step": 29255
},
{
"epoch": 3.4118470149253732,
"grad_norm": 0.4826921491070021,
"learning_rate": 1.6312170418235705e-05,
"loss": 0.3275,
"step": 29260
},
{
"epoch": 3.412430037313433,
"grad_norm": 0.7730794447605902,
"learning_rate": 1.6304643746436994e-05,
"loss": 0.3234,
"step": 29265
},
{
"epoch": 3.4130130597014925,
"grad_norm": 0.4460789197147561,
"learning_rate": 1.6297118739280483e-05,
"loss": 0.3239,
"step": 29270
},
{
"epoch": 3.413596082089552,
"grad_norm": 0.47225575055167696,
"learning_rate": 1.6289595397885074e-05,
"loss": 0.3353,
"step": 29275
},
{
"epoch": 3.4141791044776117,
"grad_norm": 0.4689193918446093,
"learning_rate": 1.6282073723369427e-05,
"loss": 0.3117,
"step": 29280
},
{
"epoch": 3.414762126865672,
"grad_norm": 0.6572589456764506,
"learning_rate": 1.6274553716851928e-05,
"loss": 0.3172,
"step": 29285
},
{
"epoch": 3.4153451492537314,
"grad_norm": 0.48956681254627316,
"learning_rate": 1.6267035379450744e-05,
"loss": 0.3249,
"step": 29290
},
{
"epoch": 3.415928171641791,
"grad_norm": 0.4307089857003831,
"learning_rate": 1.625951871228378e-05,
"loss": 0.3179,
"step": 29295
},
{
"epoch": 3.4165111940298507,
"grad_norm": 0.4394328605862848,
"learning_rate": 1.625200371646867e-05,
"loss": 0.3266,
"step": 29300
},
{
"epoch": 3.4170942164179103,
"grad_norm": 0.5080366533717237,
"learning_rate": 1.6244490393122857e-05,
"loss": 0.3227,
"step": 29305
},
{
"epoch": 3.41767723880597,
"grad_norm": 0.472944969837989,
"learning_rate": 1.6236978743363464e-05,
"loss": 0.3309,
"step": 29310
},
{
"epoch": 3.41826026119403,
"grad_norm": 0.4679474789501229,
"learning_rate": 1.622946876830744e-05,
"loss": 0.3217,
"step": 29315
},
{
"epoch": 3.4188432835820897,
"grad_norm": 0.46585469354169445,
"learning_rate": 1.622196046907141e-05,
"loss": 0.3383,
"step": 29320
},
{
"epoch": 3.4194263059701493,
"grad_norm": 0.47036204047896363,
"learning_rate": 1.62144538467718e-05,
"loss": 0.3158,
"step": 29325
},
{
"epoch": 3.420009328358209,
"grad_norm": 0.4460527976475676,
"learning_rate": 1.6206948902524783e-05,
"loss": 0.3047,
"step": 29330
},
{
"epoch": 3.4205923507462686,
"grad_norm": 0.4662483884995518,
"learning_rate": 1.6199445637446255e-05,
"loss": 0.321,
"step": 29335
},
{
"epoch": 3.421175373134328,
"grad_norm": 0.4653190427968612,
"learning_rate": 1.619194405265189e-05,
"loss": 0.3214,
"step": 29340
},
{
"epoch": 3.4217583955223883,
"grad_norm": 0.4173920207221273,
"learning_rate": 1.618444414925711e-05,
"loss": 0.3136,
"step": 29345
},
{
"epoch": 3.422341417910448,
"grad_norm": 0.4654712782520686,
"learning_rate": 1.617694592837705e-05,
"loss": 0.3128,
"step": 29350
},
{
"epoch": 3.4229244402985075,
"grad_norm": 0.48019523489175087,
"learning_rate": 1.616944939112666e-05,
"loss": 0.3112,
"step": 29355
},
{
"epoch": 3.423507462686567,
"grad_norm": 0.4820369050056036,
"learning_rate": 1.616195453862057e-05,
"loss": 0.3253,
"step": 29360
},
{
"epoch": 3.4240904850746268,
"grad_norm": 0.4337344580595188,
"learning_rate": 1.615446137197322e-05,
"loss": 0.3208,
"step": 29365
},
{
"epoch": 3.4246735074626864,
"grad_norm": 0.4857873891669292,
"learning_rate": 1.614696989229876e-05,
"loss": 0.3318,
"step": 29370
},
{
"epoch": 3.4252565298507465,
"grad_norm": 0.44245351929052856,
"learning_rate": 1.613948010071109e-05,
"loss": 0.3149,
"step": 29375
},
{
"epoch": 3.425839552238806,
"grad_norm": 0.4523023080052427,
"learning_rate": 1.6131991998323893e-05,
"loss": 0.3151,
"step": 29380
},
{
"epoch": 3.4264225746268657,
"grad_norm": 0.46242880869989617,
"learning_rate": 1.612450558625056e-05,
"loss": 0.3332,
"step": 29385
},
{
"epoch": 3.4270055970149254,
"grad_norm": 0.4335355944581716,
"learning_rate": 1.611702086560426e-05,
"loss": 0.3205,
"step": 29390
},
{
"epoch": 3.427588619402985,
"grad_norm": 0.4714346715641154,
"learning_rate": 1.61095378374979e-05,
"loss": 0.3111,
"step": 29395
},
{
"epoch": 3.4281716417910446,
"grad_norm": 0.48579038471627284,
"learning_rate": 1.6102056503044115e-05,
"loss": 0.3342,
"step": 29400
},
{
"epoch": 3.4287546641791042,
"grad_norm": 0.4680627208635204,
"learning_rate": 1.6094576863355338e-05,
"loss": 0.347,
"step": 29405
},
{
"epoch": 3.4293376865671643,
"grad_norm": 0.4452049869579787,
"learning_rate": 1.6087098919543696e-05,
"loss": 0.3125,
"step": 29410
},
{
"epoch": 3.429920708955224,
"grad_norm": 0.4465235262099271,
"learning_rate": 1.6079622672721093e-05,
"loss": 0.3431,
"step": 29415
},
{
"epoch": 3.4305037313432836,
"grad_norm": 0.46229598497620156,
"learning_rate": 1.6072148123999182e-05,
"loss": 0.3264,
"step": 29420
},
{
"epoch": 3.431086753731343,
"grad_norm": 0.45152214672311974,
"learning_rate": 1.6064675274489348e-05,
"loss": 0.3378,
"step": 29425
},
{
"epoch": 3.431669776119403,
"grad_norm": 0.4256730359566115,
"learning_rate": 1.605720412530274e-05,
"loss": 0.3107,
"step": 29430
},
{
"epoch": 3.432252798507463,
"grad_norm": 0.47213526214375995,
"learning_rate": 1.6049734677550243e-05,
"loss": 0.3287,
"step": 29435
},
{
"epoch": 3.4328358208955225,
"grad_norm": 0.5274518664129215,
"learning_rate": 1.6042266932342498e-05,
"loss": 0.3304,
"step": 29440
},
{
"epoch": 3.433418843283582,
"grad_norm": 0.4606776918727309,
"learning_rate": 1.6034800890789874e-05,
"loss": 0.3319,
"step": 29445
},
{
"epoch": 3.434001865671642,
"grad_norm": 0.4743678258705525,
"learning_rate": 1.6027336554002512e-05,
"loss": 0.328,
"step": 29450
},
{
"epoch": 3.4345848880597014,
"grad_norm": 0.47545050039761877,
"learning_rate": 1.6019873923090277e-05,
"loss": 0.3182,
"step": 29455
},
{
"epoch": 3.435167910447761,
"grad_norm": 0.4523005954933371,
"learning_rate": 1.60124129991628e-05,
"loss": 0.3239,
"step": 29460
},
{
"epoch": 3.4357509328358207,
"grad_norm": 0.43291743277189226,
"learning_rate": 1.600495378332944e-05,
"loss": 0.3173,
"step": 29465
},
{
"epoch": 3.4363339552238807,
"grad_norm": 0.4127627299113383,
"learning_rate": 1.599749627669933e-05,
"loss": 0.3095,
"step": 29470
},
{
"epoch": 3.4369169776119404,
"grad_norm": 0.467592471972227,
"learning_rate": 1.59900404803813e-05,
"loss": 0.3324,
"step": 29475
},
{
"epoch": 3.4375,
"grad_norm": 0.44253230695768114,
"learning_rate": 1.5982586395483983e-05,
"loss": 0.3379,
"step": 29480
},
{
"epoch": 3.4380830223880596,
"grad_norm": 0.4901313291509628,
"learning_rate": 1.5975134023115707e-05,
"loss": 0.3136,
"step": 29485
},
{
"epoch": 3.4386660447761193,
"grad_norm": 0.46239650748248434,
"learning_rate": 1.5967683364384595e-05,
"loss": 0.327,
"step": 29490
},
{
"epoch": 3.4392490671641793,
"grad_norm": 0.43061478261902475,
"learning_rate": 1.5960234420398464e-05,
"loss": 0.3255,
"step": 29495
},
{
"epoch": 3.439832089552239,
"grad_norm": 0.4798479801501889,
"learning_rate": 1.595278719226491e-05,
"loss": 0.3261,
"step": 29500
},
{
"epoch": 3.4404151119402986,
"grad_norm": 0.49377906906887215,
"learning_rate": 1.5945341681091268e-05,
"loss": 0.3273,
"step": 29505
},
{
"epoch": 3.440998134328358,
"grad_norm": 0.42993794672775065,
"learning_rate": 1.5937897887984605e-05,
"loss": 0.3354,
"step": 29510
},
{
"epoch": 3.441581156716418,
"grad_norm": 0.469948013542681,
"learning_rate": 1.593045581405175e-05,
"loss": 0.3409,
"step": 29515
},
{
"epoch": 3.4421641791044775,
"grad_norm": 0.4391459720289155,
"learning_rate": 1.5923015460399277e-05,
"loss": 0.3304,
"step": 29520
},
{
"epoch": 3.442747201492537,
"grad_norm": 0.47788552398134604,
"learning_rate": 1.5915576828133462e-05,
"loss": 0.3184,
"step": 29525
},
{
"epoch": 3.443330223880597,
"grad_norm": 0.45281405309650546,
"learning_rate": 1.59081399183604e-05,
"loss": 0.3406,
"step": 29530
},
{
"epoch": 3.443913246268657,
"grad_norm": 0.49233220368932884,
"learning_rate": 1.590070473218585e-05,
"loss": 0.3513,
"step": 29535
},
{
"epoch": 3.4444962686567164,
"grad_norm": 0.4282769663966438,
"learning_rate": 1.589327127071539e-05,
"loss": 0.33,
"step": 29540
},
{
"epoch": 3.445079291044776,
"grad_norm": 0.4549963003710725,
"learning_rate": 1.5885839535054277e-05,
"loss": 0.3145,
"step": 29545
},
{
"epoch": 3.4456623134328357,
"grad_norm": 0.4561056686906441,
"learning_rate": 1.587840952630755e-05,
"loss": 0.3257,
"step": 29550
},
{
"epoch": 3.4462453358208958,
"grad_norm": 0.5272509953017616,
"learning_rate": 1.587098124557997e-05,
"loss": 0.3249,
"step": 29555
},
{
"epoch": 3.4468283582089554,
"grad_norm": 0.46858598486506753,
"learning_rate": 1.5863554693976065e-05,
"loss": 0.339,
"step": 29560
},
{
"epoch": 3.447411380597015,
"grad_norm": 0.47203929165066233,
"learning_rate": 1.5856129872600083e-05,
"loss": 0.3391,
"step": 29565
},
{
"epoch": 3.4479944029850746,
"grad_norm": 0.4713804789655109,
"learning_rate": 1.584870678255604e-05,
"loss": 0.3143,
"step": 29570
},
{
"epoch": 3.4485774253731343,
"grad_norm": 0.4645038934380423,
"learning_rate": 1.5841285424947643e-05,
"loss": 0.3246,
"step": 29575
},
{
"epoch": 3.449160447761194,
"grad_norm": 0.44099698390558617,
"learning_rate": 1.5833865800878422e-05,
"loss": 0.348,
"step": 29580
},
{
"epoch": 3.4497434701492535,
"grad_norm": 0.4629400844409978,
"learning_rate": 1.5826447911451568e-05,
"loss": 0.3366,
"step": 29585
},
{
"epoch": 3.4503264925373136,
"grad_norm": 0.44877295900291836,
"learning_rate": 1.5819031757770064e-05,
"loss": 0.3243,
"step": 29590
},
{
"epoch": 3.4509095149253732,
"grad_norm": 0.4761971918320478,
"learning_rate": 1.5811617340936622e-05,
"loss": 0.3347,
"step": 29595
},
{
"epoch": 3.451492537313433,
"grad_norm": 0.47429383320207846,
"learning_rate": 1.580420466205369e-05,
"loss": 0.3327,
"step": 29600
},
{
"epoch": 3.4520755597014925,
"grad_norm": 0.4325222611319617,
"learning_rate": 1.5796793722223464e-05,
"loss": 0.3084,
"step": 29605
},
{
"epoch": 3.452658582089552,
"grad_norm": 0.49804279155651243,
"learning_rate": 1.5789384522547888e-05,
"loss": 0.3529,
"step": 29610
},
{
"epoch": 3.4532416044776117,
"grad_norm": 0.4403294750127433,
"learning_rate": 1.5781977064128627e-05,
"loss": 0.3129,
"step": 29615
},
{
"epoch": 3.453824626865672,
"grad_norm": 0.466385874116567,
"learning_rate": 1.577457134806711e-05,
"loss": 0.3067,
"step": 29620
},
{
"epoch": 3.4544076492537314,
"grad_norm": 0.4554909670671076,
"learning_rate": 1.5767167375464488e-05,
"loss": 0.3243,
"step": 29625
},
{
"epoch": 3.454990671641791,
"grad_norm": 0.47142982851626164,
"learning_rate": 1.5759765147421658e-05,
"loss": 0.3299,
"step": 29630
},
{
"epoch": 3.4555736940298507,
"grad_norm": 0.4820811528926111,
"learning_rate": 1.5752364665039265e-05,
"loss": 0.3243,
"step": 29635
},
{
"epoch": 3.4561567164179103,
"grad_norm": 0.44116226153905386,
"learning_rate": 1.5744965929417693e-05,
"loss": 0.3171,
"step": 29640
},
{
"epoch": 3.45673973880597,
"grad_norm": 0.43959800142075794,
"learning_rate": 1.5737568941657053e-05,
"loss": 0.3121,
"step": 29645
},
{
"epoch": 3.45732276119403,
"grad_norm": 0.4270239335081458,
"learning_rate": 1.573017370285722e-05,
"loss": 0.3207,
"step": 29650
},
{
"epoch": 3.4579057835820897,
"grad_norm": 0.4715974742793349,
"learning_rate": 1.57227802141178e-05,
"loss": 0.3356,
"step": 29655
},
{
"epoch": 3.4584888059701493,
"grad_norm": 0.46583689868027905,
"learning_rate": 1.57153884765381e-05,
"loss": 0.3284,
"step": 29660
},
{
"epoch": 3.459071828358209,
"grad_norm": 0.45235961163983923,
"learning_rate": 1.570799849121724e-05,
"loss": 0.3299,
"step": 29665
},
{
"epoch": 3.4596548507462686,
"grad_norm": 0.45464411904765856,
"learning_rate": 1.5700610259254018e-05,
"loss": 0.3324,
"step": 29670
},
{
"epoch": 3.460237873134328,
"grad_norm": 0.4450421944133654,
"learning_rate": 1.5693223781746996e-05,
"loss": 0.3334,
"step": 29675
},
{
"epoch": 3.4608208955223883,
"grad_norm": 0.4594034531240603,
"learning_rate": 1.5685839059794476e-05,
"loss": 0.3229,
"step": 29680
},
{
"epoch": 3.461403917910448,
"grad_norm": 0.41895290725439016,
"learning_rate": 1.5678456094494493e-05,
"loss": 0.31,
"step": 29685
},
{
"epoch": 3.4619869402985075,
"grad_norm": 0.44935042825201343,
"learning_rate": 1.5671074886944823e-05,
"loss": 0.3257,
"step": 29690
},
{
"epoch": 3.462569962686567,
"grad_norm": 0.46514451343414803,
"learning_rate": 1.5663695438242992e-05,
"loss": 0.3295,
"step": 29695
},
{
"epoch": 3.4631529850746268,
"grad_norm": 0.4602897596220332,
"learning_rate": 1.5656317749486225e-05,
"loss": 0.325,
"step": 29700
},
{
"epoch": 3.4637360074626864,
"grad_norm": 0.47791320148340927,
"learning_rate": 1.5648941821771552e-05,
"loss": 0.3124,
"step": 29705
},
{
"epoch": 3.4643190298507465,
"grad_norm": 0.4429455185747638,
"learning_rate": 1.5641567656195664e-05,
"loss": 0.3463,
"step": 29710
},
{
"epoch": 3.464902052238806,
"grad_norm": 0.5332490658060334,
"learning_rate": 1.563419525385507e-05,
"loss": 0.3343,
"step": 29715
},
{
"epoch": 3.4654850746268657,
"grad_norm": 0.4872252237806146,
"learning_rate": 1.562682461584594e-05,
"loss": 0.3534,
"step": 29720
},
{
"epoch": 3.4660680970149254,
"grad_norm": 0.4136662662407533,
"learning_rate": 1.5619455743264227e-05,
"loss": 0.334,
"step": 29725
},
{
"epoch": 3.466651119402985,
"grad_norm": 0.4599090427951163,
"learning_rate": 1.561208863720562e-05,
"loss": 0.3183,
"step": 29730
},
{
"epoch": 3.4672341417910446,
"grad_norm": 0.4915351526363331,
"learning_rate": 1.5604723298765532e-05,
"loss": 0.3139,
"step": 29735
},
{
"epoch": 3.4678171641791042,
"grad_norm": 0.5549869449077802,
"learning_rate": 1.559735972903912e-05,
"loss": 0.3345,
"step": 29740
},
{
"epoch": 3.4684001865671643,
"grad_norm": 0.4140216835519108,
"learning_rate": 1.5589997929121282e-05,
"loss": 0.3321,
"step": 29745
},
{
"epoch": 3.468983208955224,
"grad_norm": 0.4644103030378211,
"learning_rate": 1.5582637900106622e-05,
"loss": 0.3338,
"step": 29750
},
{
"epoch": 3.4695662313432836,
"grad_norm": 0.4392730181361793,
"learning_rate": 1.557527964308954e-05,
"loss": 0.3445,
"step": 29755
},
{
"epoch": 3.470149253731343,
"grad_norm": 0.46176465643720466,
"learning_rate": 1.5567923159164108e-05,
"loss": 0.3208,
"step": 29760
},
{
"epoch": 3.470732276119403,
"grad_norm": 0.4878151437183506,
"learning_rate": 1.5560568449424197e-05,
"loss": 0.336,
"step": 29765
},
{
"epoch": 3.471315298507463,
"grad_norm": 0.5059817850100247,
"learning_rate": 1.555321551496335e-05,
"loss": 0.3379,
"step": 29770
},
{
"epoch": 3.4718983208955225,
"grad_norm": 0.4449441568580068,
"learning_rate": 1.5545864356874894e-05,
"loss": 0.3694,
"step": 29775
},
{
"epoch": 3.472481343283582,
"grad_norm": 0.4667909260521094,
"learning_rate": 1.553851497625187e-05,
"loss": 0.3264,
"step": 29780
},
{
"epoch": 3.473064365671642,
"grad_norm": 0.4646907193799085,
"learning_rate": 1.5531167374187068e-05,
"loss": 0.3064,
"step": 29785
},
{
"epoch": 3.4736473880597014,
"grad_norm": 0.4966638032944251,
"learning_rate": 1.5523821551773006e-05,
"loss": 0.3505,
"step": 29790
},
{
"epoch": 3.474230410447761,
"grad_norm": 0.46284592261560986,
"learning_rate": 1.5516477510101933e-05,
"loss": 0.3303,
"step": 29795
},
{
"epoch": 3.4748134328358207,
"grad_norm": 0.459641710190575,
"learning_rate": 1.5509135250265835e-05,
"loss": 0.3198,
"step": 29800
},
{
"epoch": 3.4753964552238807,
"grad_norm": 0.41530490147794286,
"learning_rate": 1.5501794773356437e-05,
"loss": 0.3179,
"step": 29805
},
{
"epoch": 3.4759794776119404,
"grad_norm": 0.4827602927154679,
"learning_rate": 1.5494456080465198e-05,
"loss": 0.3173,
"step": 29810
},
{
"epoch": 3.4765625,
"grad_norm": 0.45721814461824417,
"learning_rate": 1.548711917268331e-05,
"loss": 0.3362,
"step": 29815
},
{
"epoch": 3.4771455223880596,
"grad_norm": 0.4500721077058827,
"learning_rate": 1.547978405110171e-05,
"loss": 0.3157,
"step": 29820
},
{
"epoch": 3.4777285447761193,
"grad_norm": 0.43683248938066466,
"learning_rate": 1.5472450716811052e-05,
"loss": 0.3502,
"step": 29825
},
{
"epoch": 3.4783115671641793,
"grad_norm": 0.44719437684965135,
"learning_rate": 1.5465119170901742e-05,
"loss": 0.3132,
"step": 29830
},
{
"epoch": 3.478894589552239,
"grad_norm": 0.468349601267782,
"learning_rate": 1.5457789414463886e-05,
"loss": 0.3233,
"step": 29835
},
{
"epoch": 3.4794776119402986,
"grad_norm": 0.44337347085310647,
"learning_rate": 1.545046144858738e-05,
"loss": 0.3499,
"step": 29840
},
{
"epoch": 3.480060634328358,
"grad_norm": 0.4269928071599079,
"learning_rate": 1.5443135274361792e-05,
"loss": 0.3156,
"step": 29845
},
{
"epoch": 3.480643656716418,
"grad_norm": 0.5329365590231341,
"learning_rate": 1.543581089287649e-05,
"loss": 0.3273,
"step": 29850
},
{
"epoch": 3.4812266791044775,
"grad_norm": 0.4883476737585097,
"learning_rate": 1.5428488305220506e-05,
"loss": 0.3291,
"step": 29855
},
{
"epoch": 3.481809701492537,
"grad_norm": 0.4440475681494304,
"learning_rate": 1.5421167512482655e-05,
"loss": 0.3462,
"step": 29860
},
{
"epoch": 3.482392723880597,
"grad_norm": 0.5066557853889835,
"learning_rate": 1.5413848515751463e-05,
"loss": 0.3184,
"step": 29865
},
{
"epoch": 3.482975746268657,
"grad_norm": 0.45645957901086526,
"learning_rate": 1.5406531316115197e-05,
"loss": 0.3312,
"step": 29870
},
{
"epoch": 3.4835587686567164,
"grad_norm": 0.421373702662619,
"learning_rate": 1.539921591466185e-05,
"loss": 0.3092,
"step": 29875
},
{
"epoch": 3.484141791044776,
"grad_norm": 0.4782413018340401,
"learning_rate": 1.539190231247917e-05,
"loss": 0.3322,
"step": 29880
},
{
"epoch": 3.4847248134328357,
"grad_norm": 0.43459425670708246,
"learning_rate": 1.5384590510654596e-05,
"loss": 0.3197,
"step": 29885
},
{
"epoch": 3.4853078358208958,
"grad_norm": 0.45425958228353447,
"learning_rate": 1.5377280510275342e-05,
"loss": 0.3108,
"step": 29890
},
{
"epoch": 3.4858908582089554,
"grad_norm": 0.4905339271792473,
"learning_rate": 1.536997231242831e-05,
"loss": 0.3283,
"step": 29895
},
{
"epoch": 3.486473880597015,
"grad_norm": 0.4527586055954125,
"learning_rate": 1.5362665918200193e-05,
"loss": 0.3278,
"step": 29900
},
{
"epoch": 3.4870569029850746,
"grad_norm": 0.4509179327793743,
"learning_rate": 1.5355361328677358e-05,
"loss": 0.3379,
"step": 29905
},
{
"epoch": 3.4876399253731343,
"grad_norm": 0.4625325556708729,
"learning_rate": 1.534805854494593e-05,
"loss": 0.3269,
"step": 29910
},
{
"epoch": 3.488222947761194,
"grad_norm": 0.4589192547153669,
"learning_rate": 1.5340757568091767e-05,
"loss": 0.3247,
"step": 29915
},
{
"epoch": 3.4888059701492535,
"grad_norm": 0.461786600897186,
"learning_rate": 1.533345839920045e-05,
"loss": 0.3238,
"step": 29920
},
{
"epoch": 3.4893889925373136,
"grad_norm": 0.4643999625352688,
"learning_rate": 1.5326161039357307e-05,
"loss": 0.3206,
"step": 29925
},
{
"epoch": 3.4899720149253732,
"grad_norm": 0.4475201148208013,
"learning_rate": 1.5318865489647383e-05,
"loss": 0.3264,
"step": 29930
},
{
"epoch": 3.490555037313433,
"grad_norm": 0.43609363372306625,
"learning_rate": 1.531157175115543e-05,
"loss": 0.3342,
"step": 29935
},
{
"epoch": 3.4911380597014925,
"grad_norm": 0.4828681148141982,
"learning_rate": 1.5304279824966e-05,
"loss": 0.3253,
"step": 29940
},
{
"epoch": 3.491721082089552,
"grad_norm": 0.47411229592868015,
"learning_rate": 1.5296989712163302e-05,
"loss": 0.3472,
"step": 29945
},
{
"epoch": 3.4923041044776117,
"grad_norm": 0.473296964249336,
"learning_rate": 1.528970141383131e-05,
"loss": 0.3473,
"step": 29950
},
{
"epoch": 3.492887126865672,
"grad_norm": 0.442274355032023,
"learning_rate": 1.528241493105373e-05,
"loss": 0.3314,
"step": 29955
},
{
"epoch": 3.4934701492537314,
"grad_norm": 0.4745079198954923,
"learning_rate": 1.5275130264913994e-05,
"loss": 0.3317,
"step": 29960
},
{
"epoch": 3.494053171641791,
"grad_norm": 0.4150336984853833,
"learning_rate": 1.5267847416495257e-05,
"loss": 0.3329,
"step": 29965
},
{
"epoch": 3.4946361940298507,
"grad_norm": 0.4879438571232143,
"learning_rate": 1.5260566386880413e-05,
"loss": 0.3257,
"step": 29970
},
{
"epoch": 3.4952192164179103,
"grad_norm": 0.46425872436194615,
"learning_rate": 1.5253287177152082e-05,
"loss": 0.3226,
"step": 29975
},
{
"epoch": 3.49580223880597,
"grad_norm": 0.46554802973905623,
"learning_rate": 1.5246009788392606e-05,
"loss": 0.3133,
"step": 29980
},
{
"epoch": 3.49638526119403,
"grad_norm": 0.44823348537045765,
"learning_rate": 1.5238734221684065e-05,
"loss": 0.3392,
"step": 29985
},
{
"epoch": 3.4969682835820897,
"grad_norm": 0.5319361981527446,
"learning_rate": 1.5231460478108268e-05,
"loss": 0.3351,
"step": 29990
},
{
"epoch": 3.4975513059701493,
"grad_norm": 0.4458190794231886,
"learning_rate": 1.5224188558746749e-05,
"loss": 0.3111,
"step": 29995
},
{
"epoch": 3.498134328358209,
"grad_norm": 0.4677344169174092,
"learning_rate": 1.5216918464680776e-05,
"loss": 0.3488,
"step": 30000
},
{
"epoch": 3.4987173507462686,
"grad_norm": 0.47493244189895345,
"learning_rate": 1.5209650196991352e-05,
"loss": 0.3187,
"step": 30005
},
{
"epoch": 3.499300373134328,
"grad_norm": 0.427928725696231,
"learning_rate": 1.520238375675917e-05,
"loss": 0.3145,
"step": 30010
},
{
"epoch": 3.4998833955223883,
"grad_norm": 0.479999978785708,
"learning_rate": 1.5195119145064712e-05,
"loss": 0.3459,
"step": 30015
},
{
"epoch": 3.500466417910448,
"grad_norm": 0.5049413499590955,
"learning_rate": 1.5187856362988123e-05,
"loss": 0.334,
"step": 30020
},
{
"epoch": 3.5010494402985075,
"grad_norm": 0.4606207617182415,
"learning_rate": 1.5180595411609347e-05,
"loss": 0.3508,
"step": 30025
},
{
"epoch": 3.501632462686567,
"grad_norm": 0.4245892946045796,
"learning_rate": 1.5173336292007994e-05,
"loss": 0.3256,
"step": 30030
},
{
"epoch": 3.5022154850746268,
"grad_norm": 0.4788508571711781,
"learning_rate": 1.5166079005263425e-05,
"loss": 0.3323,
"step": 30035
},
{
"epoch": 3.5027985074626864,
"grad_norm": 0.453151954371947,
"learning_rate": 1.5158823552454737e-05,
"loss": 0.3138,
"step": 30040
},
{
"epoch": 3.5033815298507465,
"grad_norm": 0.46792014684429817,
"learning_rate": 1.5151569934660743e-05,
"loss": 0.3496,
"step": 30045
},
{
"epoch": 3.503964552238806,
"grad_norm": 0.452465615558095,
"learning_rate": 1.5144318152959985e-05,
"loss": 0.3266,
"step": 30050
},
{
"epoch": 3.5045475746268657,
"grad_norm": 0.4647037726873906,
"learning_rate": 1.513706820843075e-05,
"loss": 0.313,
"step": 30055
},
{
"epoch": 3.5051305970149254,
"grad_norm": 0.45918369004542436,
"learning_rate": 1.5129820102151e-05,
"loss": 0.335,
"step": 30060
},
{
"epoch": 3.505713619402985,
"grad_norm": 0.46661663913134543,
"learning_rate": 1.5122573835198495e-05,
"loss": 0.3286,
"step": 30065
},
{
"epoch": 3.5062966417910446,
"grad_norm": 0.4441609496785673,
"learning_rate": 1.5115329408650658e-05,
"loss": 0.3148,
"step": 30070
},
{
"epoch": 3.5068796641791042,
"grad_norm": 0.4603939051198401,
"learning_rate": 1.5108086823584692e-05,
"loss": 0.3215,
"step": 30075
},
{
"epoch": 3.5074626865671643,
"grad_norm": 0.4531797376247531,
"learning_rate": 1.5100846081077479e-05,
"loss": 0.3264,
"step": 30080
},
{
"epoch": 3.508045708955224,
"grad_norm": 0.4564771523303664,
"learning_rate": 1.5093607182205652e-05,
"loss": 0.3194,
"step": 30085
},
{
"epoch": 3.5086287313432836,
"grad_norm": 0.4714967782602235,
"learning_rate": 1.5086370128045574e-05,
"loss": 0.3348,
"step": 30090
},
{
"epoch": 3.509211753731343,
"grad_norm": 0.4945879123232498,
"learning_rate": 1.5079134919673316e-05,
"loss": 0.3488,
"step": 30095
},
{
"epoch": 3.509794776119403,
"grad_norm": 0.4529553874663103,
"learning_rate": 1.5071901558164692e-05,
"loss": 0.3291,
"step": 30100
},
{
"epoch": 3.510377798507463,
"grad_norm": 0.4340529939580081,
"learning_rate": 1.5064670044595236e-05,
"loss": 0.3111,
"step": 30105
},
{
"epoch": 3.5109608208955225,
"grad_norm": 0.4640821458973144,
"learning_rate": 1.5057440380040184e-05,
"loss": 0.3219,
"step": 30110
},
{
"epoch": 3.511543843283582,
"grad_norm": 0.4273474161407204,
"learning_rate": 1.5050212565574551e-05,
"loss": 0.3287,
"step": 30115
},
{
"epoch": 3.512126865671642,
"grad_norm": 0.4766857444880997,
"learning_rate": 1.5042986602273017e-05,
"loss": 0.3351,
"step": 30120
},
{
"epoch": 3.5127098880597014,
"grad_norm": 0.5169461119406662,
"learning_rate": 1.503576249121002e-05,
"loss": 0.3506,
"step": 30125
},
{
"epoch": 3.513292910447761,
"grad_norm": 0.45750200079140124,
"learning_rate": 1.502854023345972e-05,
"loss": 0.3167,
"step": 30130
},
{
"epoch": 3.5138759328358207,
"grad_norm": 2.7689476172114067,
"learning_rate": 1.5021319830095995e-05,
"loss": 0.3211,
"step": 30135
},
{
"epoch": 3.5144589552238807,
"grad_norm": 0.45714542015338194,
"learning_rate": 1.5014101282192452e-05,
"loss": 0.3313,
"step": 30140
},
{
"epoch": 3.5150419776119404,
"grad_norm": 0.5064495487185795,
"learning_rate": 1.5006884590822418e-05,
"loss": 0.3443,
"step": 30145
},
{
"epoch": 3.515625,
"grad_norm": 0.4677117280305794,
"learning_rate": 1.4999669757058956e-05,
"loss": 0.3397,
"step": 30150
},
{
"epoch": 3.5162080223880596,
"grad_norm": 0.4425881887654731,
"learning_rate": 1.4992456781974826e-05,
"loss": 0.3244,
"step": 30155
},
{
"epoch": 3.5167910447761193,
"grad_norm": 0.44083925066694307,
"learning_rate": 1.498524566664253e-05,
"loss": 0.3022,
"step": 30160
},
{
"epoch": 3.5173740671641793,
"grad_norm": 0.4348221106452468,
"learning_rate": 1.4978036412134305e-05,
"loss": 0.3084,
"step": 30165
},
{
"epoch": 3.517957089552239,
"grad_norm": 0.4275335750701321,
"learning_rate": 1.4970829019522083e-05,
"loss": 0.3135,
"step": 30170
},
{
"epoch": 3.5185401119402986,
"grad_norm": 0.4693869970062322,
"learning_rate": 1.4963623489877546e-05,
"loss": 0.3225,
"step": 30175
},
{
"epoch": 3.519123134328358,
"grad_norm": 0.46492413750613176,
"learning_rate": 1.4956419824272083e-05,
"loss": 0.3391,
"step": 30180
},
{
"epoch": 3.519706156716418,
"grad_norm": 0.48108138015879837,
"learning_rate": 1.4949218023776806e-05,
"loss": 0.3229,
"step": 30185
},
{
"epoch": 3.5202891791044775,
"grad_norm": 0.48711724947857066,
"learning_rate": 1.4942018089462567e-05,
"loss": 0.3317,
"step": 30190
},
{
"epoch": 3.520872201492537,
"grad_norm": 0.4570418012517138,
"learning_rate": 1.4934820022399901e-05,
"loss": 0.3409,
"step": 30195
},
{
"epoch": 3.521455223880597,
"grad_norm": 0.469818870151866,
"learning_rate": 1.4927623823659126e-05,
"loss": 0.3076,
"step": 30200
},
{
"epoch": 3.522038246268657,
"grad_norm": 0.4428288836390382,
"learning_rate": 1.4920429494310223e-05,
"loss": 0.3203,
"step": 30205
},
{
"epoch": 3.5226212686567164,
"grad_norm": 0.4896116815348177,
"learning_rate": 1.4913237035422926e-05,
"loss": 0.3407,
"step": 30210
},
{
"epoch": 3.523204291044776,
"grad_norm": 0.44644334194840807,
"learning_rate": 1.4906046448066685e-05,
"loss": 0.3107,
"step": 30215
},
{
"epoch": 3.5237873134328357,
"grad_norm": 0.44030758438026024,
"learning_rate": 1.4898857733310673e-05,
"loss": 0.3453,
"step": 30220
},
{
"epoch": 3.5243703358208958,
"grad_norm": 0.45797676287323474,
"learning_rate": 1.4891670892223781e-05,
"loss": 0.3245,
"step": 30225
},
{
"epoch": 3.5249533582089554,
"grad_norm": 0.49234730565533935,
"learning_rate": 1.4884485925874634e-05,
"loss": 0.3398,
"step": 30230
},
{
"epoch": 3.525536380597015,
"grad_norm": 0.47360282632648765,
"learning_rate": 1.487730283533154e-05,
"loss": 0.3264,
"step": 30235
},
{
"epoch": 3.5261194029850746,
"grad_norm": 0.615572611208453,
"learning_rate": 1.4870121621662594e-05,
"loss": 0.3528,
"step": 30240
},
{
"epoch": 3.5267024253731343,
"grad_norm": 0.4720230725004802,
"learning_rate": 1.4862942285935536e-05,
"loss": 0.3225,
"step": 30245
},
{
"epoch": 3.527285447761194,
"grad_norm": 0.48480881634838957,
"learning_rate": 1.4855764829217894e-05,
"loss": 0.3283,
"step": 30250
},
{
"epoch": 3.5278684701492535,
"grad_norm": 0.46537338033193987,
"learning_rate": 1.4848589252576872e-05,
"loss": 0.3295,
"step": 30255
},
{
"epoch": 3.5284514925373136,
"grad_norm": 0.4932477814602127,
"learning_rate": 1.4841415557079413e-05,
"loss": 0.3148,
"step": 30260
},
{
"epoch": 3.5290345149253732,
"grad_norm": 0.48945859058918917,
"learning_rate": 1.4834243743792176e-05,
"loss": 0.3484,
"step": 30265
},
{
"epoch": 3.529617537313433,
"grad_norm": 0.4515938381967078,
"learning_rate": 1.482707381378154e-05,
"loss": 0.3408,
"step": 30270
},
{
"epoch": 3.5302005597014925,
"grad_norm": 0.44603366316159504,
"learning_rate": 1.4819905768113604e-05,
"loss": 0.3345,
"step": 30275
},
{
"epoch": 3.530783582089552,
"grad_norm": 0.4716672497914246,
"learning_rate": 1.4812739607854199e-05,
"loss": 0.3315,
"step": 30280
},
{
"epoch": 3.531366604477612,
"grad_norm": 0.456842676782559,
"learning_rate": 1.4805575334068842e-05,
"loss": 0.3132,
"step": 30285
},
{
"epoch": 3.5319496268656714,
"grad_norm": 0.4483232566313984,
"learning_rate": 1.479841294782282e-05,
"loss": 0.3102,
"step": 30290
},
{
"epoch": 3.5325326492537314,
"grad_norm": 0.43596943271537114,
"learning_rate": 1.4791252450181081e-05,
"loss": 0.3397,
"step": 30295
},
{
"epoch": 3.533115671641791,
"grad_norm": 0.4586826391998859,
"learning_rate": 1.4784093842208351e-05,
"loss": 0.315,
"step": 30300
},
{
"epoch": 3.5336986940298507,
"grad_norm": 0.4916541342786979,
"learning_rate": 1.4776937124969026e-05,
"loss": 0.3592,
"step": 30305
},
{
"epoch": 3.5342817164179103,
"grad_norm": 0.46179133577272885,
"learning_rate": 1.4769782299527252e-05,
"loss": 0.3102,
"step": 30310
},
{
"epoch": 3.53486473880597,
"grad_norm": 0.4340595619890763,
"learning_rate": 1.4762629366946876e-05,
"loss": 0.3209,
"step": 30315
},
{
"epoch": 3.53544776119403,
"grad_norm": 0.4743104374022143,
"learning_rate": 1.4755478328291476e-05,
"loss": 0.3168,
"step": 30320
},
{
"epoch": 3.5360307835820897,
"grad_norm": 0.4791837885523034,
"learning_rate": 1.4748329184624341e-05,
"loss": 0.3032,
"step": 30325
},
{
"epoch": 3.5366138059701493,
"grad_norm": 0.4803759644825129,
"learning_rate": 1.4741181937008485e-05,
"loss": 0.3191,
"step": 30330
},
{
"epoch": 3.537196828358209,
"grad_norm": 0.45903502528960044,
"learning_rate": 1.4734036586506628e-05,
"loss": 0.3203,
"step": 30335
},
{
"epoch": 3.5377798507462686,
"grad_norm": 0.48116858842136384,
"learning_rate": 1.4726893134181214e-05,
"loss": 0.3293,
"step": 30340
},
{
"epoch": 3.5383628731343286,
"grad_norm": 0.44050830677185454,
"learning_rate": 1.4719751581094409e-05,
"loss": 0.3323,
"step": 30345
},
{
"epoch": 3.538945895522388,
"grad_norm": 0.4587786175505095,
"learning_rate": 1.4712611928308095e-05,
"loss": 0.3318,
"step": 30350
},
{
"epoch": 3.539528917910448,
"grad_norm": 0.5043457729237092,
"learning_rate": 1.4705474176883865e-05,
"loss": 0.3353,
"step": 30355
},
{
"epoch": 3.5401119402985075,
"grad_norm": 0.4629321792633427,
"learning_rate": 1.4698338327883044e-05,
"loss": 0.3162,
"step": 30360
},
{
"epoch": 3.540694962686567,
"grad_norm": 0.4791692020631821,
"learning_rate": 1.4691204382366666e-05,
"loss": 0.3339,
"step": 30365
},
{
"epoch": 3.5412779850746268,
"grad_norm": 0.4579125183709961,
"learning_rate": 1.4684072341395454e-05,
"loss": 0.3167,
"step": 30370
},
{
"epoch": 3.5418610074626864,
"grad_norm": 0.42304258163289626,
"learning_rate": 1.4676942206029909e-05,
"loss": 0.3082,
"step": 30375
},
{
"epoch": 3.5424440298507465,
"grad_norm": 0.4724063661286411,
"learning_rate": 1.4669813977330193e-05,
"loss": 0.3227,
"step": 30380
},
{
"epoch": 3.543027052238806,
"grad_norm": 0.4686993826196236,
"learning_rate": 1.466268765635621e-05,
"loss": 0.3192,
"step": 30385
},
{
"epoch": 3.5436100746268657,
"grad_norm": 0.453485392281836,
"learning_rate": 1.4655563244167572e-05,
"loss": 0.3116,
"step": 30390
},
{
"epoch": 3.5441930970149254,
"grad_norm": 0.4519716041384925,
"learning_rate": 1.4648440741823616e-05,
"loss": 0.3285,
"step": 30395
},
{
"epoch": 3.544776119402985,
"grad_norm": 0.4876679115355501,
"learning_rate": 1.4641320150383391e-05,
"loss": 0.3209,
"step": 30400
},
{
"epoch": 3.5453591417910446,
"grad_norm": 0.46832032629958203,
"learning_rate": 1.4634201470905665e-05,
"loss": 0.3363,
"step": 30405
},
{
"epoch": 3.5459421641791042,
"grad_norm": 0.4133999250795664,
"learning_rate": 1.4627084704448895e-05,
"loss": 0.3084,
"step": 30410
},
{
"epoch": 3.5465251865671643,
"grad_norm": 0.5371970916230431,
"learning_rate": 1.4619969852071305e-05,
"loss": 0.3281,
"step": 30415
},
{
"epoch": 3.547108208955224,
"grad_norm": 0.490413599502768,
"learning_rate": 1.461285691483078e-05,
"loss": 0.3253,
"step": 30420
},
{
"epoch": 3.5476912313432836,
"grad_norm": 0.5005322929727773,
"learning_rate": 1.4605745893784973e-05,
"loss": 0.3389,
"step": 30425
},
{
"epoch": 3.548274253731343,
"grad_norm": 0.4313245945801489,
"learning_rate": 1.4598636789991199e-05,
"loss": 0.328,
"step": 30430
},
{
"epoch": 3.548857276119403,
"grad_norm": 0.4674055972639654,
"learning_rate": 1.4591529604506527e-05,
"loss": 0.3295,
"step": 30435
},
{
"epoch": 3.549440298507463,
"grad_norm": 0.4098801845115704,
"learning_rate": 1.458442433838772e-05,
"loss": 0.3314,
"step": 30440
},
{
"epoch": 3.5500233208955225,
"grad_norm": 0.4410726552405741,
"learning_rate": 1.4577320992691269e-05,
"loss": 0.3223,
"step": 30445
},
{
"epoch": 3.550606343283582,
"grad_norm": 0.4652104596232166,
"learning_rate": 1.4570219568473376e-05,
"loss": 0.3189,
"step": 30450
},
{
"epoch": 3.551189365671642,
"grad_norm": 0.45065671865836054,
"learning_rate": 1.4563120066789953e-05,
"loss": 0.3268,
"step": 30455
},
{
"epoch": 3.5517723880597014,
"grad_norm": 0.47720768682642084,
"learning_rate": 1.4556022488696614e-05,
"loss": 0.3201,
"step": 30460
},
{
"epoch": 3.552355410447761,
"grad_norm": 0.46531219747261804,
"learning_rate": 1.4548926835248722e-05,
"loss": 0.3165,
"step": 30465
},
{
"epoch": 3.5529384328358207,
"grad_norm": 0.4644734524214457,
"learning_rate": 1.4541833107501312e-05,
"loss": 0.3441,
"step": 30470
},
{
"epoch": 3.5535214552238807,
"grad_norm": 0.5028933687869508,
"learning_rate": 1.453474130650918e-05,
"loss": 0.3354,
"step": 30475
},
{
"epoch": 3.5541044776119404,
"grad_norm": 0.44612356750149024,
"learning_rate": 1.4527651433326786e-05,
"loss": 0.3283,
"step": 30480
},
{
"epoch": 3.5546875,
"grad_norm": 0.4747894134423976,
"learning_rate": 1.4520563489008333e-05,
"loss": 0.3125,
"step": 30485
},
{
"epoch": 3.5552705223880596,
"grad_norm": 0.48242015364768864,
"learning_rate": 1.4513477474607729e-05,
"loss": 0.3382,
"step": 30490
},
{
"epoch": 3.5558535447761193,
"grad_norm": 0.459976643763842,
"learning_rate": 1.4506393391178597e-05,
"loss": 0.3309,
"step": 30495
},
{
"epoch": 3.5564365671641793,
"grad_norm": 0.4353100122061999,
"learning_rate": 1.4499311239774277e-05,
"loss": 0.3443,
"step": 30500
},
{
"epoch": 3.557019589552239,
"grad_norm": 0.4668148688982469,
"learning_rate": 1.449223102144781e-05,
"loss": 0.3356,
"step": 30505
},
{
"epoch": 3.5576026119402986,
"grad_norm": 0.42776316532602726,
"learning_rate": 1.4485152737251972e-05,
"loss": 0.3319,
"step": 30510
},
{
"epoch": 3.558185634328358,
"grad_norm": 0.532413465286466,
"learning_rate": 1.4478076388239214e-05,
"loss": 0.3419,
"step": 30515
},
{
"epoch": 3.558768656716418,
"grad_norm": 0.4452158253276118,
"learning_rate": 1.4471001975461735e-05,
"loss": 0.34,
"step": 30520
},
{
"epoch": 3.5593516791044775,
"grad_norm": 0.45648202388103803,
"learning_rate": 1.4463929499971424e-05,
"loss": 0.3331,
"step": 30525
},
{
"epoch": 3.559934701492537,
"grad_norm": 0.45474354594715216,
"learning_rate": 1.4456858962819897e-05,
"loss": 0.3307,
"step": 30530
},
{
"epoch": 3.560517723880597,
"grad_norm": 0.5143629576840156,
"learning_rate": 1.4449790365058474e-05,
"loss": 0.3213,
"step": 30535
},
{
"epoch": 3.561100746268657,
"grad_norm": 0.48541349392452526,
"learning_rate": 1.4442723707738199e-05,
"loss": 0.3321,
"step": 30540
},
{
"epoch": 3.5616837686567164,
"grad_norm": 0.4448177983046585,
"learning_rate": 1.4435658991909784e-05,
"loss": 0.3389,
"step": 30545
},
{
"epoch": 3.562266791044776,
"grad_norm": 0.4373481608964463,
"learning_rate": 1.4428596218623722e-05,
"loss": 0.3381,
"step": 30550
},
{
"epoch": 3.5628498134328357,
"grad_norm": 0.46052473260830273,
"learning_rate": 1.4421535388930146e-05,
"loss": 0.3417,
"step": 30555
},
{
"epoch": 3.5634328358208958,
"grad_norm": 0.4926637056965854,
"learning_rate": 1.4414476503878968e-05,
"loss": 0.3301,
"step": 30560
},
{
"epoch": 3.5640158582089554,
"grad_norm": 0.46336727427198904,
"learning_rate": 1.440741956451975e-05,
"loss": 0.3209,
"step": 30565
},
{
"epoch": 3.564598880597015,
"grad_norm": 0.453847692957194,
"learning_rate": 1.4400364571901803e-05,
"loss": 0.3384,
"step": 30570
},
{
"epoch": 3.5651819029850746,
"grad_norm": 0.4747549448739118,
"learning_rate": 1.4393311527074132e-05,
"loss": 0.3478,
"step": 30575
},
{
"epoch": 3.5657649253731343,
"grad_norm": 0.4363988368659044,
"learning_rate": 1.4386260431085457e-05,
"loss": 0.3205,
"step": 30580
},
{
"epoch": 3.566347947761194,
"grad_norm": 0.47010044342853596,
"learning_rate": 1.4379211284984213e-05,
"loss": 0.3349,
"step": 30585
},
{
"epoch": 3.5669309701492535,
"grad_norm": 0.45336617538481033,
"learning_rate": 1.4372164089818546e-05,
"loss": 0.3463,
"step": 30590
},
{
"epoch": 3.5675139925373136,
"grad_norm": 0.5103809496340458,
"learning_rate": 1.4365118846636283e-05,
"loss": 0.3386,
"step": 30595
},
{
"epoch": 3.5680970149253732,
"grad_norm": 0.48231877924813954,
"learning_rate": 1.4358075556485016e-05,
"loss": 0.3322,
"step": 30600
},
{
"epoch": 3.568680037313433,
"grad_norm": 0.4284399873001586,
"learning_rate": 1.4351034220411983e-05,
"loss": 0.3121,
"step": 30605
},
{
"epoch": 3.5692630597014925,
"grad_norm": 0.4665428934088034,
"learning_rate": 1.4343994839464192e-05,
"loss": 0.3221,
"step": 30610
},
{
"epoch": 3.569846082089552,
"grad_norm": 0.500284749218801,
"learning_rate": 1.4336957414688313e-05,
"loss": 0.3292,
"step": 30615
},
{
"epoch": 3.570429104477612,
"grad_norm": 0.444873404022475,
"learning_rate": 1.4329921947130748e-05,
"loss": 0.308,
"step": 30620
},
{
"epoch": 3.5710121268656714,
"grad_norm": 0.4377960289701257,
"learning_rate": 1.4322888437837604e-05,
"loss": 0.3139,
"step": 30625
},
{
"epoch": 3.5715951492537314,
"grad_norm": 0.49744359940608707,
"learning_rate": 1.43158568878547e-05,
"loss": 0.3313,
"step": 30630
},
{
"epoch": 3.572178171641791,
"grad_norm": 0.463954140641339,
"learning_rate": 1.4308827298227551e-05,
"loss": 0.3263,
"step": 30635
},
{
"epoch": 3.5727611940298507,
"grad_norm": 0.45837969153942526,
"learning_rate": 1.430179967000141e-05,
"loss": 0.325,
"step": 30640
},
{
"epoch": 3.5733442164179103,
"grad_norm": 0.451186575985876,
"learning_rate": 1.4294774004221185e-05,
"loss": 0.3294,
"step": 30645
},
{
"epoch": 3.57392723880597,
"grad_norm": 0.4580687873676085,
"learning_rate": 1.4287750301931557e-05,
"loss": 0.3306,
"step": 30650
},
{
"epoch": 3.57451026119403,
"grad_norm": 0.47140307322636815,
"learning_rate": 1.4280728564176866e-05,
"loss": 0.3389,
"step": 30655
},
{
"epoch": 3.5750932835820897,
"grad_norm": 0.4410827472739807,
"learning_rate": 1.4273708792001182e-05,
"loss": 0.3164,
"step": 30660
},
{
"epoch": 3.5756763059701493,
"grad_norm": 0.4440255101002687,
"learning_rate": 1.4266690986448273e-05,
"loss": 0.3394,
"step": 30665
},
{
"epoch": 3.576259328358209,
"grad_norm": 0.5035070979999087,
"learning_rate": 1.4259675148561627e-05,
"loss": 0.3518,
"step": 30670
},
{
"epoch": 3.5768423507462686,
"grad_norm": 0.454344166676145,
"learning_rate": 1.4252661279384427e-05,
"loss": 0.3319,
"step": 30675
},
{
"epoch": 3.5774253731343286,
"grad_norm": 0.5376653647152443,
"learning_rate": 1.424564937995957e-05,
"loss": 0.3532,
"step": 30680
},
{
"epoch": 3.578008395522388,
"grad_norm": 0.5173844403556396,
"learning_rate": 1.4238639451329666e-05,
"loss": 0.3395,
"step": 30685
},
{
"epoch": 3.578591417910448,
"grad_norm": 0.47301287423528704,
"learning_rate": 1.4231631494537007e-05,
"loss": 0.3208,
"step": 30690
},
{
"epoch": 3.5791744402985075,
"grad_norm": 0.4635506259497985,
"learning_rate": 1.4224625510623618e-05,
"loss": 0.3272,
"step": 30695
},
{
"epoch": 3.579757462686567,
"grad_norm": 0.5085480217776777,
"learning_rate": 1.4217621500631222e-05,
"loss": 0.3359,
"step": 30700
},
{
"epoch": 3.5803404850746268,
"grad_norm": 0.5790643628800386,
"learning_rate": 1.4210619465601249e-05,
"loss": 0.3283,
"step": 30705
},
{
"epoch": 3.5809235074626864,
"grad_norm": 0.4351335248079688,
"learning_rate": 1.4203619406574833e-05,
"loss": 0.3152,
"step": 30710
},
{
"epoch": 3.5815065298507465,
"grad_norm": 0.5269015701887055,
"learning_rate": 1.4196621324592818e-05,
"loss": 0.34,
"step": 30715
},
{
"epoch": 3.582089552238806,
"grad_norm": 0.480888650874225,
"learning_rate": 1.4189625220695746e-05,
"loss": 0.3245,
"step": 30720
},
{
"epoch": 3.5826725746268657,
"grad_norm": 0.41132331450331927,
"learning_rate": 1.4182631095923884e-05,
"loss": 0.3014,
"step": 30725
},
{
"epoch": 3.5832555970149254,
"grad_norm": 0.4517273439773864,
"learning_rate": 1.4175638951317166e-05,
"loss": 0.3526,
"step": 30730
},
{
"epoch": 3.583838619402985,
"grad_norm": 0.47586947349794534,
"learning_rate": 1.4168648787915289e-05,
"loss": 0.3388,
"step": 30735
},
{
"epoch": 3.5844216417910446,
"grad_norm": 0.4399132035009382,
"learning_rate": 1.41616606067576e-05,
"loss": 0.3391,
"step": 30740
},
{
"epoch": 3.5850046641791042,
"grad_norm": 0.4704910767583765,
"learning_rate": 1.4154674408883179e-05,
"loss": 0.324,
"step": 30745
},
{
"epoch": 3.5855876865671643,
"grad_norm": 0.5182597288204652,
"learning_rate": 1.4147690195330815e-05,
"loss": 0.3237,
"step": 30750
},
{
"epoch": 3.586170708955224,
"grad_norm": 0.4700820162789669,
"learning_rate": 1.4140707967138983e-05,
"loss": 0.3167,
"step": 30755
},
{
"epoch": 3.5867537313432836,
"grad_norm": 0.5084414884924573,
"learning_rate": 1.413372772534588e-05,
"loss": 0.3284,
"step": 30760
},
{
"epoch": 3.587336753731343,
"grad_norm": 0.4427609210397223,
"learning_rate": 1.4126749470989408e-05,
"loss": 0.349,
"step": 30765
},
{
"epoch": 3.587919776119403,
"grad_norm": 0.45517836181530963,
"learning_rate": 1.411977320510714e-05,
"loss": 0.3343,
"step": 30770
},
{
"epoch": 3.588502798507463,
"grad_norm": 0.48551626657276176,
"learning_rate": 1.4112798928736415e-05,
"loss": 0.3268,
"step": 30775
},
{
"epoch": 3.5890858208955225,
"grad_norm": 0.465285706126102,
"learning_rate": 1.410582664291421e-05,
"loss": 0.3262,
"step": 30780
},
{
"epoch": 3.589668843283582,
"grad_norm": 0.43831796135475604,
"learning_rate": 1.4098856348677263e-05,
"loss": 0.3371,
"step": 30785
},
{
"epoch": 3.590251865671642,
"grad_norm": 0.4495296181918966,
"learning_rate": 1.4091888047061974e-05,
"loss": 0.3249,
"step": 30790
},
{
"epoch": 3.5908348880597014,
"grad_norm": 0.4730025916124849,
"learning_rate": 1.4084921739104465e-05,
"loss": 0.3173,
"step": 30795
},
{
"epoch": 3.591417910447761,
"grad_norm": 0.446813303352069,
"learning_rate": 1.4077957425840563e-05,
"loss": 0.3148,
"step": 30800
},
{
"epoch": 3.5920009328358207,
"grad_norm": 0.43225037294348756,
"learning_rate": 1.407099510830579e-05,
"loss": 0.3097,
"step": 30805
},
{
"epoch": 3.5925839552238807,
"grad_norm": 0.4547572890001812,
"learning_rate": 1.406403478753538e-05,
"loss": 0.324,
"step": 30810
},
{
"epoch": 3.5931669776119404,
"grad_norm": 0.47153439536213787,
"learning_rate": 1.4057076464564272e-05,
"loss": 0.3182,
"step": 30815
},
{
"epoch": 3.59375,
"grad_norm": 0.44347109812292,
"learning_rate": 1.4050120140427081e-05,
"loss": 0.3239,
"step": 30820
},
{
"epoch": 3.5943330223880596,
"grad_norm": 0.44664336753678724,
"learning_rate": 1.4043165816158172e-05,
"loss": 0.3211,
"step": 30825
},
{
"epoch": 3.5949160447761193,
"grad_norm": 0.4332024839803423,
"learning_rate": 1.4036213492791561e-05,
"loss": 0.3201,
"step": 30830
},
{
"epoch": 3.5954990671641793,
"grad_norm": 0.4974923109860408,
"learning_rate": 1.4029263171361018e-05,
"loss": 0.3223,
"step": 30835
},
{
"epoch": 3.596082089552239,
"grad_norm": 0.459723389493139,
"learning_rate": 1.4022314852899968e-05,
"loss": 0.3309,
"step": 30840
},
{
"epoch": 3.5966651119402986,
"grad_norm": 0.4928228806000406,
"learning_rate": 1.4015368538441567e-05,
"loss": 0.3431,
"step": 30845
},
{
"epoch": 3.597248134328358,
"grad_norm": 0.46940709708411177,
"learning_rate": 1.4008424229018668e-05,
"loss": 0.3212,
"step": 30850
},
{
"epoch": 3.597831156716418,
"grad_norm": 0.45432976582878304,
"learning_rate": 1.4001481925663818e-05,
"loss": 0.3122,
"step": 30855
},
{
"epoch": 3.5984141791044775,
"grad_norm": 0.48216334391261273,
"learning_rate": 1.3994541629409275e-05,
"loss": 0.2973,
"step": 30860
},
{
"epoch": 3.598997201492537,
"grad_norm": 0.4598175603708983,
"learning_rate": 1.3987603341287003e-05,
"loss": 0.3325,
"step": 30865
},
{
"epoch": 3.599580223880597,
"grad_norm": 0.42948881682522444,
"learning_rate": 1.398066706232864e-05,
"loss": 0.3295,
"step": 30870
},
{
"epoch": 3.600163246268657,
"grad_norm": 0.5255344992401619,
"learning_rate": 1.3973732793565553e-05,
"loss": 0.343,
"step": 30875
},
{
"epoch": 3.6007462686567164,
"grad_norm": 0.4937335780902401,
"learning_rate": 1.3966800536028802e-05,
"loss": 0.3469,
"step": 30880
},
{
"epoch": 3.601329291044776,
"grad_norm": 0.4300854945636151,
"learning_rate": 1.3959870290749148e-05,
"loss": 0.2882,
"step": 30885
},
{
"epoch": 3.6019123134328357,
"grad_norm": 0.4111604436999921,
"learning_rate": 1.395294205875705e-05,
"loss": 0.3022,
"step": 30890
},
{
"epoch": 3.6024953358208958,
"grad_norm": 0.44701929226917175,
"learning_rate": 1.394601584108267e-05,
"loss": 0.3225,
"step": 30895
},
{
"epoch": 3.6030783582089554,
"grad_norm": 0.4527450630108651,
"learning_rate": 1.3939091638755882e-05,
"loss": 0.3108,
"step": 30900
},
{
"epoch": 3.603661380597015,
"grad_norm": 0.45689457674625045,
"learning_rate": 1.3932169452806221e-05,
"loss": 0.3384,
"step": 30905
},
{
"epoch": 3.6042444029850746,
"grad_norm": 0.4739513240901544,
"learning_rate": 1.3925249284262984e-05,
"loss": 0.3261,
"step": 30910
},
{
"epoch": 3.6048274253731343,
"grad_norm": 0.46259649913304157,
"learning_rate": 1.3918331134155104e-05,
"loss": 0.3382,
"step": 30915
},
{
"epoch": 3.605410447761194,
"grad_norm": 0.5476694026778616,
"learning_rate": 1.3911415003511258e-05,
"loss": 0.3241,
"step": 30920
},
{
"epoch": 3.6059934701492535,
"grad_norm": 0.5187947756610837,
"learning_rate": 1.3904500893359809e-05,
"loss": 0.3443,
"step": 30925
},
{
"epoch": 3.6065764925373136,
"grad_norm": 0.44187033269340575,
"learning_rate": 1.3897588804728818e-05,
"loss": 0.3271,
"step": 30930
},
{
"epoch": 3.6071595149253732,
"grad_norm": 0.42632349752822357,
"learning_rate": 1.3890678738646038e-05,
"loss": 0.3356,
"step": 30935
},
{
"epoch": 3.607742537313433,
"grad_norm": 0.5299985861528723,
"learning_rate": 1.3883770696138946e-05,
"loss": 0.3532,
"step": 30940
},
{
"epoch": 3.6083255597014925,
"grad_norm": 0.4936740664032516,
"learning_rate": 1.3876864678234674e-05,
"loss": 0.3329,
"step": 30945
},
{
"epoch": 3.608908582089552,
"grad_norm": 0.47305652469131637,
"learning_rate": 1.3869960685960118e-05,
"loss": 0.3295,
"step": 30950
},
{
"epoch": 3.609491604477612,
"grad_norm": 0.4784904276737314,
"learning_rate": 1.3863058720341799e-05,
"loss": 0.3189,
"step": 30955
},
{
"epoch": 3.6100746268656714,
"grad_norm": 0.4753044017218112,
"learning_rate": 1.3856158782406007e-05,
"loss": 0.3272,
"step": 30960
},
{
"epoch": 3.6106576492537314,
"grad_norm": 0.5188411601643111,
"learning_rate": 1.3849260873178671e-05,
"loss": 0.3478,
"step": 30965
},
{
"epoch": 3.611240671641791,
"grad_norm": 0.47575059374767503,
"learning_rate": 1.3842364993685453e-05,
"loss": 0.3105,
"step": 30970
},
{
"epoch": 3.6118236940298507,
"grad_norm": 0.4746714512523541,
"learning_rate": 1.3835471144951705e-05,
"loss": 0.3342,
"step": 30975
},
{
"epoch": 3.6124067164179103,
"grad_norm": 0.4252946453989581,
"learning_rate": 1.3828579328002473e-05,
"loss": 0.3216,
"step": 30980
},
{
"epoch": 3.61298973880597,
"grad_norm": 0.4708342921734325,
"learning_rate": 1.382168954386251e-05,
"loss": 0.3368,
"step": 30985
},
{
"epoch": 3.61357276119403,
"grad_norm": 0.44174710006528406,
"learning_rate": 1.3814801793556264e-05,
"loss": 0.3133,
"step": 30990
},
{
"epoch": 3.6141557835820897,
"grad_norm": 0.4498021523154961,
"learning_rate": 1.3807916078107851e-05,
"loss": 0.3114,
"step": 30995
},
{
"epoch": 3.6147388059701493,
"grad_norm": 0.44302185317222054,
"learning_rate": 1.3801032398541153e-05,
"loss": 0.3347,
"step": 31000
},
{
"epoch": 3.615321828358209,
"grad_norm": 0.4790054963282025,
"learning_rate": 1.3794150755879665e-05,
"loss": 0.3236,
"step": 31005
},
{
"epoch": 3.6159048507462686,
"grad_norm": 0.47125434231183194,
"learning_rate": 1.3787271151146658e-05,
"loss": 0.3239,
"step": 31010
},
{
"epoch": 3.6164878731343286,
"grad_norm": 0.46755831006417453,
"learning_rate": 1.3780393585365043e-05,
"loss": 0.3219,
"step": 31015
},
{
"epoch": 3.617070895522388,
"grad_norm": 0.4784430740098405,
"learning_rate": 1.3773518059557445e-05,
"loss": 0.3306,
"step": 31020
},
{
"epoch": 3.617653917910448,
"grad_norm": 0.44043478119241186,
"learning_rate": 1.3766644574746202e-05,
"loss": 0.3205,
"step": 31025
},
{
"epoch": 3.6182369402985075,
"grad_norm": 0.4292599765900938,
"learning_rate": 1.3759773131953321e-05,
"loss": 0.3284,
"step": 31030
},
{
"epoch": 3.618819962686567,
"grad_norm": 0.4783147794772432,
"learning_rate": 1.3752903732200535e-05,
"loss": 0.326,
"step": 31035
},
{
"epoch": 3.6194029850746268,
"grad_norm": 0.4561539018021913,
"learning_rate": 1.3746036376509252e-05,
"loss": 0.3302,
"step": 31040
},
{
"epoch": 3.6199860074626864,
"grad_norm": 0.46498336262089046,
"learning_rate": 1.3739171065900577e-05,
"loss": 0.3182,
"step": 31045
},
{
"epoch": 3.6205690298507465,
"grad_norm": 0.45313508210464637,
"learning_rate": 1.3732307801395322e-05,
"loss": 0.3251,
"step": 31050
},
{
"epoch": 3.621152052238806,
"grad_norm": 0.4563700798823811,
"learning_rate": 1.3725446584013984e-05,
"loss": 0.3142,
"step": 31055
},
{
"epoch": 3.6217350746268657,
"grad_norm": 0.45174279163432074,
"learning_rate": 1.3718587414776756e-05,
"loss": 0.3241,
"step": 31060
},
{
"epoch": 3.6223180970149254,
"grad_norm": 0.49544856606238696,
"learning_rate": 1.3711730294703545e-05,
"loss": 0.3444,
"step": 31065
},
{
"epoch": 3.622901119402985,
"grad_norm": 0.4744435909483976,
"learning_rate": 1.3704875224813928e-05,
"loss": 0.3338,
"step": 31070
},
{
"epoch": 3.6234841417910446,
"grad_norm": 0.4682710577465102,
"learning_rate": 1.3698022206127201e-05,
"loss": 0.3306,
"step": 31075
},
{
"epoch": 3.6240671641791042,
"grad_norm": 0.46582122906724877,
"learning_rate": 1.3691171239662315e-05,
"loss": 0.3412,
"step": 31080
},
{
"epoch": 3.6246501865671643,
"grad_norm": 0.5003394888146652,
"learning_rate": 1.3684322326437976e-05,
"loss": 0.3349,
"step": 31085
},
{
"epoch": 3.625233208955224,
"grad_norm": 0.4492810392951438,
"learning_rate": 1.3677475467472534e-05,
"loss": 0.3329,
"step": 31090
},
{
"epoch": 3.6258162313432836,
"grad_norm": 0.4272140190782571,
"learning_rate": 1.367063066378405e-05,
"loss": 0.314,
"step": 31095
},
{
"epoch": 3.626399253731343,
"grad_norm": 0.45743973011498534,
"learning_rate": 1.366378791639028e-05,
"loss": 0.3143,
"step": 31100
},
{
"epoch": 3.626982276119403,
"grad_norm": 0.4905799603140713,
"learning_rate": 1.3656947226308686e-05,
"loss": 0.3515,
"step": 31105
},
{
"epoch": 3.627565298507463,
"grad_norm": 0.5056921514198293,
"learning_rate": 1.3650108594556404e-05,
"loss": 0.3273,
"step": 31110
},
{
"epoch": 3.6281483208955225,
"grad_norm": 0.4460771657180042,
"learning_rate": 1.3643272022150286e-05,
"loss": 0.3027,
"step": 31115
},
{
"epoch": 3.628731343283582,
"grad_norm": 0.46123327257003033,
"learning_rate": 1.3636437510106836e-05,
"loss": 0.3407,
"step": 31120
},
{
"epoch": 3.629314365671642,
"grad_norm": 0.45680882201875966,
"learning_rate": 1.3629605059442315e-05,
"loss": 0.3225,
"step": 31125
},
{
"epoch": 3.6298973880597014,
"grad_norm": 0.4351080943532293,
"learning_rate": 1.362277467117261e-05,
"loss": 0.3198,
"step": 31130
},
{
"epoch": 3.630480410447761,
"grad_norm": 0.45381964677387365,
"learning_rate": 1.3615946346313368e-05,
"loss": 0.3414,
"step": 31135
},
{
"epoch": 3.6310634328358207,
"grad_norm": 0.5275309351178956,
"learning_rate": 1.3609120085879872e-05,
"loss": 0.3313,
"step": 31140
},
{
"epoch": 3.6316464552238807,
"grad_norm": 0.4898873747004423,
"learning_rate": 1.360229589088713e-05,
"loss": 0.3227,
"step": 31145
},
{
"epoch": 3.6322294776119404,
"grad_norm": 0.4507389438058164,
"learning_rate": 1.3595473762349825e-05,
"loss": 0.3068,
"step": 31150
},
{
"epoch": 3.6328125,
"grad_norm": 0.5061611948054943,
"learning_rate": 1.3588653701282359e-05,
"loss": 0.344,
"step": 31155
},
{
"epoch": 3.6333955223880596,
"grad_norm": 0.48318510481799876,
"learning_rate": 1.3581835708698796e-05,
"loss": 0.3343,
"step": 31160
},
{
"epoch": 3.6339785447761193,
"grad_norm": 0.48286218202518044,
"learning_rate": 1.357501978561292e-05,
"loss": 0.3227,
"step": 31165
},
{
"epoch": 3.6345615671641793,
"grad_norm": 0.5035539243855311,
"learning_rate": 1.3568205933038164e-05,
"loss": 0.3423,
"step": 31170
},
{
"epoch": 3.635144589552239,
"grad_norm": 0.4665797479912174,
"learning_rate": 1.3561394151987722e-05,
"loss": 0.3301,
"step": 31175
},
{
"epoch": 3.6357276119402986,
"grad_norm": 0.4761977558837576,
"learning_rate": 1.3554584443474405e-05,
"loss": 0.3318,
"step": 31180
},
{
"epoch": 3.636310634328358,
"grad_norm": 0.43260977443560905,
"learning_rate": 1.3547776808510782e-05,
"loss": 0.3284,
"step": 31185
},
{
"epoch": 3.636893656716418,
"grad_norm": 0.4189791268300447,
"learning_rate": 1.3540971248109063e-05,
"loss": 0.3031,
"step": 31190
},
{
"epoch": 3.6374766791044775,
"grad_norm": 0.481164745706147,
"learning_rate": 1.3534167763281175e-05,
"loss": 0.3234,
"step": 31195
},
{
"epoch": 3.638059701492537,
"grad_norm": 0.49315789036743013,
"learning_rate": 1.352736635503873e-05,
"loss": 0.3188,
"step": 31200
},
{
"epoch": 3.638642723880597,
"grad_norm": 0.455394463954907,
"learning_rate": 1.3520567024393033e-05,
"loss": 0.3302,
"step": 31205
},
{
"epoch": 3.639225746268657,
"grad_norm": 0.4533384008353099,
"learning_rate": 1.3513769772355083e-05,
"loss": 0.3245,
"step": 31210
},
{
"epoch": 3.6398087686567164,
"grad_norm": 0.4542700047601782,
"learning_rate": 1.350697459993556e-05,
"loss": 0.3247,
"step": 31215
},
{
"epoch": 3.640391791044776,
"grad_norm": 0.4506638594465722,
"learning_rate": 1.3500181508144855e-05,
"loss": 0.3276,
"step": 31220
},
{
"epoch": 3.6409748134328357,
"grad_norm": 0.4736122106785712,
"learning_rate": 1.3493390497993014e-05,
"loss": 0.3346,
"step": 31225
},
{
"epoch": 3.6415578358208958,
"grad_norm": 0.45184832963680616,
"learning_rate": 1.3486601570489809e-05,
"loss": 0.3294,
"step": 31230
},
{
"epoch": 3.6421408582089554,
"grad_norm": 0.47620322229028306,
"learning_rate": 1.3479814726644686e-05,
"loss": 0.3391,
"step": 31235
},
{
"epoch": 3.642723880597015,
"grad_norm": 0.44130327890535087,
"learning_rate": 1.3473029967466779e-05,
"loss": 0.3102,
"step": 31240
},
{
"epoch": 3.6433069029850746,
"grad_norm": 0.4927547189276676,
"learning_rate": 1.3466247293964922e-05,
"loss": 0.3244,
"step": 31245
},
{
"epoch": 3.6438899253731343,
"grad_norm": 0.4761645226753004,
"learning_rate": 1.3459466707147644e-05,
"loss": 0.3433,
"step": 31250
},
{
"epoch": 3.644472947761194,
"grad_norm": 0.43866264751097683,
"learning_rate": 1.3452688208023123e-05,
"loss": 0.3218,
"step": 31255
},
{
"epoch": 3.6450559701492535,
"grad_norm": 0.4646522793504169,
"learning_rate": 1.3445911797599293e-05,
"loss": 0.3431,
"step": 31260
},
{
"epoch": 3.6456389925373136,
"grad_norm": 0.48836516077234127,
"learning_rate": 1.3439137476883706e-05,
"loss": 0.3368,
"step": 31265
},
{
"epoch": 3.6462220149253732,
"grad_norm": 0.46470621472139234,
"learning_rate": 1.3432365246883671e-05,
"loss": 0.3181,
"step": 31270
},
{
"epoch": 3.646805037313433,
"grad_norm": 0.44878768581362055,
"learning_rate": 1.3425595108606138e-05,
"loss": 0.3261,
"step": 31275
},
{
"epoch": 3.6473880597014925,
"grad_norm": 0.4500183550241171,
"learning_rate": 1.3418827063057754e-05,
"loss": 0.3094,
"step": 31280
},
{
"epoch": 3.647971082089552,
"grad_norm": 0.4651200613887222,
"learning_rate": 1.3412061111244877e-05,
"loss": 0.3234,
"step": 31285
},
{
"epoch": 3.648554104477612,
"grad_norm": 0.4346814567803131,
"learning_rate": 1.3405297254173532e-05,
"loss": 0.3361,
"step": 31290
},
{
"epoch": 3.6491371268656714,
"grad_norm": 0.4945965432343019,
"learning_rate": 1.3398535492849442e-05,
"loss": 0.317,
"step": 31295
},
{
"epoch": 3.6497201492537314,
"grad_norm": 0.5081834516913957,
"learning_rate": 1.3391775828278023e-05,
"loss": 0.3253,
"step": 31300
},
{
"epoch": 3.650303171641791,
"grad_norm": 0.4759108436963413,
"learning_rate": 1.338501826146435e-05,
"loss": 0.3233,
"step": 31305
},
{
"epoch": 3.6508861940298507,
"grad_norm": 0.5018965996529728,
"learning_rate": 1.3378262793413237e-05,
"loss": 0.3347,
"step": 31310
},
{
"epoch": 3.6514692164179103,
"grad_norm": 0.4455252560222746,
"learning_rate": 1.3371509425129131e-05,
"loss": 0.3245,
"step": 31315
},
{
"epoch": 3.65205223880597,
"grad_norm": 0.47711378350122946,
"learning_rate": 1.3364758157616219e-05,
"loss": 0.3129,
"step": 31320
},
{
"epoch": 3.65263526119403,
"grad_norm": 0.4418243434050547,
"learning_rate": 1.3358008991878329e-05,
"loss": 0.3417,
"step": 31325
},
{
"epoch": 3.6532182835820897,
"grad_norm": 0.5501608681321434,
"learning_rate": 1.3351261928919007e-05,
"loss": 0.3616,
"step": 31330
},
{
"epoch": 3.6538013059701493,
"grad_norm": 0.5282777264117906,
"learning_rate": 1.3344516969741471e-05,
"loss": 0.3321,
"step": 31335
},
{
"epoch": 3.654384328358209,
"grad_norm": 0.4498665078075985,
"learning_rate": 1.3337774115348639e-05,
"loss": 0.3001,
"step": 31340
},
{
"epoch": 3.6549673507462686,
"grad_norm": 0.4629014153670532,
"learning_rate": 1.3331033366743101e-05,
"loss": 0.3079,
"step": 31345
},
{
"epoch": 3.6555503731343286,
"grad_norm": 0.46894722870541605,
"learning_rate": 1.3324294724927155e-05,
"loss": 0.3003,
"step": 31350
},
{
"epoch": 3.656133395522388,
"grad_norm": 0.49917555020652415,
"learning_rate": 1.3317558190902752e-05,
"loss": 0.3379,
"step": 31355
},
{
"epoch": 3.656716417910448,
"grad_norm": 0.5067293504331861,
"learning_rate": 1.3310823765671571e-05,
"loss": 0.3567,
"step": 31360
},
{
"epoch": 3.6572994402985075,
"grad_norm": 0.4914156511449992,
"learning_rate": 1.3304091450234934e-05,
"loss": 0.3388,
"step": 31365
},
{
"epoch": 3.657882462686567,
"grad_norm": 0.4553188970482324,
"learning_rate": 1.32973612455939e-05,
"loss": 0.3098,
"step": 31370
},
{
"epoch": 3.6584654850746268,
"grad_norm": 0.47881768265702407,
"learning_rate": 1.3290633152749166e-05,
"loss": 0.3346,
"step": 31375
},
{
"epoch": 3.6590485074626864,
"grad_norm": 0.48698493411426985,
"learning_rate": 1.3283907172701135e-05,
"loss": 0.3395,
"step": 31380
},
{
"epoch": 3.6596315298507465,
"grad_norm": 0.4406234083551808,
"learning_rate": 1.3277183306449902e-05,
"loss": 0.3039,
"step": 31385
},
{
"epoch": 3.660214552238806,
"grad_norm": 0.4615525504738181,
"learning_rate": 1.3270461554995243e-05,
"loss": 0.3311,
"step": 31390
},
{
"epoch": 3.6607975746268657,
"grad_norm": 0.4256954983134391,
"learning_rate": 1.3263741919336608e-05,
"loss": 0.3226,
"step": 31395
},
{
"epoch": 3.6613805970149254,
"grad_norm": 0.45771343012051446,
"learning_rate": 1.3257024400473162e-05,
"loss": 0.3087,
"step": 31400
},
{
"epoch": 3.661963619402985,
"grad_norm": 0.458134612632429,
"learning_rate": 1.3250308999403715e-05,
"loss": 0.3087,
"step": 31405
},
{
"epoch": 3.6625466417910446,
"grad_norm": 0.4895246636519347,
"learning_rate": 1.3243595717126792e-05,
"loss": 0.3318,
"step": 31410
},
{
"epoch": 3.6631296641791042,
"grad_norm": 0.4704772905461017,
"learning_rate": 1.3236884554640589e-05,
"loss": 0.3164,
"step": 31415
},
{
"epoch": 3.6637126865671643,
"grad_norm": 0.43385243655184663,
"learning_rate": 1.3230175512943e-05,
"loss": 0.3141,
"step": 31420
},
{
"epoch": 3.664295708955224,
"grad_norm": 0.44902480668516553,
"learning_rate": 1.3223468593031585e-05,
"loss": 0.3347,
"step": 31425
},
{
"epoch": 3.6648787313432836,
"grad_norm": 0.441123561924162,
"learning_rate": 1.3216763795903608e-05,
"loss": 0.2972,
"step": 31430
},
{
"epoch": 3.665461753731343,
"grad_norm": 0.4451649744099363,
"learning_rate": 1.321006112255601e-05,
"loss": 0.323,
"step": 31435
},
{
"epoch": 3.666044776119403,
"grad_norm": 0.4718936590855251,
"learning_rate": 1.3203360573985394e-05,
"loss": 0.3391,
"step": 31440
},
{
"epoch": 3.666627798507463,
"grad_norm": 0.48261580115801245,
"learning_rate": 1.31966621511881e-05,
"loss": 0.3145,
"step": 31445
},
{
"epoch": 3.6672108208955225,
"grad_norm": 0.44949290096900263,
"learning_rate": 1.3189965855160088e-05,
"loss": 0.315,
"step": 31450
},
{
"epoch": 3.667793843283582,
"grad_norm": 0.45535760627703,
"learning_rate": 1.318327168689705e-05,
"loss": 0.3195,
"step": 31455
},
{
"epoch": 3.668376865671642,
"grad_norm": 0.5133884073554863,
"learning_rate": 1.3176579647394338e-05,
"loss": 0.3213,
"step": 31460
},
{
"epoch": 3.6689598880597014,
"grad_norm": 0.48407952311970437,
"learning_rate": 1.3169889737647001e-05,
"loss": 0.3238,
"step": 31465
},
{
"epoch": 3.669542910447761,
"grad_norm": 0.4712862288057401,
"learning_rate": 1.3163201958649757e-05,
"loss": 0.3113,
"step": 31470
},
{
"epoch": 3.6701259328358207,
"grad_norm": 0.4468691551751411,
"learning_rate": 1.315651631139703e-05,
"loss": 0.3152,
"step": 31475
},
{
"epoch": 3.6707089552238807,
"grad_norm": 0.45416081387858553,
"learning_rate": 1.314983279688288e-05,
"loss": 0.3303,
"step": 31480
},
{
"epoch": 3.6712919776119404,
"grad_norm": 0.46364841940380214,
"learning_rate": 1.3143151416101118e-05,
"loss": 0.3342,
"step": 31485
},
{
"epoch": 3.671875,
"grad_norm": 0.4699111081809452,
"learning_rate": 1.3136472170045171e-05,
"loss": 0.3244,
"step": 31490
},
{
"epoch": 3.6724580223880596,
"grad_norm": 0.4511674288857646,
"learning_rate": 1.3129795059708209e-05,
"loss": 0.3346,
"step": 31495
},
{
"epoch": 3.6730410447761193,
"grad_norm": 0.5140414045658062,
"learning_rate": 1.3123120086083026e-05,
"loss": 0.319,
"step": 31500
},
{
"epoch": 3.6736240671641793,
"grad_norm": 0.49386677317891153,
"learning_rate": 1.3116447250162145e-05,
"loss": 0.3391,
"step": 31505
},
{
"epoch": 3.674207089552239,
"grad_norm": 0.4724215098111002,
"learning_rate": 1.3109776552937742e-05,
"loss": 0.3327,
"step": 31510
},
{
"epoch": 3.6747901119402986,
"grad_norm": 0.48079247659077834,
"learning_rate": 1.3103107995401689e-05,
"loss": 0.3278,
"step": 31515
},
{
"epoch": 3.675373134328358,
"grad_norm": 0.4673231523048643,
"learning_rate": 1.3096441578545544e-05,
"loss": 0.3396,
"step": 31520
},
{
"epoch": 3.675956156716418,
"grad_norm": 0.49948194068544777,
"learning_rate": 1.3089777303360534e-05,
"loss": 0.3243,
"step": 31525
},
{
"epoch": 3.6765391791044775,
"grad_norm": 0.46695038827477636,
"learning_rate": 1.308311517083756e-05,
"loss": 0.3123,
"step": 31530
},
{
"epoch": 3.677122201492537,
"grad_norm": 0.5344194598937325,
"learning_rate": 1.3076455181967246e-05,
"loss": 0.3344,
"step": 31535
},
{
"epoch": 3.677705223880597,
"grad_norm": 0.4784688103852506,
"learning_rate": 1.306979733773983e-05,
"loss": 0.3169,
"step": 31540
},
{
"epoch": 3.678288246268657,
"grad_norm": 0.48524859817677335,
"learning_rate": 1.3063141639145313e-05,
"loss": 0.3281,
"step": 31545
},
{
"epoch": 3.6788712686567164,
"grad_norm": 0.420086663938523,
"learning_rate": 1.3056488087173302e-05,
"loss": 0.3211,
"step": 31550
},
{
"epoch": 3.679454291044776,
"grad_norm": 0.47507384117275503,
"learning_rate": 1.3049836682813127e-05,
"loss": 0.3246,
"step": 31555
},
{
"epoch": 3.6800373134328357,
"grad_norm": 0.43644833261680144,
"learning_rate": 1.3043187427053788e-05,
"loss": 0.312,
"step": 31560
},
{
"epoch": 3.6806203358208958,
"grad_norm": 0.5023075759999288,
"learning_rate": 1.3036540320883967e-05,
"loss": 0.3392,
"step": 31565
},
{
"epoch": 3.6812033582089554,
"grad_norm": 0.49592396130851113,
"learning_rate": 1.3029895365292018e-05,
"loss": 0.3275,
"step": 31570
},
{
"epoch": 3.681786380597015,
"grad_norm": 0.48003364363140033,
"learning_rate": 1.3023252561266e-05,
"loss": 0.3341,
"step": 31575
},
{
"epoch": 3.6823694029850746,
"grad_norm": 0.45634224992258754,
"learning_rate": 1.3016611909793613e-05,
"loss": 0.3255,
"step": 31580
},
{
"epoch": 3.6829524253731343,
"grad_norm": 0.4701313857081696,
"learning_rate": 1.3009973411862264e-05,
"loss": 0.3177,
"step": 31585
},
{
"epoch": 3.683535447761194,
"grad_norm": 0.4344048690540488,
"learning_rate": 1.3003337068459037e-05,
"loss": 0.328,
"step": 31590
},
{
"epoch": 3.6841184701492535,
"grad_norm": 0.49223030335934287,
"learning_rate": 1.29967028805707e-05,
"loss": 0.3466,
"step": 31595
},
{
"epoch": 3.6847014925373136,
"grad_norm": 0.41996922306859436,
"learning_rate": 1.2990070849183678e-05,
"loss": 0.3282,
"step": 31600
},
{
"epoch": 3.6852845149253732,
"grad_norm": 0.5101676986445276,
"learning_rate": 1.2983440975284101e-05,
"loss": 0.3302,
"step": 31605
},
{
"epoch": 3.685867537313433,
"grad_norm": 0.49682199517574394,
"learning_rate": 1.2976813259857773e-05,
"loss": 0.2948,
"step": 31610
},
{
"epoch": 3.6864505597014925,
"grad_norm": 0.4631508954087268,
"learning_rate": 1.2970187703890149e-05,
"loss": 0.3347,
"step": 31615
},
{
"epoch": 3.687033582089552,
"grad_norm": 0.47885669120682994,
"learning_rate": 1.2963564308366416e-05,
"loss": 0.3261,
"step": 31620
},
{
"epoch": 3.687616604477612,
"grad_norm": 0.48543010067184783,
"learning_rate": 1.2956943074271388e-05,
"loss": 0.3414,
"step": 31625
},
{
"epoch": 3.6881996268656714,
"grad_norm": 0.4984091597801686,
"learning_rate": 1.295032400258958e-05,
"loss": 0.3308,
"step": 31630
},
{
"epoch": 3.6887826492537314,
"grad_norm": 0.46677066628033437,
"learning_rate": 1.2943707094305196e-05,
"loss": 0.3312,
"step": 31635
},
{
"epoch": 3.689365671641791,
"grad_norm": 0.4773263411727992,
"learning_rate": 1.2937092350402097e-05,
"loss": 0.3155,
"step": 31640
},
{
"epoch": 3.6899486940298507,
"grad_norm": 0.42832969149943423,
"learning_rate": 1.2930479771863835e-05,
"loss": 0.3408,
"step": 31645
},
{
"epoch": 3.6905317164179103,
"grad_norm": 0.45644740360498687,
"learning_rate": 1.2923869359673646e-05,
"loss": 0.3264,
"step": 31650
},
{
"epoch": 3.69111473880597,
"grad_norm": 0.44582819806361973,
"learning_rate": 1.2917261114814414e-05,
"loss": 0.3336,
"step": 31655
},
{
"epoch": 3.69169776119403,
"grad_norm": 0.45150250658632707,
"learning_rate": 1.2910655038268749e-05,
"loss": 0.3073,
"step": 31660
},
{
"epoch": 3.6922807835820897,
"grad_norm": 0.4326064459735902,
"learning_rate": 1.2904051131018877e-05,
"loss": 0.3327,
"step": 31665
},
{
"epoch": 3.6928638059701493,
"grad_norm": 0.45941836798825053,
"learning_rate": 1.2897449394046773e-05,
"loss": 0.327,
"step": 31670
},
{
"epoch": 3.693446828358209,
"grad_norm": 0.5060818866839104,
"learning_rate": 1.2890849828334029e-05,
"loss": 0.3183,
"step": 31675
},
{
"epoch": 3.6940298507462686,
"grad_norm": 0.48612612030918756,
"learning_rate": 1.2884252434861938e-05,
"loss": 0.2979,
"step": 31680
},
{
"epoch": 3.6946128731343286,
"grad_norm": 0.49800443359831537,
"learning_rate": 1.2877657214611477e-05,
"loss": 0.3155,
"step": 31685
},
{
"epoch": 3.695195895522388,
"grad_norm": 0.4948702647432376,
"learning_rate": 1.2871064168563291e-05,
"loss": 0.3196,
"step": 31690
},
{
"epoch": 3.695778917910448,
"grad_norm": 0.44893136195759986,
"learning_rate": 1.2864473297697693e-05,
"loss": 0.322,
"step": 31695
},
{
"epoch": 3.6963619402985075,
"grad_norm": 0.45850928885668435,
"learning_rate": 1.2857884602994706e-05,
"loss": 0.3275,
"step": 31700
},
{
"epoch": 3.696944962686567,
"grad_norm": 0.4672951709579247,
"learning_rate": 1.2851298085433972e-05,
"loss": 0.3318,
"step": 31705
},
{
"epoch": 3.6975279850746268,
"grad_norm": 0.5054266321428105,
"learning_rate": 1.2844713745994871e-05,
"loss": 0.3474,
"step": 31710
},
{
"epoch": 3.6981110074626864,
"grad_norm": 0.49467511543653125,
"learning_rate": 1.283813158565641e-05,
"loss": 0.3328,
"step": 31715
},
{
"epoch": 3.6986940298507465,
"grad_norm": 0.5031012319202796,
"learning_rate": 1.2831551605397321e-05,
"loss": 0.3221,
"step": 31720
},
{
"epoch": 3.699277052238806,
"grad_norm": 0.45112489659844346,
"learning_rate": 1.2824973806195955e-05,
"loss": 0.3006,
"step": 31725
},
{
"epoch": 3.6998600746268657,
"grad_norm": 0.4464073300376547,
"learning_rate": 1.2818398189030383e-05,
"loss": 0.3196,
"step": 31730
},
{
"epoch": 3.7004430970149254,
"grad_norm": 0.429807181591747,
"learning_rate": 1.2811824754878331e-05,
"loss": 0.326,
"step": 31735
},
{
"epoch": 3.701026119402985,
"grad_norm": 0.4507615629116169,
"learning_rate": 1.2805253504717213e-05,
"loss": 0.3188,
"step": 31740
},
{
"epoch": 3.7016091417910446,
"grad_norm": 0.45645840827116146,
"learning_rate": 1.27986844395241e-05,
"loss": 0.3308,
"step": 31745
},
{
"epoch": 3.7021921641791042,
"grad_norm": 0.4385297201540902,
"learning_rate": 1.2792117560275766e-05,
"loss": 0.3368,
"step": 31750
},
{
"epoch": 3.7027751865671643,
"grad_norm": 0.5116371698941201,
"learning_rate": 1.2785552867948625e-05,
"loss": 0.3425,
"step": 31755
},
{
"epoch": 3.703358208955224,
"grad_norm": 0.4744805080188085,
"learning_rate": 1.2778990363518785e-05,
"loss": 0.3261,
"step": 31760
},
{
"epoch": 3.7039412313432836,
"grad_norm": 0.4938922902251075,
"learning_rate": 1.277243004796204e-05,
"loss": 0.3517,
"step": 31765
},
{
"epoch": 3.704524253731343,
"grad_norm": 0.4846248190781154,
"learning_rate": 1.2765871922253835e-05,
"loss": 0.3265,
"step": 31770
},
{
"epoch": 3.705107276119403,
"grad_norm": 0.4366723589008353,
"learning_rate": 1.2759315987369306e-05,
"loss": 0.3361,
"step": 31775
},
{
"epoch": 3.705690298507463,
"grad_norm": 0.4675989131888052,
"learning_rate": 1.2752762244283255e-05,
"loss": 0.3484,
"step": 31780
},
{
"epoch": 3.7062733208955225,
"grad_norm": 0.46144740806748774,
"learning_rate": 1.2746210693970171e-05,
"loss": 0.3296,
"step": 31785
},
{
"epoch": 3.706856343283582,
"grad_norm": 0.49553855145425413,
"learning_rate": 1.2739661337404185e-05,
"loss": 0.331,
"step": 31790
},
{
"epoch": 3.707439365671642,
"grad_norm": 0.4482578805259561,
"learning_rate": 1.2733114175559149e-05,
"loss": 0.3293,
"step": 31795
},
{
"epoch": 3.7080223880597014,
"grad_norm": 0.4791044612725869,
"learning_rate": 1.2726569209408545e-05,
"loss": 0.3103,
"step": 31800
},
{
"epoch": 3.708605410447761,
"grad_norm": 0.4638006436199728,
"learning_rate": 1.272002643992555e-05,
"loss": 0.3319,
"step": 31805
},
{
"epoch": 3.7091884328358207,
"grad_norm": 0.4608440531145325,
"learning_rate": 1.2713485868083014e-05,
"loss": 0.3418,
"step": 31810
},
{
"epoch": 3.7097714552238807,
"grad_norm": 0.47277131107852466,
"learning_rate": 1.270694749485346e-05,
"loss": 0.3258,
"step": 31815
},
{
"epoch": 3.7103544776119404,
"grad_norm": 0.5124155258559975,
"learning_rate": 1.2700411321209078e-05,
"loss": 0.3298,
"step": 31820
},
{
"epoch": 3.7109375,
"grad_norm": 0.49050737819867396,
"learning_rate": 1.2693877348121742e-05,
"loss": 0.3475,
"step": 31825
},
{
"epoch": 3.7115205223880596,
"grad_norm": 0.46803011403512135,
"learning_rate": 1.2687345576562965e-05,
"loss": 0.3299,
"step": 31830
},
{
"epoch": 3.7121035447761193,
"grad_norm": 0.46483350384816363,
"learning_rate": 1.2680816007503998e-05,
"loss": 0.3118,
"step": 31835
},
{
"epoch": 3.7126865671641793,
"grad_norm": 0.42148401608250813,
"learning_rate": 1.2674288641915688e-05,
"loss": 0.304,
"step": 31840
},
{
"epoch": 3.713269589552239,
"grad_norm": 0.46259206012009374,
"learning_rate": 1.2667763480768625e-05,
"loss": 0.3316,
"step": 31845
},
{
"epoch": 3.7138526119402986,
"grad_norm": 0.48018614383398484,
"learning_rate": 1.2661240525033016e-05,
"loss": 0.3368,
"step": 31850
},
{
"epoch": 3.714435634328358,
"grad_norm": 0.4609350877561927,
"learning_rate": 1.2654719775678764e-05,
"loss": 0.339,
"step": 31855
},
{
"epoch": 3.715018656716418,
"grad_norm": 0.4546027197923379,
"learning_rate": 1.264820123367545e-05,
"loss": 0.3319,
"step": 31860
},
{
"epoch": 3.7156016791044775,
"grad_norm": 0.45717221896297866,
"learning_rate": 1.2641684899992314e-05,
"loss": 0.3372,
"step": 31865
},
{
"epoch": 3.716184701492537,
"grad_norm": 0.4670576032693017,
"learning_rate": 1.2635170775598271e-05,
"loss": 0.3186,
"step": 31870
},
{
"epoch": 3.716767723880597,
"grad_norm": 0.4617922031560612,
"learning_rate": 1.2628658861461923e-05,
"loss": 0.322,
"step": 31875
},
{
"epoch": 3.717350746268657,
"grad_norm": 0.48166851839959623,
"learning_rate": 1.26221491585515e-05,
"loss": 0.3331,
"step": 31880
},
{
"epoch": 3.7179337686567164,
"grad_norm": 0.4985631006094774,
"learning_rate": 1.2615641667834962e-05,
"loss": 0.3168,
"step": 31885
},
{
"epoch": 3.718516791044776,
"grad_norm": 0.519262460358683,
"learning_rate": 1.2609136390279886e-05,
"loss": 0.3537,
"step": 31890
},
{
"epoch": 3.7190998134328357,
"grad_norm": 0.47523741070448344,
"learning_rate": 1.2602633326853569e-05,
"loss": 0.3387,
"step": 31895
},
{
"epoch": 3.7196828358208958,
"grad_norm": 0.4900817921587001,
"learning_rate": 1.2596132478522938e-05,
"loss": 0.3295,
"step": 31900
},
{
"epoch": 3.7202658582089554,
"grad_norm": 0.4657727449280486,
"learning_rate": 1.258963384625461e-05,
"loss": 0.3437,
"step": 31905
},
{
"epoch": 3.720848880597015,
"grad_norm": 0.42124359394423505,
"learning_rate": 1.258313743101487e-05,
"loss": 0.3117,
"step": 31910
},
{
"epoch": 3.7214319029850746,
"grad_norm": 0.4599675295215162,
"learning_rate": 1.2576643233769669e-05,
"loss": 0.3389,
"step": 31915
},
{
"epoch": 3.7220149253731343,
"grad_norm": 0.46923324974649605,
"learning_rate": 1.2570151255484639e-05,
"loss": 0.3201,
"step": 31920
},
{
"epoch": 3.722597947761194,
"grad_norm": 0.4591247602003022,
"learning_rate": 1.2563661497125073e-05,
"loss": 0.3171,
"step": 31925
},
{
"epoch": 3.7231809701492535,
"grad_norm": 0.4339479858104873,
"learning_rate": 1.2557173959655932e-05,
"loss": 0.3247,
"step": 31930
},
{
"epoch": 3.7237639925373136,
"grad_norm": 0.49242388227375533,
"learning_rate": 1.2550688644041861e-05,
"loss": 0.3337,
"step": 31935
},
{
"epoch": 3.7243470149253732,
"grad_norm": 0.4409374650015577,
"learning_rate": 1.2544205551247148e-05,
"loss": 0.3198,
"step": 31940
},
{
"epoch": 3.724930037313433,
"grad_norm": 0.4929748884710064,
"learning_rate": 1.2537724682235776e-05,
"loss": 0.339,
"step": 31945
},
{
"epoch": 3.7255130597014925,
"grad_norm": 0.4542569726267021,
"learning_rate": 1.253124603797139e-05,
"loss": 0.3068,
"step": 31950
},
{
"epoch": 3.726096082089552,
"grad_norm": 0.47521964948638573,
"learning_rate": 1.2524769619417297e-05,
"loss": 0.3217,
"step": 31955
},
{
"epoch": 3.726679104477612,
"grad_norm": 0.46060947529613755,
"learning_rate": 1.251829542753648e-05,
"loss": 0.3367,
"step": 31960
},
{
"epoch": 3.7272621268656714,
"grad_norm": 0.46462461609170735,
"learning_rate": 1.2511823463291595e-05,
"loss": 0.3162,
"step": 31965
},
{
"epoch": 3.7278451492537314,
"grad_norm": 0.513210399503573,
"learning_rate": 1.2505353727644958e-05,
"loss": 0.3318,
"step": 31970
},
{
"epoch": 3.728428171641791,
"grad_norm": 0.4789985779106394,
"learning_rate": 1.2498886221558547e-05,
"loss": 0.3435,
"step": 31975
},
{
"epoch": 3.7290111940298507,
"grad_norm": 0.4860609060195966,
"learning_rate": 1.249242094599404e-05,
"loss": 0.3366,
"step": 31980
},
{
"epoch": 3.7295942164179103,
"grad_norm": 0.42586287306760157,
"learning_rate": 1.2485957901912736e-05,
"loss": 0.328,
"step": 31985
},
{
"epoch": 3.73017723880597,
"grad_norm": 0.46580715460925837,
"learning_rate": 1.2479497090275643e-05,
"loss": 0.3263,
"step": 31990
},
{
"epoch": 3.73076026119403,
"grad_norm": 0.46649738570614,
"learning_rate": 1.2473038512043417e-05,
"loss": 0.3414,
"step": 31995
},
{
"epoch": 3.7313432835820897,
"grad_norm": 0.47731033900291076,
"learning_rate": 1.246658216817639e-05,
"loss": 0.3403,
"step": 32000
},
{
"epoch": 3.7319263059701493,
"grad_norm": 0.4817616983967962,
"learning_rate": 1.2460128059634556e-05,
"loss": 0.3123,
"step": 32005
},
{
"epoch": 3.732509328358209,
"grad_norm": 0.5627787936908184,
"learning_rate": 1.2453676187377588e-05,
"loss": 0.3207,
"step": 32010
},
{
"epoch": 3.7330923507462686,
"grad_norm": 0.49174042244665306,
"learning_rate": 1.2447226552364792e-05,
"loss": 0.3363,
"step": 32015
},
{
"epoch": 3.7336753731343286,
"grad_norm": 0.4882290714936515,
"learning_rate": 1.2440779155555202e-05,
"loss": 0.3167,
"step": 32020
},
{
"epoch": 3.734258395522388,
"grad_norm": 0.45821859591672176,
"learning_rate": 1.2434333997907448e-05,
"loss": 0.316,
"step": 32025
},
{
"epoch": 3.734841417910448,
"grad_norm": 0.4181102718574339,
"learning_rate": 1.24278910803799e-05,
"loss": 0.3339,
"step": 32030
},
{
"epoch": 3.7354244402985075,
"grad_norm": 0.4701952700834935,
"learning_rate": 1.2421450403930529e-05,
"loss": 0.3516,
"step": 32035
},
{
"epoch": 3.736007462686567,
"grad_norm": 0.4469984835355792,
"learning_rate": 1.2415011969517016e-05,
"loss": 0.3227,
"step": 32040
},
{
"epoch": 3.7365904850746268,
"grad_norm": 0.4605442280735747,
"learning_rate": 1.2408575778096686e-05,
"loss": 0.3127,
"step": 32045
},
{
"epoch": 3.7371735074626864,
"grad_norm": 0.4756742820583785,
"learning_rate": 1.2402141830626547e-05,
"loss": 0.3315,
"step": 32050
},
{
"epoch": 3.7377565298507465,
"grad_norm": 0.4675689486501223,
"learning_rate": 1.2395710128063263e-05,
"loss": 0.3299,
"step": 32055
},
{
"epoch": 3.738339552238806,
"grad_norm": 0.5287087304801841,
"learning_rate": 1.2389280671363175e-05,
"loss": 0.3029,
"step": 32060
},
{
"epoch": 3.7389225746268657,
"grad_norm": 0.4764204681060229,
"learning_rate": 1.2382853461482255e-05,
"loss": 0.3203,
"step": 32065
},
{
"epoch": 3.7395055970149254,
"grad_norm": 0.47976503952454813,
"learning_rate": 1.2376428499376201e-05,
"loss": 0.3254,
"step": 32070
},
{
"epoch": 3.740088619402985,
"grad_norm": 0.44102137168488564,
"learning_rate": 1.2370005786000314e-05,
"loss": 0.3242,
"step": 32075
},
{
"epoch": 3.7406716417910446,
"grad_norm": 0.448509879831629,
"learning_rate": 1.2363585322309615e-05,
"loss": 0.328,
"step": 32080
},
{
"epoch": 3.7412546641791042,
"grad_norm": 0.4638741438273131,
"learning_rate": 1.2357167109258748e-05,
"loss": 0.327,
"step": 32085
},
{
"epoch": 3.7418376865671643,
"grad_norm": 0.457911526246863,
"learning_rate": 1.2350751147802047e-05,
"loss": 0.3355,
"step": 32090
},
{
"epoch": 3.742420708955224,
"grad_norm": 0.4553662582416192,
"learning_rate": 1.23443374388935e-05,
"loss": 0.3056,
"step": 32095
},
{
"epoch": 3.7430037313432836,
"grad_norm": 0.4553970154189041,
"learning_rate": 1.2337925983486768e-05,
"loss": 0.338,
"step": 32100
},
{
"epoch": 3.743586753731343,
"grad_norm": 0.47841457127273723,
"learning_rate": 1.2331516782535172e-05,
"loss": 0.3099,
"step": 32105
},
{
"epoch": 3.744169776119403,
"grad_norm": 0.4315875427584779,
"learning_rate": 1.2325109836991703e-05,
"loss": 0.3212,
"step": 32110
},
{
"epoch": 3.744752798507463,
"grad_norm": 0.5325946492349596,
"learning_rate": 1.2318705147809006e-05,
"loss": 0.343,
"step": 32115
},
{
"epoch": 3.7453358208955225,
"grad_norm": 0.483822559676552,
"learning_rate": 1.2312302715939394e-05,
"loss": 0.3036,
"step": 32120
},
{
"epoch": 3.745918843283582,
"grad_norm": 0.4743002706450139,
"learning_rate": 1.2305902542334854e-05,
"loss": 0.3088,
"step": 32125
},
{
"epoch": 3.746501865671642,
"grad_norm": 2.044759677569916,
"learning_rate": 1.2299504627947029e-05,
"loss": 0.3454,
"step": 32130
},
{
"epoch": 3.7470848880597014,
"grad_norm": 0.4217148784562264,
"learning_rate": 1.2293108973727224e-05,
"loss": 0.3337,
"step": 32135
},
{
"epoch": 3.747667910447761,
"grad_norm": 0.46320034017587036,
"learning_rate": 1.2286715580626418e-05,
"loss": 0.3375,
"step": 32140
},
{
"epoch": 3.7482509328358207,
"grad_norm": 0.4409040520149116,
"learning_rate": 1.228032444959525e-05,
"loss": 0.3284,
"step": 32145
},
{
"epoch": 3.7488339552238807,
"grad_norm": 0.5085748463935102,
"learning_rate": 1.2273935581584e-05,
"loss": 0.331,
"step": 32150
},
{
"epoch": 3.7494169776119404,
"grad_norm": 0.4331591632851916,
"learning_rate": 1.2267548977542656e-05,
"loss": 0.3183,
"step": 32155
},
{
"epoch": 3.75,
"grad_norm": 0.449574801438169,
"learning_rate": 1.2261164638420832e-05,
"loss": 0.3103,
"step": 32160
},
{
"epoch": 3.7505830223880596,
"grad_norm": 0.48238367955161604,
"learning_rate": 1.2254782565167817e-05,
"loss": 0.3207,
"step": 32165
},
{
"epoch": 3.7511660447761193,
"grad_norm": 0.47403035375974273,
"learning_rate": 1.2248402758732568e-05,
"loss": 0.3324,
"step": 32170
},
{
"epoch": 3.7517490671641793,
"grad_norm": 0.45266625798584026,
"learning_rate": 1.22420252200637e-05,
"loss": 0.3365,
"step": 32175
},
{
"epoch": 3.752332089552239,
"grad_norm": 0.46073069022823204,
"learning_rate": 1.2235649950109492e-05,
"loss": 0.319,
"step": 32180
},
{
"epoch": 3.7529151119402986,
"grad_norm": 0.48925534472831805,
"learning_rate": 1.2229276949817894e-05,
"loss": 0.3276,
"step": 32185
},
{
"epoch": 3.753498134328358,
"grad_norm": 0.4443209482708789,
"learning_rate": 1.222290622013649e-05,
"loss": 0.3184,
"step": 32190
},
{
"epoch": 3.754081156716418,
"grad_norm": 0.5255221148089022,
"learning_rate": 1.221653776201257e-05,
"loss": 0.341,
"step": 32195
},
{
"epoch": 3.7546641791044775,
"grad_norm": 0.4793235964058894,
"learning_rate": 1.2210171576393037e-05,
"loss": 0.3383,
"step": 32200
},
{
"epoch": 3.755247201492537,
"grad_norm": 0.4938430126386554,
"learning_rate": 1.220380766422451e-05,
"loss": 0.3197,
"step": 32205
},
{
"epoch": 3.755830223880597,
"grad_norm": 0.4765637872045808,
"learning_rate": 1.2197446026453219e-05,
"loss": 0.3262,
"step": 32210
},
{
"epoch": 3.756413246268657,
"grad_norm": 0.4683225388276627,
"learning_rate": 1.219108666402509e-05,
"loss": 0.3651,
"step": 32215
},
{
"epoch": 3.7569962686567164,
"grad_norm": 0.45420723492269827,
"learning_rate": 1.2184729577885695e-05,
"loss": 0.3377,
"step": 32220
},
{
"epoch": 3.757579291044776,
"grad_norm": 0.43816465509182606,
"learning_rate": 1.2178374768980275e-05,
"loss": 0.3232,
"step": 32225
},
{
"epoch": 3.7581623134328357,
"grad_norm": 0.44977683835597543,
"learning_rate": 1.2172022238253727e-05,
"loss": 0.3268,
"step": 32230
},
{
"epoch": 3.7587453358208958,
"grad_norm": 0.4541173885854834,
"learning_rate": 1.2165671986650618e-05,
"loss": 0.3276,
"step": 32235
},
{
"epoch": 3.7593283582089554,
"grad_norm": 0.5592475270580155,
"learning_rate": 1.2159324015115148e-05,
"loss": 0.3519,
"step": 32240
},
{
"epoch": 3.759911380597015,
"grad_norm": 0.44534625705153374,
"learning_rate": 1.2152978324591233e-05,
"loss": 0.3384,
"step": 32245
},
{
"epoch": 3.7604944029850746,
"grad_norm": 0.44667025876219807,
"learning_rate": 1.2146634916022383e-05,
"loss": 0.3404,
"step": 32250
},
{
"epoch": 3.7610774253731343,
"grad_norm": 0.43161392006810906,
"learning_rate": 1.214029379035183e-05,
"loss": 0.3141,
"step": 32255
},
{
"epoch": 3.761660447761194,
"grad_norm": 0.4984373533108043,
"learning_rate": 1.2133954948522423e-05,
"loss": 0.343,
"step": 32260
},
{
"epoch": 3.7622434701492535,
"grad_norm": 0.4875573311512853,
"learning_rate": 1.2127618391476688e-05,
"loss": 0.3329,
"step": 32265
},
{
"epoch": 3.7628264925373136,
"grad_norm": 0.4676666518858329,
"learning_rate": 1.2121284120156812e-05,
"loss": 0.3124,
"step": 32270
},
{
"epoch": 3.7634095149253732,
"grad_norm": 0.46558430530465317,
"learning_rate": 1.2114952135504642e-05,
"loss": 0.3157,
"step": 32275
},
{
"epoch": 3.763992537313433,
"grad_norm": 0.444194527943588,
"learning_rate": 1.210862243846168e-05,
"loss": 0.3348,
"step": 32280
},
{
"epoch": 3.7645755597014925,
"grad_norm": 0.4774140083616483,
"learning_rate": 1.2102295029969104e-05,
"loss": 0.3334,
"step": 32285
},
{
"epoch": 3.765158582089552,
"grad_norm": 0.4887350469136481,
"learning_rate": 1.209596991096772e-05,
"loss": 0.3266,
"step": 32290
},
{
"epoch": 3.765741604477612,
"grad_norm": 0.4697718880107236,
"learning_rate": 1.2089647082398022e-05,
"loss": 0.3297,
"step": 32295
},
{
"epoch": 3.7663246268656714,
"grad_norm": 0.48329095287260077,
"learning_rate": 1.2083326545200154e-05,
"loss": 0.3494,
"step": 32300
},
{
"epoch": 3.7669076492537314,
"grad_norm": 0.4906534119761042,
"learning_rate": 1.2077008300313915e-05,
"loss": 0.3346,
"step": 32305
},
{
"epoch": 3.767490671641791,
"grad_norm": 0.459370468202657,
"learning_rate": 1.2070692348678776e-05,
"loss": 0.3137,
"step": 32310
},
{
"epoch": 3.7680736940298507,
"grad_norm": 0.41422423338049147,
"learning_rate": 1.2064378691233851e-05,
"loss": 0.3051,
"step": 32315
},
{
"epoch": 3.7686567164179103,
"grad_norm": 0.5547554233691531,
"learning_rate": 1.205806732891793e-05,
"loss": 0.3369,
"step": 32320
},
{
"epoch": 3.76923973880597,
"grad_norm": 0.44835155064718296,
"learning_rate": 1.205175826266943e-05,
"loss": 0.3165,
"step": 32325
},
{
"epoch": 3.76982276119403,
"grad_norm": 0.44559403853101776,
"learning_rate": 1.2045451493426483e-05,
"loss": 0.3352,
"step": 32330
},
{
"epoch": 3.7704057835820897,
"grad_norm": 0.4433241737057994,
"learning_rate": 1.2039147022126815e-05,
"loss": 0.3137,
"step": 32335
},
{
"epoch": 3.7709888059701493,
"grad_norm": 0.4367302619279283,
"learning_rate": 1.2032844849707853e-05,
"loss": 0.3003,
"step": 32340
},
{
"epoch": 3.771571828358209,
"grad_norm": 0.44623453743128016,
"learning_rate": 1.2026544977106669e-05,
"loss": 0.3216,
"step": 32345
},
{
"epoch": 3.7721548507462686,
"grad_norm": 0.47066659887948226,
"learning_rate": 1.202024740525999e-05,
"loss": 0.3343,
"step": 32350
},
{
"epoch": 3.7727378731343286,
"grad_norm": 0.4729895133636887,
"learning_rate": 1.2013952135104209e-05,
"loss": 0.3244,
"step": 32355
},
{
"epoch": 3.773320895522388,
"grad_norm": 0.42145515561435953,
"learning_rate": 1.2007659167575377e-05,
"loss": 0.3002,
"step": 32360
},
{
"epoch": 3.773903917910448,
"grad_norm": 0.4740490777351826,
"learning_rate": 1.2001368503609179e-05,
"loss": 0.3385,
"step": 32365
},
{
"epoch": 3.7744869402985075,
"grad_norm": 0.4311640262971568,
"learning_rate": 1.1995080144141004e-05,
"loss": 0.3092,
"step": 32370
},
{
"epoch": 3.775069962686567,
"grad_norm": 0.43809698981656,
"learning_rate": 1.1988794090105842e-05,
"loss": 0.3111,
"step": 32375
},
{
"epoch": 3.7756529850746268,
"grad_norm": 0.47687228833874734,
"learning_rate": 1.1982510342438395e-05,
"loss": 0.3334,
"step": 32380
},
{
"epoch": 3.7762360074626864,
"grad_norm": 0.4240495396943132,
"learning_rate": 1.1976228902072981e-05,
"loss": 0.3063,
"step": 32385
},
{
"epoch": 3.7768190298507465,
"grad_norm": 0.47655406701332764,
"learning_rate": 1.1969949769943587e-05,
"loss": 0.3324,
"step": 32390
},
{
"epoch": 3.777402052238806,
"grad_norm": 0.466356782032414,
"learning_rate": 1.196367294698387e-05,
"loss": 0.3244,
"step": 32395
},
{
"epoch": 3.7779850746268657,
"grad_norm": 0.48898401058095925,
"learning_rate": 1.195739843412713e-05,
"loss": 0.3292,
"step": 32400
},
{
"epoch": 3.7785680970149254,
"grad_norm": 0.45772056201959277,
"learning_rate": 1.1951126232306325e-05,
"loss": 0.3227,
"step": 32405
},
{
"epoch": 3.779151119402985,
"grad_norm": 0.4373854684473265,
"learning_rate": 1.1944856342454078e-05,
"loss": 0.3375,
"step": 32410
},
{
"epoch": 3.7797341417910446,
"grad_norm": 0.44894386778120476,
"learning_rate": 1.1938588765502643e-05,
"loss": 0.3385,
"step": 32415
},
{
"epoch": 3.7803171641791042,
"grad_norm": 0.5119645475216401,
"learning_rate": 1.1932323502383978e-05,
"loss": 0.3209,
"step": 32420
},
{
"epoch": 3.7809001865671643,
"grad_norm": 0.4764171671627143,
"learning_rate": 1.1926060554029636e-05,
"loss": 0.3235,
"step": 32425
},
{
"epoch": 3.781483208955224,
"grad_norm": 0.5176406273951388,
"learning_rate": 1.1919799921370888e-05,
"loss": 0.3365,
"step": 32430
},
{
"epoch": 3.7820662313432836,
"grad_norm": 0.5064991390036546,
"learning_rate": 1.1913541605338605e-05,
"loss": 0.3134,
"step": 32435
},
{
"epoch": 3.782649253731343,
"grad_norm": 0.49993873940907857,
"learning_rate": 1.1907285606863351e-05,
"loss": 0.3226,
"step": 32440
},
{
"epoch": 3.783232276119403,
"grad_norm": 0.5135022484796244,
"learning_rate": 1.190103192687533e-05,
"loss": 0.3342,
"step": 32445
},
{
"epoch": 3.783815298507463,
"grad_norm": 0.4597044748870591,
"learning_rate": 1.1894780566304406e-05,
"loss": 0.3255,
"step": 32450
},
{
"epoch": 3.7843983208955225,
"grad_norm": 0.46585929236718493,
"learning_rate": 1.1888531526080095e-05,
"loss": 0.3458,
"step": 32455
},
{
"epoch": 3.784981343283582,
"grad_norm": 0.4673428208687415,
"learning_rate": 1.1882284807131576e-05,
"loss": 0.3338,
"step": 32460
},
{
"epoch": 3.785564365671642,
"grad_norm": 0.43258356419151617,
"learning_rate": 1.1876040410387653e-05,
"loss": 0.318,
"step": 32465
},
{
"epoch": 3.7861473880597014,
"grad_norm": 0.41681281349956106,
"learning_rate": 1.1869798336776845e-05,
"loss": 0.3294,
"step": 32470
},
{
"epoch": 3.786730410447761,
"grad_norm": 0.4674194498301845,
"learning_rate": 1.1863558587227256e-05,
"loss": 0.3152,
"step": 32475
},
{
"epoch": 3.7873134328358207,
"grad_norm": 0.4513428447564988,
"learning_rate": 1.1857321162666692e-05,
"loss": 0.3341,
"step": 32480
},
{
"epoch": 3.7878964552238807,
"grad_norm": 0.42564313689155603,
"learning_rate": 1.1851086064022596e-05,
"loss": 0.329,
"step": 32485
},
{
"epoch": 3.7884794776119404,
"grad_norm": 0.43746810166454736,
"learning_rate": 1.1844853292222066e-05,
"loss": 0.3249,
"step": 32490
},
{
"epoch": 3.7890625,
"grad_norm": 0.45328960422956577,
"learning_rate": 1.1838622848191857e-05,
"loss": 0.331,
"step": 32495
},
{
"epoch": 3.7896455223880596,
"grad_norm": 0.44084270016519495,
"learning_rate": 1.1832394732858377e-05,
"loss": 0.2871,
"step": 32500
},
{
"epoch": 3.7902285447761193,
"grad_norm": 0.47713810295263487,
"learning_rate": 1.1826168947147693e-05,
"loss": 0.3416,
"step": 32505
},
{
"epoch": 3.7908115671641793,
"grad_norm": 0.4655338329444578,
"learning_rate": 1.1819945491985504e-05,
"loss": 0.3449,
"step": 32510
},
{
"epoch": 3.791394589552239,
"grad_norm": 0.488389264161232,
"learning_rate": 1.1813724368297187e-05,
"loss": 0.3268,
"step": 32515
},
{
"epoch": 3.7919776119402986,
"grad_norm": 0.46751089493196396,
"learning_rate": 1.1807505577007765e-05,
"loss": 0.3333,
"step": 32520
},
{
"epoch": 3.792560634328358,
"grad_norm": 0.48137074011958036,
"learning_rate": 1.1801289119041909e-05,
"loss": 0.3356,
"step": 32525
},
{
"epoch": 3.793143656716418,
"grad_norm": 0.4797731430916417,
"learning_rate": 1.179507499532395e-05,
"loss": 0.3448,
"step": 32530
},
{
"epoch": 3.7937266791044775,
"grad_norm": 0.48443341575171117,
"learning_rate": 1.1788863206777875e-05,
"loss": 0.3379,
"step": 32535
},
{
"epoch": 3.794309701492537,
"grad_norm": 0.43114955714166386,
"learning_rate": 1.1782653754327295e-05,
"loss": 0.3249,
"step": 32540
},
{
"epoch": 3.794892723880597,
"grad_norm": 0.5020365866277783,
"learning_rate": 1.1776446638895523e-05,
"loss": 0.3399,
"step": 32545
},
{
"epoch": 3.795475746268657,
"grad_norm": 0.45577173307467767,
"learning_rate": 1.1770241861405475e-05,
"loss": 0.3126,
"step": 32550
},
{
"epoch": 3.7960587686567164,
"grad_norm": 0.4358709337137338,
"learning_rate": 1.1764039422779765e-05,
"loss": 0.3233,
"step": 32555
},
{
"epoch": 3.796641791044776,
"grad_norm": 0.4940993977369946,
"learning_rate": 1.1757839323940616e-05,
"loss": 0.331,
"step": 32560
},
{
"epoch": 3.7972248134328357,
"grad_norm": 0.4909816286910687,
"learning_rate": 1.1751641565809931e-05,
"loss": 0.3388,
"step": 32565
},
{
"epoch": 3.7978078358208958,
"grad_norm": 0.4524352131225283,
"learning_rate": 1.1745446149309257e-05,
"loss": 0.3031,
"step": 32570
},
{
"epoch": 3.7983908582089554,
"grad_norm": 0.4580994040900444,
"learning_rate": 1.1739253075359796e-05,
"loss": 0.3296,
"step": 32575
},
{
"epoch": 3.798973880597015,
"grad_norm": 0.4580068147597908,
"learning_rate": 1.1733062344882396e-05,
"loss": 0.3531,
"step": 32580
},
{
"epoch": 3.7995569029850746,
"grad_norm": 0.46620247870181725,
"learning_rate": 1.1726873958797565e-05,
"loss": 0.3321,
"step": 32585
},
{
"epoch": 3.8001399253731343,
"grad_norm": 0.5971353874873305,
"learning_rate": 1.1720687918025434e-05,
"loss": 0.3126,
"step": 32590
},
{
"epoch": 3.800722947761194,
"grad_norm": 0.4390347678214194,
"learning_rate": 1.1714504223485844e-05,
"loss": 0.3202,
"step": 32595
},
{
"epoch": 3.8013059701492535,
"grad_norm": 0.48452309368607444,
"learning_rate": 1.1708322876098215e-05,
"loss": 0.3277,
"step": 32600
},
{
"epoch": 3.8018889925373136,
"grad_norm": 0.4593093720055001,
"learning_rate": 1.1702143876781687e-05,
"loss": 0.3377,
"step": 32605
},
{
"epoch": 3.8024720149253732,
"grad_norm": 0.42618677367571434,
"learning_rate": 1.1695967226454996e-05,
"loss": 0.3168,
"step": 32610
},
{
"epoch": 3.803055037313433,
"grad_norm": 0.4414796121902606,
"learning_rate": 1.1689792926036555e-05,
"loss": 0.3271,
"step": 32615
},
{
"epoch": 3.8036380597014925,
"grad_norm": 0.4733229542104905,
"learning_rate": 1.1683620976444426e-05,
"loss": 0.3463,
"step": 32620
},
{
"epoch": 3.804221082089552,
"grad_norm": 0.46288076213503515,
"learning_rate": 1.1677451378596318e-05,
"loss": 0.3546,
"step": 32625
},
{
"epoch": 3.804804104477612,
"grad_norm": 0.44912203881342816,
"learning_rate": 1.1671284133409592e-05,
"loss": 0.3125,
"step": 32630
},
{
"epoch": 3.8053871268656714,
"grad_norm": 0.4297942344755935,
"learning_rate": 1.1665119241801257e-05,
"loss": 0.3278,
"step": 32635
},
{
"epoch": 3.8059701492537314,
"grad_norm": 0.41257148676184574,
"learning_rate": 1.1658956704687974e-05,
"loss": 0.3151,
"step": 32640
},
{
"epoch": 3.806553171641791,
"grad_norm": 0.454786923252004,
"learning_rate": 1.165279652298606e-05,
"loss": 0.3416,
"step": 32645
},
{
"epoch": 3.8071361940298507,
"grad_norm": 0.4857749800750693,
"learning_rate": 1.1646638697611462e-05,
"loss": 0.3351,
"step": 32650
},
{
"epoch": 3.8077192164179103,
"grad_norm": 0.45067500520051373,
"learning_rate": 1.1640483229479794e-05,
"loss": 0.3439,
"step": 32655
},
{
"epoch": 3.80830223880597,
"grad_norm": 0.4823448438799639,
"learning_rate": 1.1634330119506317e-05,
"loss": 0.3296,
"step": 32660
},
{
"epoch": 3.80888526119403,
"grad_norm": 0.47307038382014316,
"learning_rate": 1.162817936860594e-05,
"loss": 0.3521,
"step": 32665
},
{
"epoch": 3.8094682835820897,
"grad_norm": 0.44750460757509564,
"learning_rate": 1.1622030977693221e-05,
"loss": 0.3021,
"step": 32670
},
{
"epoch": 3.8100513059701493,
"grad_norm": 0.4693317663106042,
"learning_rate": 1.1615884947682364e-05,
"loss": 0.34,
"step": 32675
},
{
"epoch": 3.810634328358209,
"grad_norm": 0.4615728105995682,
"learning_rate": 1.1609741279487236e-05,
"loss": 0.3472,
"step": 32680
},
{
"epoch": 3.8112173507462686,
"grad_norm": 0.4486343742363925,
"learning_rate": 1.1603599974021317e-05,
"loss": 0.3005,
"step": 32685
},
{
"epoch": 3.8118003731343286,
"grad_norm": 0.5027177392080457,
"learning_rate": 1.1597461032197788e-05,
"loss": 0.3401,
"step": 32690
},
{
"epoch": 3.812383395522388,
"grad_norm": 0.44719318802271446,
"learning_rate": 1.1591324454929433e-05,
"loss": 0.3341,
"step": 32695
},
{
"epoch": 3.812966417910448,
"grad_norm": 0.45142182524978125,
"learning_rate": 1.1585190243128707e-05,
"loss": 0.3241,
"step": 32700
},
{
"epoch": 3.8135494402985075,
"grad_norm": 0.4769584469070269,
"learning_rate": 1.1579058397707707e-05,
"loss": 0.3164,
"step": 32705
},
{
"epoch": 3.814132462686567,
"grad_norm": 0.4761466214469882,
"learning_rate": 1.1572928919578186e-05,
"loss": 0.3173,
"step": 32710
},
{
"epoch": 3.8147154850746268,
"grad_norm": 0.43987912155833664,
"learning_rate": 1.1566801809651532e-05,
"loss": 0.3259,
"step": 32715
},
{
"epoch": 3.8152985074626864,
"grad_norm": 0.44197945084467494,
"learning_rate": 1.15606770688388e-05,
"loss": 0.3343,
"step": 32720
},
{
"epoch": 3.8158815298507465,
"grad_norm": 0.490769456821821,
"learning_rate": 1.1554554698050652e-05,
"loss": 0.3221,
"step": 32725
},
{
"epoch": 3.816464552238806,
"grad_norm": 0.4370166373735806,
"learning_rate": 1.154843469819746e-05,
"loss": 0.3146,
"step": 32730
},
{
"epoch": 3.8170475746268657,
"grad_norm": 0.4094388795690992,
"learning_rate": 1.1542317070189181e-05,
"loss": 0.3257,
"step": 32735
},
{
"epoch": 3.8176305970149254,
"grad_norm": 0.5294554908674985,
"learning_rate": 1.1536201814935473e-05,
"loss": 0.2999,
"step": 32740
},
{
"epoch": 3.818213619402985,
"grad_norm": 0.3973754804614025,
"learning_rate": 1.1530088933345595e-05,
"loss": 0.3059,
"step": 32745
},
{
"epoch": 3.8187966417910446,
"grad_norm": 0.4460908479567948,
"learning_rate": 1.152397842632848e-05,
"loss": 0.3399,
"step": 32750
},
{
"epoch": 3.8193796641791042,
"grad_norm": 0.45097839645187715,
"learning_rate": 1.1517870294792709e-05,
"loss": 0.3293,
"step": 32755
},
{
"epoch": 3.8199626865671643,
"grad_norm": 0.4501929828562514,
"learning_rate": 1.1511764539646494e-05,
"loss": 0.3331,
"step": 32760
},
{
"epoch": 3.820545708955224,
"grad_norm": 0.4679727027485936,
"learning_rate": 1.1505661161797707e-05,
"loss": 0.315,
"step": 32765
},
{
"epoch": 3.8211287313432836,
"grad_norm": 0.47639889209726183,
"learning_rate": 1.1499560162153866e-05,
"loss": 0.3297,
"step": 32770
},
{
"epoch": 3.821711753731343,
"grad_norm": 0.5031297037643795,
"learning_rate": 1.1493461541622114e-05,
"loss": 0.3245,
"step": 32775
},
{
"epoch": 3.822294776119403,
"grad_norm": 0.5005275814073943,
"learning_rate": 1.1487365301109281e-05,
"loss": 0.3494,
"step": 32780
},
{
"epoch": 3.822877798507463,
"grad_norm": 0.4678153501476471,
"learning_rate": 1.1481271441521796e-05,
"loss": 0.3315,
"step": 32785
},
{
"epoch": 3.8234608208955225,
"grad_norm": 0.47138667915063925,
"learning_rate": 1.147517996376578e-05,
"loss": 0.3147,
"step": 32790
},
{
"epoch": 3.824043843283582,
"grad_norm": 0.4446152429445292,
"learning_rate": 1.1469090868746961e-05,
"loss": 0.3107,
"step": 32795
},
{
"epoch": 3.824626865671642,
"grad_norm": 0.4628791326023051,
"learning_rate": 1.1463004157370735e-05,
"loss": 0.3368,
"step": 32800
},
{
"epoch": 3.8252098880597014,
"grad_norm": 0.47672171053911006,
"learning_rate": 1.1456919830542138e-05,
"loss": 0.3302,
"step": 32805
},
{
"epoch": 3.825792910447761,
"grad_norm": 0.4833603423512201,
"learning_rate": 1.1450837889165852e-05,
"loss": 0.3266,
"step": 32810
},
{
"epoch": 3.8263759328358207,
"grad_norm": 0.5082904135940178,
"learning_rate": 1.14447583341462e-05,
"loss": 0.3489,
"step": 32815
},
{
"epoch": 3.8269589552238807,
"grad_norm": 0.4746803078733811,
"learning_rate": 1.1438681166387162e-05,
"loss": 0.3418,
"step": 32820
},
{
"epoch": 3.8275419776119404,
"grad_norm": 0.46944595952476803,
"learning_rate": 1.1432606386792341e-05,
"loss": 0.3126,
"step": 32825
},
{
"epoch": 3.828125,
"grad_norm": 0.45320397573807447,
"learning_rate": 1.1426533996265008e-05,
"loss": 0.3284,
"step": 32830
},
{
"epoch": 3.8287080223880596,
"grad_norm": 0.4815212010253675,
"learning_rate": 1.1420463995708064e-05,
"loss": 0.3301,
"step": 32835
},
{
"epoch": 3.8292910447761193,
"grad_norm": 0.4455388520982315,
"learning_rate": 1.1414396386024064e-05,
"loss": 0.2994,
"step": 32840
},
{
"epoch": 3.8298740671641793,
"grad_norm": 0.5022441632403734,
"learning_rate": 1.1408331168115203e-05,
"loss": 0.3471,
"step": 32845
},
{
"epoch": 3.830457089552239,
"grad_norm": 0.487466620138546,
"learning_rate": 1.140226834288332e-05,
"loss": 0.3252,
"step": 32850
},
{
"epoch": 3.8310401119402986,
"grad_norm": 0.4577422473409993,
"learning_rate": 1.1396207911229903e-05,
"loss": 0.3442,
"step": 32855
},
{
"epoch": 3.831623134328358,
"grad_norm": 0.4097207699401092,
"learning_rate": 1.1390149874056065e-05,
"loss": 0.3358,
"step": 32860
},
{
"epoch": 3.832206156716418,
"grad_norm": 0.4877820747731223,
"learning_rate": 1.1384094232262602e-05,
"loss": 0.3215,
"step": 32865
},
{
"epoch": 3.8327891791044775,
"grad_norm": 0.4833693289223179,
"learning_rate": 1.1378040986749912e-05,
"loss": 0.3364,
"step": 32870
},
{
"epoch": 3.833372201492537,
"grad_norm": 0.461400786860475,
"learning_rate": 1.137199013841806e-05,
"loss": 0.3283,
"step": 32875
},
{
"epoch": 3.833955223880597,
"grad_norm": 0.47158608400881774,
"learning_rate": 1.1365941688166747e-05,
"loss": 0.3344,
"step": 32880
},
{
"epoch": 3.834538246268657,
"grad_norm": 0.4680783563143886,
"learning_rate": 1.1359895636895324e-05,
"loss": 0.3336,
"step": 32885
},
{
"epoch": 3.8351212686567164,
"grad_norm": 0.45279641518224417,
"learning_rate": 1.1353851985502777e-05,
"loss": 0.3208,
"step": 32890
},
{
"epoch": 3.835704291044776,
"grad_norm": 0.4574662744031216,
"learning_rate": 1.1347810734887747e-05,
"loss": 0.3305,
"step": 32895
},
{
"epoch": 3.8362873134328357,
"grad_norm": 0.4328061385653668,
"learning_rate": 1.134177188594849e-05,
"loss": 0.3209,
"step": 32900
},
{
"epoch": 3.8368703358208958,
"grad_norm": 0.4886698936922826,
"learning_rate": 1.1335735439582952e-05,
"loss": 0.3554,
"step": 32905
},
{
"epoch": 3.8374533582089554,
"grad_norm": 0.48878378205522216,
"learning_rate": 1.1329701396688669e-05,
"loss": 0.3381,
"step": 32910
},
{
"epoch": 3.838036380597015,
"grad_norm": 0.4803671045616649,
"learning_rate": 1.132366975816287e-05,
"loss": 0.3268,
"step": 32915
},
{
"epoch": 3.8386194029850746,
"grad_norm": 0.503144613393508,
"learning_rate": 1.1317640524902383e-05,
"loss": 0.3464,
"step": 32920
},
{
"epoch": 3.8392024253731343,
"grad_norm": 0.4587130891790112,
"learning_rate": 1.1311613697803703e-05,
"loss": 0.3239,
"step": 32925
},
{
"epoch": 3.839785447761194,
"grad_norm": 0.4539351130193337,
"learning_rate": 1.1305589277762965e-05,
"loss": 0.3472,
"step": 32930
},
{
"epoch": 3.8403684701492535,
"grad_norm": 0.450259232588213,
"learning_rate": 1.1299567265675939e-05,
"loss": 0.3076,
"step": 32935
},
{
"epoch": 3.8409514925373136,
"grad_norm": 0.42848948841402396,
"learning_rate": 1.129354766243804e-05,
"loss": 0.3174,
"step": 32940
},
{
"epoch": 3.8415345149253732,
"grad_norm": 0.4619131535601183,
"learning_rate": 1.1287530468944332e-05,
"loss": 0.3107,
"step": 32945
},
{
"epoch": 3.842117537313433,
"grad_norm": 0.5382638189612068,
"learning_rate": 1.1281515686089497e-05,
"loss": 0.3219,
"step": 32950
},
{
"epoch": 3.8427005597014925,
"grad_norm": 0.4759530067107466,
"learning_rate": 1.1275503314767901e-05,
"loss": 0.3239,
"step": 32955
},
{
"epoch": 3.843283582089552,
"grad_norm": 0.46973304991536674,
"learning_rate": 1.1269493355873498e-05,
"loss": 0.3408,
"step": 32960
},
{
"epoch": 3.843866604477612,
"grad_norm": 0.44085510836492364,
"learning_rate": 1.126348581029994e-05,
"loss": 0.3293,
"step": 32965
},
{
"epoch": 3.8444496268656714,
"grad_norm": 0.449990620750623,
"learning_rate": 1.1257480678940469e-05,
"loss": 0.3026,
"step": 32970
},
{
"epoch": 3.8450326492537314,
"grad_norm": 0.4514383287544397,
"learning_rate": 1.1251477962687998e-05,
"loss": 0.3115,
"step": 32975
},
{
"epoch": 3.845615671641791,
"grad_norm": 0.43436002421258607,
"learning_rate": 1.1245477662435076e-05,
"loss": 0.3152,
"step": 32980
},
{
"epoch": 3.8461986940298507,
"grad_norm": 0.48597649213632177,
"learning_rate": 1.1239479779073885e-05,
"loss": 0.3169,
"step": 32985
},
{
"epoch": 3.8467817164179103,
"grad_norm": 0.44564284318614067,
"learning_rate": 1.123348431349626e-05,
"loss": 0.3332,
"step": 32990
},
{
"epoch": 3.84736473880597,
"grad_norm": 0.4102286231617167,
"learning_rate": 1.1227491266593668e-05,
"loss": 0.3261,
"step": 32995
},
{
"epoch": 3.84794776119403,
"grad_norm": 0.4651315474062523,
"learning_rate": 1.1221500639257204e-05,
"loss": 0.3179,
"step": 33000
},
{
"epoch": 3.8485307835820897,
"grad_norm": 0.4872571147764733,
"learning_rate": 1.121551243237764e-05,
"loss": 0.3287,
"step": 33005
},
{
"epoch": 3.8491138059701493,
"grad_norm": 0.4675344374427744,
"learning_rate": 1.1209526646845346e-05,
"loss": 0.3345,
"step": 33010
},
{
"epoch": 3.849696828358209,
"grad_norm": 0.4400261379252843,
"learning_rate": 1.1203543283550355e-05,
"loss": 0.3108,
"step": 33015
},
{
"epoch": 3.8502798507462686,
"grad_norm": 0.4430965742244254,
"learning_rate": 1.1197562343382341e-05,
"loss": 0.3395,
"step": 33020
},
{
"epoch": 3.8508628731343286,
"grad_norm": 0.45735532099931,
"learning_rate": 1.119158382723061e-05,
"loss": 0.3278,
"step": 33025
},
{
"epoch": 3.851445895522388,
"grad_norm": 0.4687951359586848,
"learning_rate": 1.118560773598411e-05,
"loss": 0.3231,
"step": 33030
},
{
"epoch": 3.852028917910448,
"grad_norm": 0.46140836800880514,
"learning_rate": 1.1179634070531427e-05,
"loss": 0.3398,
"step": 33035
},
{
"epoch": 3.8526119402985075,
"grad_norm": 0.4615435808577384,
"learning_rate": 1.1173662831760798e-05,
"loss": 0.336,
"step": 33040
},
{
"epoch": 3.853194962686567,
"grad_norm": 0.498488469657942,
"learning_rate": 1.1167694020560071e-05,
"loss": 0.3461,
"step": 33045
},
{
"epoch": 3.8537779850746268,
"grad_norm": 0.5473689760854433,
"learning_rate": 1.1161727637816762e-05,
"loss": 0.3294,
"step": 33050
},
{
"epoch": 3.8543610074626864,
"grad_norm": 0.4668962224968966,
"learning_rate": 1.1155763684418013e-05,
"loss": 0.3426,
"step": 33055
},
{
"epoch": 3.8549440298507465,
"grad_norm": 0.4405721965189786,
"learning_rate": 1.1149802161250607e-05,
"loss": 0.3221,
"step": 33060
},
{
"epoch": 3.855527052238806,
"grad_norm": 0.47858635454662596,
"learning_rate": 1.1143843069200965e-05,
"loss": 0.323,
"step": 33065
},
{
"epoch": 3.8561100746268657,
"grad_norm": 0.43614195310448467,
"learning_rate": 1.1137886409155158e-05,
"loss": 0.3183,
"step": 33070
},
{
"epoch": 3.8566930970149254,
"grad_norm": 0.5362034447270819,
"learning_rate": 1.1131932181998856e-05,
"loss": 0.3537,
"step": 33075
},
{
"epoch": 3.857276119402985,
"grad_norm": 0.4455931691239378,
"learning_rate": 1.1125980388617425e-05,
"loss": 0.3312,
"step": 33080
},
{
"epoch": 3.8578591417910446,
"grad_norm": 0.46523903730440463,
"learning_rate": 1.1120031029895816e-05,
"loss": 0.3251,
"step": 33085
},
{
"epoch": 3.8584421641791042,
"grad_norm": 0.5282719718396226,
"learning_rate": 1.1114084106718667e-05,
"loss": 0.3405,
"step": 33090
},
{
"epoch": 3.8590251865671643,
"grad_norm": 0.45428995715896703,
"learning_rate": 1.1108139619970207e-05,
"loss": 0.3399,
"step": 33095
},
{
"epoch": 3.859608208955224,
"grad_norm": 0.4469221003103343,
"learning_rate": 1.1102197570534334e-05,
"loss": 0.3184,
"step": 33100
},
{
"epoch": 3.8601912313432836,
"grad_norm": 0.4305500784793551,
"learning_rate": 1.1096257959294572e-05,
"loss": 0.3296,
"step": 33105
},
{
"epoch": 3.860774253731343,
"grad_norm": 0.49759181520564827,
"learning_rate": 1.1090320787134085e-05,
"loss": 0.3293,
"step": 33110
},
{
"epoch": 3.861357276119403,
"grad_norm": 0.48592985818559,
"learning_rate": 1.1084386054935669e-05,
"loss": 0.3186,
"step": 33115
},
{
"epoch": 3.861940298507463,
"grad_norm": 0.45724255869740565,
"learning_rate": 1.1078453763581776e-05,
"loss": 0.321,
"step": 33120
},
{
"epoch": 3.8625233208955225,
"grad_norm": 0.459645642938318,
"learning_rate": 1.1072523913954455e-05,
"loss": 0.3245,
"step": 33125
},
{
"epoch": 3.863106343283582,
"grad_norm": 0.5223399497373401,
"learning_rate": 1.1066596506935447e-05,
"loss": 0.3346,
"step": 33130
},
{
"epoch": 3.863689365671642,
"grad_norm": 0.47712012114642316,
"learning_rate": 1.1060671543406074e-05,
"loss": 0.327,
"step": 33135
},
{
"epoch": 3.8642723880597014,
"grad_norm": 0.48271037227275715,
"learning_rate": 1.1054749024247348e-05,
"loss": 0.3324,
"step": 33140
},
{
"epoch": 3.864855410447761,
"grad_norm": 0.4315080688638868,
"learning_rate": 1.1048828950339867e-05,
"loss": 0.3079,
"step": 33145
},
{
"epoch": 3.8654384328358207,
"grad_norm": 0.44450042431301534,
"learning_rate": 1.1042911322563903e-05,
"loss": 0.3259,
"step": 33150
},
{
"epoch": 3.8660214552238807,
"grad_norm": 0.4503506105615425,
"learning_rate": 1.1036996141799347e-05,
"loss": 0.3399,
"step": 33155
},
{
"epoch": 3.8666044776119404,
"grad_norm": 0.4775715826622604,
"learning_rate": 1.103108340892573e-05,
"loss": 0.3366,
"step": 33160
},
{
"epoch": 3.8671875,
"grad_norm": 0.4747788035082357,
"learning_rate": 1.1025173124822213e-05,
"loss": 0.3184,
"step": 33165
},
{
"epoch": 3.8677705223880596,
"grad_norm": 0.46296396496129416,
"learning_rate": 1.1019265290367616e-05,
"loss": 0.3401,
"step": 33170
},
{
"epoch": 3.8683535447761193,
"grad_norm": 0.430501096671592,
"learning_rate": 1.1013359906440353e-05,
"loss": 0.3008,
"step": 33175
},
{
"epoch": 3.8689365671641793,
"grad_norm": 0.437483851632213,
"learning_rate": 1.100745697391852e-05,
"loss": 0.3122,
"step": 33180
},
{
"epoch": 3.869519589552239,
"grad_norm": 0.4713102482347539,
"learning_rate": 1.1001556493679812e-05,
"loss": 0.3223,
"step": 33185
},
{
"epoch": 3.8701026119402986,
"grad_norm": 0.4496255994957661,
"learning_rate": 1.099565846660158e-05,
"loss": 0.3342,
"step": 33190
},
{
"epoch": 3.870685634328358,
"grad_norm": 0.48147533231731876,
"learning_rate": 1.0989762893560798e-05,
"loss": 0.3372,
"step": 33195
},
{
"epoch": 3.871268656716418,
"grad_norm": 0.44251577833344247,
"learning_rate": 1.0983869775434091e-05,
"loss": 0.3323,
"step": 33200
},
{
"epoch": 3.8718516791044775,
"grad_norm": 0.48410716018750416,
"learning_rate": 1.0977979113097702e-05,
"loss": 0.324,
"step": 33205
},
{
"epoch": 3.872434701492537,
"grad_norm": 0.44439583909512353,
"learning_rate": 1.097209090742752e-05,
"loss": 0.3061,
"step": 33210
},
{
"epoch": 3.873017723880597,
"grad_norm": 0.47159658683048755,
"learning_rate": 1.096620515929907e-05,
"loss": 0.3182,
"step": 33215
},
{
"epoch": 3.873600746268657,
"grad_norm": 0.4701875766362371,
"learning_rate": 1.096032186958749e-05,
"loss": 0.33,
"step": 33220
},
{
"epoch": 3.8741837686567164,
"grad_norm": 0.48776139257811485,
"learning_rate": 1.095444103916758e-05,
"loss": 0.3306,
"step": 33225
},
{
"epoch": 3.874766791044776,
"grad_norm": 0.48917754383258233,
"learning_rate": 1.0948562668913763e-05,
"loss": 0.3315,
"step": 33230
},
{
"epoch": 3.8753498134328357,
"grad_norm": 0.42845203482849836,
"learning_rate": 1.0942686759700092e-05,
"loss": 0.3361,
"step": 33235
},
{
"epoch": 3.8759328358208958,
"grad_norm": 0.4521128584339785,
"learning_rate": 1.0936813312400263e-05,
"loss": 0.3432,
"step": 33240
},
{
"epoch": 3.8765158582089554,
"grad_norm": 0.46657270299240866,
"learning_rate": 1.0930942327887605e-05,
"loss": 0.3333,
"step": 33245
},
{
"epoch": 3.877098880597015,
"grad_norm": 0.42492942434563746,
"learning_rate": 1.092507380703506e-05,
"loss": 0.3254,
"step": 33250
},
{
"epoch": 3.8776819029850746,
"grad_norm": 0.44468334518498764,
"learning_rate": 1.0919207750715243e-05,
"loss": 0.314,
"step": 33255
},
{
"epoch": 3.8782649253731343,
"grad_norm": 0.4651406759628273,
"learning_rate": 1.091334415980036e-05,
"loss": 0.3153,
"step": 33260
},
{
"epoch": 3.878847947761194,
"grad_norm": 0.4458658650524649,
"learning_rate": 1.0907483035162291e-05,
"loss": 0.3217,
"step": 33265
},
{
"epoch": 3.8794309701492535,
"grad_norm": 0.4663762736179635,
"learning_rate": 1.0901624377672513e-05,
"loss": 0.3061,
"step": 33270
},
{
"epoch": 3.8800139925373136,
"grad_norm": 0.4238179445095949,
"learning_rate": 1.0895768188202158e-05,
"loss": 0.3382,
"step": 33275
},
{
"epoch": 3.8805970149253732,
"grad_norm": 0.47716965895054936,
"learning_rate": 1.0889914467621986e-05,
"loss": 0.3275,
"step": 33280
},
{
"epoch": 3.881180037313433,
"grad_norm": 0.45739653512658146,
"learning_rate": 1.0884063216802388e-05,
"loss": 0.3217,
"step": 33285
},
{
"epoch": 3.8817630597014925,
"grad_norm": 0.48181114217634596,
"learning_rate": 1.0878214436613387e-05,
"loss": 0.3087,
"step": 33290
},
{
"epoch": 3.882346082089552,
"grad_norm": 0.4721078496089997,
"learning_rate": 1.0872368127924654e-05,
"loss": 0.3287,
"step": 33295
},
{
"epoch": 3.882929104477612,
"grad_norm": 0.474256957024035,
"learning_rate": 1.0866524291605452e-05,
"loss": 0.3301,
"step": 33300
},
{
"epoch": 3.8835121268656714,
"grad_norm": 0.43908918833917066,
"learning_rate": 1.0860682928524732e-05,
"loss": 0.3148,
"step": 33305
},
{
"epoch": 3.8840951492537314,
"grad_norm": 0.4688668078959364,
"learning_rate": 1.0854844039551023e-05,
"loss": 0.3072,
"step": 33310
},
{
"epoch": 3.884678171641791,
"grad_norm": 0.49775451058081527,
"learning_rate": 1.0849007625552539e-05,
"loss": 0.311,
"step": 33315
},
{
"epoch": 3.8852611940298507,
"grad_norm": 0.4565491640661001,
"learning_rate": 1.0843173687397079e-05,
"loss": 0.3254,
"step": 33320
},
{
"epoch": 3.8858442164179103,
"grad_norm": 0.5054496025514279,
"learning_rate": 1.0837342225952097e-05,
"loss": 0.3444,
"step": 33325
},
{
"epoch": 3.88642723880597,
"grad_norm": 0.4641649546547003,
"learning_rate": 1.0831513242084681e-05,
"loss": 0.3271,
"step": 33330
},
{
"epoch": 3.88701026119403,
"grad_norm": 0.43977199230377334,
"learning_rate": 1.0825686736661541e-05,
"loss": 0.3058,
"step": 33335
},
{
"epoch": 3.8875932835820897,
"grad_norm": 0.43046203770627206,
"learning_rate": 1.0819862710549025e-05,
"loss": 0.3295,
"step": 33340
},
{
"epoch": 3.8881763059701493,
"grad_norm": 0.4502064285248263,
"learning_rate": 1.0814041164613107e-05,
"loss": 0.3107,
"step": 33345
},
{
"epoch": 3.888759328358209,
"grad_norm": 0.49018547403922796,
"learning_rate": 1.0808222099719396e-05,
"loss": 0.33,
"step": 33350
},
{
"epoch": 3.8893423507462686,
"grad_norm": 0.44719309919402267,
"learning_rate": 1.0802405516733138e-05,
"loss": 0.3327,
"step": 33355
},
{
"epoch": 3.8899253731343286,
"grad_norm": 0.4349662933456471,
"learning_rate": 1.0796591416519192e-05,
"loss": 0.3073,
"step": 33360
},
{
"epoch": 3.890508395522388,
"grad_norm": 0.4769310745163692,
"learning_rate": 1.0790779799942063e-05,
"loss": 0.3294,
"step": 33365
},
{
"epoch": 3.891091417910448,
"grad_norm": 0.4633243435475389,
"learning_rate": 1.0784970667865882e-05,
"loss": 0.3389,
"step": 33370
},
{
"epoch": 3.8916744402985075,
"grad_norm": 0.4348569876234952,
"learning_rate": 1.0779164021154417e-05,
"loss": 0.3206,
"step": 33375
},
{
"epoch": 3.892257462686567,
"grad_norm": 0.4508919294527058,
"learning_rate": 1.0773359860671054e-05,
"loss": 0.3236,
"step": 33380
},
{
"epoch": 3.8928404850746268,
"grad_norm": 0.4176659453803,
"learning_rate": 1.0767558187278817e-05,
"loss": 0.3059,
"step": 33385
},
{
"epoch": 3.8934235074626864,
"grad_norm": 0.4936745756161462,
"learning_rate": 1.0761759001840371e-05,
"loss": 0.3656,
"step": 33390
},
{
"epoch": 3.8940065298507465,
"grad_norm": 0.4864860346779509,
"learning_rate": 1.0755962305217973e-05,
"loss": 0.3161,
"step": 33395
},
{
"epoch": 3.894589552238806,
"grad_norm": 0.47301983061706776,
"learning_rate": 1.0750168098273569e-05,
"loss": 0.3316,
"step": 33400
},
{
"epoch": 3.8951725746268657,
"grad_norm": 0.44945043215503067,
"learning_rate": 1.074437638186868e-05,
"loss": 0.3235,
"step": 33405
},
{
"epoch": 3.8957555970149254,
"grad_norm": 0.5170131988652965,
"learning_rate": 1.073858715686448e-05,
"loss": 0.3535,
"step": 33410
},
{
"epoch": 3.896338619402985,
"grad_norm": 0.44865396114012535,
"learning_rate": 1.0732800424121779e-05,
"loss": 0.3252,
"step": 33415
},
{
"epoch": 3.8969216417910446,
"grad_norm": 0.49849390150774087,
"learning_rate": 1.0727016184501e-05,
"loss": 0.3494,
"step": 33420
},
{
"epoch": 3.8975046641791042,
"grad_norm": 0.4575638153722119,
"learning_rate": 1.0721234438862213e-05,
"loss": 0.3244,
"step": 33425
},
{
"epoch": 3.8980876865671643,
"grad_norm": 0.4972029790663982,
"learning_rate": 1.0715455188065112e-05,
"loss": 0.3216,
"step": 33430
},
{
"epoch": 3.898670708955224,
"grad_norm": 0.4997182569608143,
"learning_rate": 1.0709678432968995e-05,
"loss": 0.3526,
"step": 33435
},
{
"epoch": 3.8992537313432836,
"grad_norm": 0.4745691563520591,
"learning_rate": 1.0703904174432836e-05,
"loss": 0.3265,
"step": 33440
},
{
"epoch": 3.899836753731343,
"grad_norm": 0.5168420977177586,
"learning_rate": 1.0698132413315188e-05,
"loss": 0.3311,
"step": 33445
},
{
"epoch": 3.900419776119403,
"grad_norm": 0.4455063996811109,
"learning_rate": 1.069236315047428e-05,
"loss": 0.3419,
"step": 33450
},
{
"epoch": 3.901002798507463,
"grad_norm": 0.47340676730818415,
"learning_rate": 1.0686596386767928e-05,
"loss": 0.3333,
"step": 33455
},
{
"epoch": 3.9015858208955225,
"grad_norm": 0.4809870026301231,
"learning_rate": 1.0680832123053603e-05,
"loss": 0.3274,
"step": 33460
},
{
"epoch": 3.902168843283582,
"grad_norm": 0.4465458990605889,
"learning_rate": 1.067507036018839e-05,
"loss": 0.3447,
"step": 33465
},
{
"epoch": 3.902751865671642,
"grad_norm": 0.4465058729904804,
"learning_rate": 1.0669311099029014e-05,
"loss": 0.3475,
"step": 33470
},
{
"epoch": 3.9033348880597014,
"grad_norm": 0.5116129137143816,
"learning_rate": 1.066355434043182e-05,
"loss": 0.3439,
"step": 33475
},
{
"epoch": 3.903917910447761,
"grad_norm": 0.4717166940202731,
"learning_rate": 1.0657800085252789e-05,
"loss": 0.3349,
"step": 33480
},
{
"epoch": 3.9045009328358207,
"grad_norm": 0.49214369973716376,
"learning_rate": 1.0652048334347503e-05,
"loss": 0.3384,
"step": 33485
},
{
"epoch": 3.9050839552238807,
"grad_norm": 0.4754204891436952,
"learning_rate": 1.064629908857122e-05,
"loss": 0.3252,
"step": 33490
},
{
"epoch": 3.9056669776119404,
"grad_norm": 0.4368586840816342,
"learning_rate": 1.0640552348778772e-05,
"loss": 0.3375,
"step": 33495
},
{
"epoch": 3.90625,
"grad_norm": 0.4529978997743157,
"learning_rate": 1.0634808115824668e-05,
"loss": 0.3234,
"step": 33500
},
{
"epoch": 3.9068330223880596,
"grad_norm": 0.49149399879697336,
"learning_rate": 1.0629066390563002e-05,
"loss": 0.321,
"step": 33505
},
{
"epoch": 3.9074160447761193,
"grad_norm": 0.4509824332176206,
"learning_rate": 1.062332717384752e-05,
"loss": 0.3581,
"step": 33510
},
{
"epoch": 3.9079990671641793,
"grad_norm": 0.4537482827730715,
"learning_rate": 1.061759046653159e-05,
"loss": 0.3393,
"step": 33515
},
{
"epoch": 3.908582089552239,
"grad_norm": 0.5105268910887601,
"learning_rate": 1.0611856269468203e-05,
"loss": 0.3447,
"step": 33520
},
{
"epoch": 3.9091651119402986,
"grad_norm": 0.4901727461174683,
"learning_rate": 1.0606124583509983e-05,
"loss": 0.3255,
"step": 33525
},
{
"epoch": 3.909748134328358,
"grad_norm": 0.4340808643488463,
"learning_rate": 1.0600395409509177e-05,
"loss": 0.3421,
"step": 33530
},
{
"epoch": 3.910331156716418,
"grad_norm": 0.484371758609476,
"learning_rate": 1.0594668748317643e-05,
"loss": 0.3099,
"step": 33535
},
{
"epoch": 3.9109141791044775,
"grad_norm": 0.4553888233254177,
"learning_rate": 1.0588944600786907e-05,
"loss": 0.2952,
"step": 33540
},
{
"epoch": 3.911497201492537,
"grad_norm": 0.482461618635314,
"learning_rate": 1.0583222967768076e-05,
"loss": 0.3072,
"step": 33545
},
{
"epoch": 3.912080223880597,
"grad_norm": 0.4991090354947881,
"learning_rate": 1.0577503850111903e-05,
"loss": 0.3241,
"step": 33550
},
{
"epoch": 3.912663246268657,
"grad_norm": 0.432300133663098,
"learning_rate": 1.0571787248668774e-05,
"loss": 0.3403,
"step": 33555
},
{
"epoch": 3.9132462686567164,
"grad_norm": 0.48530227575772117,
"learning_rate": 1.0566073164288687e-05,
"loss": 0.3301,
"step": 33560
},
{
"epoch": 3.913829291044776,
"grad_norm": 0.4873409308216426,
"learning_rate": 1.0560361597821273e-05,
"loss": 0.3247,
"step": 33565
},
{
"epoch": 3.9144123134328357,
"grad_norm": 0.4080850363735003,
"learning_rate": 1.0554652550115788e-05,
"loss": 0.3137,
"step": 33570
},
{
"epoch": 3.9149953358208958,
"grad_norm": 0.4453352465572526,
"learning_rate": 1.054894602202112e-05,
"loss": 0.3084,
"step": 33575
},
{
"epoch": 3.9155783582089554,
"grad_norm": 0.46849618867634657,
"learning_rate": 1.0543242014385758e-05,
"loss": 0.3101,
"step": 33580
},
{
"epoch": 3.916161380597015,
"grad_norm": 0.501994941153192,
"learning_rate": 1.0537540528057844e-05,
"loss": 0.3325,
"step": 33585
},
{
"epoch": 3.9167444029850746,
"grad_norm": 0.44566099017892913,
"learning_rate": 1.0531841563885134e-05,
"loss": 0.3099,
"step": 33590
},
{
"epoch": 3.9173274253731343,
"grad_norm": 0.5558749333263552,
"learning_rate": 1.0526145122715007e-05,
"loss": 0.326,
"step": 33595
},
{
"epoch": 3.917910447761194,
"grad_norm": 0.43950209811951646,
"learning_rate": 1.052045120539447e-05,
"loss": 0.3151,
"step": 33600
},
{
"epoch": 3.9184934701492535,
"grad_norm": 0.5072366338265899,
"learning_rate": 1.051475981277016e-05,
"loss": 0.3456,
"step": 33605
},
{
"epoch": 3.9190764925373136,
"grad_norm": 0.5196266090165022,
"learning_rate": 1.050907094568832e-05,
"loss": 0.3609,
"step": 33610
},
{
"epoch": 3.9196595149253732,
"grad_norm": 0.4711784241295961,
"learning_rate": 1.0503384604994846e-05,
"loss": 0.323,
"step": 33615
},
{
"epoch": 3.920242537313433,
"grad_norm": 0.4854892163180719,
"learning_rate": 1.0497700791535221e-05,
"loss": 0.3353,
"step": 33620
},
{
"epoch": 3.9208255597014925,
"grad_norm": 0.5257714374863696,
"learning_rate": 1.04920195061546e-05,
"loss": 0.3431,
"step": 33625
},
{
"epoch": 3.921408582089552,
"grad_norm": 0.5062598555312798,
"learning_rate": 1.0486340749697716e-05,
"loss": 0.3495,
"step": 33630
},
{
"epoch": 3.921991604477612,
"grad_norm": 0.48392908035583404,
"learning_rate": 1.0480664523008948e-05,
"loss": 0.3264,
"step": 33635
},
{
"epoch": 3.9225746268656714,
"grad_norm": 0.4650271530262864,
"learning_rate": 1.0474990826932301e-05,
"loss": 0.3253,
"step": 33640
},
{
"epoch": 3.9231576492537314,
"grad_norm": 0.5065185766738228,
"learning_rate": 1.0469319662311403e-05,
"loss": 0.3442,
"step": 33645
},
{
"epoch": 3.923740671641791,
"grad_norm": 0.4508854584980229,
"learning_rate": 1.0463651029989492e-05,
"loss": 0.335,
"step": 33650
},
{
"epoch": 3.9243236940298507,
"grad_norm": 0.5485095643962583,
"learning_rate": 1.0457984930809452e-05,
"loss": 0.3683,
"step": 33655
},
{
"epoch": 3.9249067164179103,
"grad_norm": 0.5218275787456073,
"learning_rate": 1.0452321365613758e-05,
"loss": 0.3506,
"step": 33660
},
{
"epoch": 3.92548973880597,
"grad_norm": 0.4757228072173987,
"learning_rate": 1.0446660335244551e-05,
"loss": 0.328,
"step": 33665
},
{
"epoch": 3.92607276119403,
"grad_norm": 0.49953106921690893,
"learning_rate": 1.0441001840543548e-05,
"loss": 0.3359,
"step": 33670
},
{
"epoch": 3.9266557835820897,
"grad_norm": 0.453266974029994,
"learning_rate": 1.0435345882352144e-05,
"loss": 0.3173,
"step": 33675
},
{
"epoch": 3.9272388059701493,
"grad_norm": 0.5206236913269021,
"learning_rate": 1.0429692461511298e-05,
"loss": 0.3386,
"step": 33680
},
{
"epoch": 3.927821828358209,
"grad_norm": 0.47800687038688183,
"learning_rate": 1.0424041578861626e-05,
"loss": 0.322,
"step": 33685
},
{
"epoch": 3.9284048507462686,
"grad_norm": 0.44966064859975596,
"learning_rate": 1.041839323524337e-05,
"loss": 0.3324,
"step": 33690
},
{
"epoch": 3.9289878731343286,
"grad_norm": 0.48893228822481516,
"learning_rate": 1.0412747431496372e-05,
"loss": 0.3328,
"step": 33695
},
{
"epoch": 3.929570895522388,
"grad_norm": 0.42419307008604823,
"learning_rate": 1.0407104168460116e-05,
"loss": 0.319,
"step": 33700
},
{
"epoch": 3.930153917910448,
"grad_norm": 0.47660750708650085,
"learning_rate": 1.0401463446973708e-05,
"loss": 0.3419,
"step": 33705
},
{
"epoch": 3.9307369402985075,
"grad_norm": 0.43585377683092885,
"learning_rate": 1.0395825267875846e-05,
"loss": 0.3083,
"step": 33710
},
{
"epoch": 3.931319962686567,
"grad_norm": 0.4780579368292095,
"learning_rate": 1.0390189632004905e-05,
"loss": 0.3239,
"step": 33715
},
{
"epoch": 3.9319029850746268,
"grad_norm": 0.44920571184233704,
"learning_rate": 1.0384556540198825e-05,
"loss": 0.3262,
"step": 33720
},
{
"epoch": 3.9324860074626864,
"grad_norm": 0.47175046583080993,
"learning_rate": 1.0378925993295202e-05,
"loss": 0.3293,
"step": 33725
},
{
"epoch": 3.9330690298507465,
"grad_norm": 0.46480283167238134,
"learning_rate": 1.0373297992131242e-05,
"loss": 0.3242,
"step": 33730
},
{
"epoch": 3.933652052238806,
"grad_norm": 0.4175856958165451,
"learning_rate": 1.0367672537543777e-05,
"loss": 0.3083,
"step": 33735
},
{
"epoch": 3.9342350746268657,
"grad_norm": 0.41018167877598666,
"learning_rate": 1.0362049630369259e-05,
"loss": 0.309,
"step": 33740
},
{
"epoch": 3.9348180970149254,
"grad_norm": 0.4955778837967272,
"learning_rate": 1.0356429271443757e-05,
"loss": 0.3282,
"step": 33745
},
{
"epoch": 3.935401119402985,
"grad_norm": 0.4265611463098694,
"learning_rate": 1.0350811461602974e-05,
"loss": 0.3255,
"step": 33750
},
{
"epoch": 3.9359841417910446,
"grad_norm": 0.5561878417502005,
"learning_rate": 1.0345196201682212e-05,
"loss": 0.3307,
"step": 33755
},
{
"epoch": 3.9365671641791042,
"grad_norm": 0.44307740143964736,
"learning_rate": 1.033958349251641e-05,
"loss": 0.3163,
"step": 33760
},
{
"epoch": 3.9371501865671643,
"grad_norm": 0.5090739166525471,
"learning_rate": 1.0333973334940125e-05,
"loss": 0.3483,
"step": 33765
},
{
"epoch": 3.937733208955224,
"grad_norm": 0.4525542709236959,
"learning_rate": 1.0328365729787536e-05,
"loss": 0.323,
"step": 33770
},
{
"epoch": 3.9383162313432836,
"grad_norm": 0.46050542647327514,
"learning_rate": 1.0322760677892437e-05,
"loss": 0.3288,
"step": 33775
},
{
"epoch": 3.938899253731343,
"grad_norm": 0.45907426668515966,
"learning_rate": 1.0317158180088254e-05,
"loss": 0.3445,
"step": 33780
},
{
"epoch": 3.939482276119403,
"grad_norm": 0.5230425517412971,
"learning_rate": 1.0311558237208006e-05,
"loss": 0.3262,
"step": 33785
},
{
"epoch": 3.940065298507463,
"grad_norm": 0.4715806732489677,
"learning_rate": 1.0305960850084373e-05,
"loss": 0.3278,
"step": 33790
},
{
"epoch": 3.9406483208955225,
"grad_norm": 0.4801936056919124,
"learning_rate": 1.030036601954961e-05,
"loss": 0.3471,
"step": 33795
},
{
"epoch": 3.941231343283582,
"grad_norm": 0.4793655410602897,
"learning_rate": 1.0294773746435638e-05,
"loss": 0.3509,
"step": 33800
},
{
"epoch": 3.941814365671642,
"grad_norm": 0.4383407971147484,
"learning_rate": 1.028918403157396e-05,
"loss": 0.3416,
"step": 33805
},
{
"epoch": 3.9423973880597014,
"grad_norm": 0.47283544342158557,
"learning_rate": 1.0283596875795718e-05,
"loss": 0.3367,
"step": 33810
},
{
"epoch": 3.942980410447761,
"grad_norm": 0.5137500935771644,
"learning_rate": 1.0278012279931665e-05,
"loss": 0.3569,
"step": 33815
},
{
"epoch": 3.9435634328358207,
"grad_norm": 0.4524030901963742,
"learning_rate": 1.0272430244812175e-05,
"loss": 0.3186,
"step": 33820
},
{
"epoch": 3.9441464552238807,
"grad_norm": 0.5053690630460168,
"learning_rate": 1.0266850771267253e-05,
"loss": 0.3424,
"step": 33825
},
{
"epoch": 3.9447294776119404,
"grad_norm": 0.4200738173375548,
"learning_rate": 1.0261273860126514e-05,
"loss": 0.3028,
"step": 33830
},
{
"epoch": 3.9453125,
"grad_norm": 0.467065619389046,
"learning_rate": 1.0255699512219166e-05,
"loss": 0.3232,
"step": 33835
},
{
"epoch": 3.9458955223880596,
"grad_norm": 0.46116333523776293,
"learning_rate": 1.0250127728374098e-05,
"loss": 0.3136,
"step": 33840
},
{
"epoch": 3.9464785447761193,
"grad_norm": 0.45348991820287743,
"learning_rate": 1.0244558509419748e-05,
"loss": 0.315,
"step": 33845
},
{
"epoch": 3.9470615671641793,
"grad_norm": 0.4517076456612718,
"learning_rate": 1.023899185618423e-05,
"loss": 0.3208,
"step": 33850
},
{
"epoch": 3.947644589552239,
"grad_norm": 0.44252536059953496,
"learning_rate": 1.023342776949524e-05,
"loss": 0.323,
"step": 33855
},
{
"epoch": 3.9482276119402986,
"grad_norm": 0.4811269923189408,
"learning_rate": 1.0227866250180105e-05,
"loss": 0.3246,
"step": 33860
},
{
"epoch": 3.948810634328358,
"grad_norm": 0.45067652056352503,
"learning_rate": 1.022230729906577e-05,
"loss": 0.3308,
"step": 33865
},
{
"epoch": 3.949393656716418,
"grad_norm": 0.42938991082270744,
"learning_rate": 1.02167509169788e-05,
"loss": 0.3154,
"step": 33870
},
{
"epoch": 3.9499766791044775,
"grad_norm": 0.4637541076276976,
"learning_rate": 1.0211197104745373e-05,
"loss": 0.3501,
"step": 33875
},
{
"epoch": 3.950559701492537,
"grad_norm": 0.4505223220118978,
"learning_rate": 1.02056458631913e-05,
"loss": 0.3286,
"step": 33880
},
{
"epoch": 3.951142723880597,
"grad_norm": 0.47479324238117204,
"learning_rate": 1.020009719314197e-05,
"loss": 0.3388,
"step": 33885
},
{
"epoch": 3.951725746268657,
"grad_norm": 0.44894587425131394,
"learning_rate": 1.0194551095422447e-05,
"loss": 0.323,
"step": 33890
},
{
"epoch": 3.9523087686567164,
"grad_norm": 0.46095727378674345,
"learning_rate": 1.0189007570857363e-05,
"loss": 0.3222,
"step": 33895
},
{
"epoch": 3.952891791044776,
"grad_norm": 0.4632299502106948,
"learning_rate": 1.0183466620270996e-05,
"loss": 0.3386,
"step": 33900
},
{
"epoch": 3.9534748134328357,
"grad_norm": 0.46950464682753884,
"learning_rate": 1.0177928244487225e-05,
"loss": 0.3335,
"step": 33905
},
{
"epoch": 3.9540578358208958,
"grad_norm": 0.4300936607612698,
"learning_rate": 1.0172392444329561e-05,
"loss": 0.3211,
"step": 33910
},
{
"epoch": 3.9546408582089554,
"grad_norm": 0.42780701592683107,
"learning_rate": 1.0166859220621122e-05,
"loss": 0.3147,
"step": 33915
},
{
"epoch": 3.955223880597015,
"grad_norm": 0.46160418788896695,
"learning_rate": 1.0161328574184645e-05,
"loss": 0.3255,
"step": 33920
},
{
"epoch": 3.9558069029850746,
"grad_norm": 0.47917701468417334,
"learning_rate": 1.015580050584249e-05,
"loss": 0.3258,
"step": 33925
},
{
"epoch": 3.9563899253731343,
"grad_norm": 0.4850823979231493,
"learning_rate": 1.0150275016416613e-05,
"loss": 0.3076,
"step": 33930
},
{
"epoch": 3.956972947761194,
"grad_norm": 0.49446454624379393,
"learning_rate": 1.0144752106728613e-05,
"loss": 0.3455,
"step": 33935
},
{
"epoch": 3.9575559701492535,
"grad_norm": 0.45225342777195443,
"learning_rate": 1.0139231777599689e-05,
"loss": 0.3278,
"step": 33940
},
{
"epoch": 3.9581389925373136,
"grad_norm": 0.4417728280258304,
"learning_rate": 1.0133714029850667e-05,
"loss": 0.3338,
"step": 33945
},
{
"epoch": 3.9587220149253732,
"grad_norm": 0.47526603632268744,
"learning_rate": 1.0128198864301976e-05,
"loss": 0.3188,
"step": 33950
},
{
"epoch": 3.959305037313433,
"grad_norm": 0.4566346351343713,
"learning_rate": 1.0122686281773674e-05,
"loss": 0.3214,
"step": 33955
},
{
"epoch": 3.9598880597014925,
"grad_norm": 0.46242712128959035,
"learning_rate": 1.0117176283085419e-05,
"loss": 0.3208,
"step": 33960
},
{
"epoch": 3.960471082089552,
"grad_norm": 0.4421784234059182,
"learning_rate": 1.0111668869056515e-05,
"loss": 0.3322,
"step": 33965
},
{
"epoch": 3.961054104477612,
"grad_norm": 0.44062643183665456,
"learning_rate": 1.0106164040505835e-05,
"loss": 0.3305,
"step": 33970
},
{
"epoch": 3.9616371268656714,
"grad_norm": 0.47666506355668165,
"learning_rate": 1.0100661798251923e-05,
"loss": 0.3329,
"step": 33975
},
{
"epoch": 3.9622201492537314,
"grad_norm": 0.4696572293050511,
"learning_rate": 1.009516214311289e-05,
"loss": 0.3291,
"step": 33980
},
{
"epoch": 3.962803171641791,
"grad_norm": 0.44796711592830507,
"learning_rate": 1.0089665075906485e-05,
"loss": 0.3021,
"step": 33985
},
{
"epoch": 3.9633861940298507,
"grad_norm": 0.4215620720448473,
"learning_rate": 1.0084170597450073e-05,
"loss": 0.3357,
"step": 33990
},
{
"epoch": 3.9639692164179103,
"grad_norm": 0.5135757878352224,
"learning_rate": 1.0078678708560627e-05,
"loss": 0.3279,
"step": 33995
},
{
"epoch": 3.96455223880597,
"grad_norm": 0.4382478647685149,
"learning_rate": 1.0073189410054742e-05,
"loss": 0.3329,
"step": 34000
},
{
"epoch": 3.96513526119403,
"grad_norm": 0.47441238354061055,
"learning_rate": 1.0067702702748627e-05,
"loss": 0.3113,
"step": 34005
},
{
"epoch": 3.9657182835820897,
"grad_norm": 0.47045652294160595,
"learning_rate": 1.0062218587458085e-05,
"loss": 0.3277,
"step": 34010
},
{
"epoch": 3.9663013059701493,
"grad_norm": 0.46033420707207967,
"learning_rate": 1.005673706499858e-05,
"loss": 0.3344,
"step": 34015
},
{
"epoch": 3.966884328358209,
"grad_norm": 0.48294006873083045,
"learning_rate": 1.0051258136185132e-05,
"loss": 0.3236,
"step": 34020
},
{
"epoch": 3.9674673507462686,
"grad_norm": 0.48696980348070035,
"learning_rate": 1.004578180183243e-05,
"loss": 0.3407,
"step": 34025
},
{
"epoch": 3.9680503731343286,
"grad_norm": 0.45801264492862054,
"learning_rate": 1.0040308062754738e-05,
"loss": 0.332,
"step": 34030
},
{
"epoch": 3.968633395522388,
"grad_norm": 0.45810702477293136,
"learning_rate": 1.0034836919765953e-05,
"loss": 0.32,
"step": 34035
},
{
"epoch": 3.969216417910448,
"grad_norm": 0.48611654623430944,
"learning_rate": 1.0029368373679583e-05,
"loss": 0.3286,
"step": 34040
},
{
"epoch": 3.9697994402985075,
"grad_norm": 0.4428633768030733,
"learning_rate": 1.002390242530874e-05,
"loss": 0.3099,
"step": 34045
},
{
"epoch": 3.970382462686567,
"grad_norm": 0.4519262796809441,
"learning_rate": 1.001843907546617e-05,
"loss": 0.3242,
"step": 34050
},
{
"epoch": 3.9709654850746268,
"grad_norm": 0.4952882194557926,
"learning_rate": 1.0012978324964214e-05,
"loss": 0.3361,
"step": 34055
},
{
"epoch": 3.9715485074626864,
"grad_norm": 0.45588317582345006,
"learning_rate": 1.0007520174614836e-05,
"loss": 0.2993,
"step": 34060
},
{
"epoch": 3.9721315298507465,
"grad_norm": 0.4690490984838446,
"learning_rate": 1.0002064625229613e-05,
"loss": 0.3176,
"step": 34065
},
{
"epoch": 3.972714552238806,
"grad_norm": 0.47169734458793017,
"learning_rate": 9.996611677619719e-06,
"loss": 0.3294,
"step": 34070
},
{
"epoch": 3.9732975746268657,
"grad_norm": 0.4356789270614016,
"learning_rate": 9.991161332595978e-06,
"loss": 0.3101,
"step": 34075
},
{
"epoch": 3.9738805970149254,
"grad_norm": 0.4398300251087639,
"learning_rate": 9.98571359096878e-06,
"loss": 0.3216,
"step": 34080
},
{
"epoch": 3.974463619402985,
"grad_norm": 0.5044326152170167,
"learning_rate": 9.980268453548172e-06,
"loss": 0.3348,
"step": 34085
},
{
"epoch": 3.9750466417910446,
"grad_norm": 0.4908949137397319,
"learning_rate": 9.97482592114378e-06,
"loss": 0.3452,
"step": 34090
},
{
"epoch": 3.9756296641791042,
"grad_norm": 0.4388141472417088,
"learning_rate": 9.969385994564862e-06,
"loss": 0.3186,
"step": 34095
},
{
"epoch": 3.9762126865671643,
"grad_norm": 0.4979654560935064,
"learning_rate": 9.96394867462028e-06,
"loss": 0.3412,
"step": 34100
},
{
"epoch": 3.976795708955224,
"grad_norm": 0.5235472401417905,
"learning_rate": 9.958513962118521e-06,
"loss": 0.32,
"step": 34105
},
{
"epoch": 3.9773787313432836,
"grad_norm": 0.4253250863988798,
"learning_rate": 9.953081857867665e-06,
"loss": 0.3323,
"step": 34110
},
{
"epoch": 3.977961753731343,
"grad_norm": 0.44316158786191384,
"learning_rate": 9.947652362675418e-06,
"loss": 0.2972,
"step": 34115
},
{
"epoch": 3.978544776119403,
"grad_norm": 0.4600520895546006,
"learning_rate": 9.94222547734909e-06,
"loss": 0.3256,
"step": 34120
},
{
"epoch": 3.979127798507463,
"grad_norm": 0.45442856356849437,
"learning_rate": 9.936801202695607e-06,
"loss": 0.3435,
"step": 34125
},
{
"epoch": 3.9797108208955225,
"grad_norm": 0.4611334100716397,
"learning_rate": 9.93137953952151e-06,
"loss": 0.3181,
"step": 34130
},
{
"epoch": 3.980293843283582,
"grad_norm": 0.4535727228122911,
"learning_rate": 9.925960488632948e-06,
"loss": 0.3193,
"step": 34135
},
{
"epoch": 3.980876865671642,
"grad_norm": 0.4688141137143561,
"learning_rate": 9.92054405083569e-06,
"loss": 0.3346,
"step": 34140
},
{
"epoch": 3.9814598880597014,
"grad_norm": 0.4533556132404464,
"learning_rate": 9.915130226935081e-06,
"loss": 0.3392,
"step": 34145
},
{
"epoch": 3.982042910447761,
"grad_norm": 0.5034582760137185,
"learning_rate": 9.90971901773614e-06,
"loss": 0.318,
"step": 34150
},
{
"epoch": 3.9826259328358207,
"grad_norm": 0.44084630090610327,
"learning_rate": 9.904310424043432e-06,
"loss": 0.3228,
"step": 34155
},
{
"epoch": 3.9832089552238807,
"grad_norm": 0.6637125022574998,
"learning_rate": 9.898904446661188e-06,
"loss": 0.3248,
"step": 34160
},
{
"epoch": 3.9837919776119404,
"grad_norm": 0.46728347647795127,
"learning_rate": 9.89350108639321e-06,
"loss": 0.3165,
"step": 34165
},
{
"epoch": 3.984375,
"grad_norm": 0.4702358150362368,
"learning_rate": 9.888100344042926e-06,
"loss": 0.3313,
"step": 34170
},
{
"epoch": 3.9849580223880596,
"grad_norm": 0.43476736172092595,
"learning_rate": 9.88270222041338e-06,
"loss": 0.3159,
"step": 34175
},
{
"epoch": 3.9855410447761193,
"grad_norm": 0.4531721630039909,
"learning_rate": 9.87730671630722e-06,
"loss": 0.3148,
"step": 34180
},
{
"epoch": 3.9861240671641793,
"grad_norm": 0.46481035080290545,
"learning_rate": 9.871913832526702e-06,
"loss": 0.3294,
"step": 34185
},
{
"epoch": 3.986707089552239,
"grad_norm": 0.4524958217254779,
"learning_rate": 9.866523569873708e-06,
"loss": 0.3376,
"step": 34190
},
{
"epoch": 3.9872901119402986,
"grad_norm": 0.4877276855479251,
"learning_rate": 9.861135929149695e-06,
"loss": 0.3211,
"step": 34195
},
{
"epoch": 3.987873134328358,
"grad_norm": 0.4217362588631423,
"learning_rate": 9.855750911155784e-06,
"loss": 0.3225,
"step": 34200
},
{
"epoch": 3.988456156716418,
"grad_norm": 4.920054886385086,
"learning_rate": 9.850368516692643e-06,
"loss": 0.3257,
"step": 34205
},
{
"epoch": 3.9890391791044775,
"grad_norm": 0.4452963636002685,
"learning_rate": 9.844988746560615e-06,
"loss": 0.3187,
"step": 34210
},
{
"epoch": 3.989622201492537,
"grad_norm": 0.4222014745462168,
"learning_rate": 9.839611601559597e-06,
"loss": 0.2971,
"step": 34215
},
{
"epoch": 3.990205223880597,
"grad_norm": 0.4926216363486367,
"learning_rate": 9.834237082489126e-06,
"loss": 0.3473,
"step": 34220
},
{
"epoch": 3.990788246268657,
"grad_norm": 0.4813548007313419,
"learning_rate": 9.828865190148342e-06,
"loss": 0.3265,
"step": 34225
},
{
"epoch": 3.9913712686567164,
"grad_norm": 0.47602805355492867,
"learning_rate": 9.823495925335995e-06,
"loss": 0.3292,
"step": 34230
},
{
"epoch": 3.991954291044776,
"grad_norm": 0.4573633687387112,
"learning_rate": 9.81812928885044e-06,
"loss": 0.3297,
"step": 34235
},
{
"epoch": 3.9925373134328357,
"grad_norm": 0.4959606630490096,
"learning_rate": 9.812765281489655e-06,
"loss": 0.3663,
"step": 34240
},
{
"epoch": 3.9931203358208958,
"grad_norm": 0.5063115538753646,
"learning_rate": 9.807403904051194e-06,
"loss": 0.3234,
"step": 34245
},
{
"epoch": 3.9937033582089554,
"grad_norm": 0.4838249620180325,
"learning_rate": 9.802045157332269e-06,
"loss": 0.3288,
"step": 34250
},
{
"epoch": 3.994286380597015,
"grad_norm": 0.46547509225290806,
"learning_rate": 9.796689042129652e-06,
"loss": 0.3084,
"step": 34255
},
{
"epoch": 3.9948694029850746,
"grad_norm": 0.5084441558896075,
"learning_rate": 9.79133555923976e-06,
"loss": 0.3341,
"step": 34260
},
{
"epoch": 3.9954524253731343,
"grad_norm": 0.4782270556677637,
"learning_rate": 9.785984709458602e-06,
"loss": 0.3358,
"step": 34265
},
{
"epoch": 3.996035447761194,
"grad_norm": 0.4669360357065058,
"learning_rate": 9.780636493581797e-06,
"loss": 0.3349,
"step": 34270
},
{
"epoch": 3.9966184701492535,
"grad_norm": 0.47936948411575514,
"learning_rate": 9.775290912404569e-06,
"loss": 0.3174,
"step": 34275
},
{
"epoch": 3.9972014925373136,
"grad_norm": 0.4750311734313681,
"learning_rate": 9.76994796672176e-06,
"loss": 0.3357,
"step": 34280
},
{
"epoch": 3.9977845149253732,
"grad_norm": 0.43884871232542555,
"learning_rate": 9.764607657327818e-06,
"loss": 0.3256,
"step": 34285
},
{
"epoch": 3.998367537313433,
"grad_norm": 0.5001795660217874,
"learning_rate": 9.759269985016786e-06,
"loss": 0.3546,
"step": 34290
},
{
"epoch": 3.9989505597014925,
"grad_norm": 0.42042507950707164,
"learning_rate": 9.753934950582333e-06,
"loss": 0.341,
"step": 34295
},
{
"epoch": 3.999533582089552,
"grad_norm": 0.4523835748984335,
"learning_rate": 9.748602554817721e-06,
"loss": 0.321,
"step": 34300
},
{
"epoch": 4.000116604477612,
"grad_norm": 0.4531600037167153,
"learning_rate": 9.743272798515829e-06,
"loss": 0.3042,
"step": 34305
},
{
"epoch": 4.000699626865671,
"grad_norm": 0.4108782613340139,
"learning_rate": 9.737945682469145e-06,
"loss": 0.2542,
"step": 34310
},
{
"epoch": 4.0012826492537314,
"grad_norm": 0.4349396955742546,
"learning_rate": 9.732621207469761e-06,
"loss": 0.2688,
"step": 34315
},
{
"epoch": 4.001865671641791,
"grad_norm": 0.4804686197522778,
"learning_rate": 9.72729937430936e-06,
"loss": 0.2487,
"step": 34320
},
{
"epoch": 4.002448694029851,
"grad_norm": 0.5048592222069496,
"learning_rate": 9.72198018377927e-06,
"loss": 0.2587,
"step": 34325
},
{
"epoch": 4.003031716417911,
"grad_norm": 0.46600498950028896,
"learning_rate": 9.716663636670375e-06,
"loss": 0.2505,
"step": 34330
},
{
"epoch": 4.00361473880597,
"grad_norm": 0.5158150099952746,
"learning_rate": 9.71134973377323e-06,
"loss": 0.2639,
"step": 34335
},
{
"epoch": 4.00419776119403,
"grad_norm": 0.5012120609047553,
"learning_rate": 9.706038475877938e-06,
"loss": 0.2581,
"step": 34340
},
{
"epoch": 4.004780783582089,
"grad_norm": 0.5258548596606887,
"learning_rate": 9.700729863774233e-06,
"loss": 0.2626,
"step": 34345
},
{
"epoch": 4.005363805970149,
"grad_norm": 0.4691002753690056,
"learning_rate": 9.69542389825146e-06,
"loss": 0.2607,
"step": 34350
},
{
"epoch": 4.005946828358209,
"grad_norm": 0.5207991899052077,
"learning_rate": 9.690120580098566e-06,
"loss": 0.254,
"step": 34355
},
{
"epoch": 4.0065298507462686,
"grad_norm": 0.4865108266583932,
"learning_rate": 9.6848199101041e-06,
"loss": 0.2678,
"step": 34360
},
{
"epoch": 4.007112873134329,
"grad_norm": 0.5057030285566712,
"learning_rate": 9.67952188905623e-06,
"loss": 0.2625,
"step": 34365
},
{
"epoch": 4.007695895522388,
"grad_norm": 0.4991036815732012,
"learning_rate": 9.674226517742705e-06,
"loss": 0.2421,
"step": 34370
},
{
"epoch": 4.008278917910448,
"grad_norm": 0.5259789567259351,
"learning_rate": 9.668933796950913e-06,
"loss": 0.2616,
"step": 34375
},
{
"epoch": 4.008861940298507,
"grad_norm": 0.4905941013631162,
"learning_rate": 9.66364372746781e-06,
"loss": 0.2439,
"step": 34380
},
{
"epoch": 4.009444962686567,
"grad_norm": 0.511369017220589,
"learning_rate": 9.658356310080007e-06,
"loss": 0.2561,
"step": 34385
},
{
"epoch": 4.010027985074627,
"grad_norm": 0.5533099144993038,
"learning_rate": 9.653071545573667e-06,
"loss": 0.2776,
"step": 34390
},
{
"epoch": 4.010611007462686,
"grad_norm": 0.5128247997774668,
"learning_rate": 9.647789434734594e-06,
"loss": 0.2625,
"step": 34395
},
{
"epoch": 4.0111940298507465,
"grad_norm": 0.4701134532348846,
"learning_rate": 9.64250997834819e-06,
"loss": 0.2418,
"step": 34400
},
{
"epoch": 4.011777052238806,
"grad_norm": 0.45810143573615186,
"learning_rate": 9.637233177199452e-06,
"loss": 0.2579,
"step": 34405
},
{
"epoch": 4.012360074626866,
"grad_norm": 0.4867578503297538,
"learning_rate": 9.631959032072997e-06,
"loss": 0.2444,
"step": 34410
},
{
"epoch": 4.012943097014926,
"grad_norm": 0.49294893674379886,
"learning_rate": 9.626687543753041e-06,
"loss": 0.2415,
"step": 34415
},
{
"epoch": 4.013526119402985,
"grad_norm": 0.5085206872787971,
"learning_rate": 9.621418713023389e-06,
"loss": 0.2729,
"step": 34420
},
{
"epoch": 4.014109141791045,
"grad_norm": 0.5347517791479994,
"learning_rate": 9.616152540667488e-06,
"loss": 0.2492,
"step": 34425
},
{
"epoch": 4.014692164179104,
"grad_norm": 0.5074309706958307,
"learning_rate": 9.61088902746835e-06,
"loss": 0.2531,
"step": 34430
},
{
"epoch": 4.015275186567164,
"grad_norm": 0.4975697039738303,
"learning_rate": 9.605628174208617e-06,
"loss": 0.2357,
"step": 34435
},
{
"epoch": 4.0158582089552235,
"grad_norm": 0.4977799738759334,
"learning_rate": 9.60036998167052e-06,
"loss": 0.2356,
"step": 34440
},
{
"epoch": 4.016441231343284,
"grad_norm": 0.534198214468689,
"learning_rate": 9.595114450635911e-06,
"loss": 0.2589,
"step": 34445
},
{
"epoch": 4.017024253731344,
"grad_norm": 0.5444308932824548,
"learning_rate": 9.589861581886232e-06,
"loss": 0.2846,
"step": 34450
},
{
"epoch": 4.017607276119403,
"grad_norm": 0.5074586532370339,
"learning_rate": 9.584611376202534e-06,
"loss": 0.2475,
"step": 34455
},
{
"epoch": 4.018190298507463,
"grad_norm": 0.4941333458896794,
"learning_rate": 9.579363834365484e-06,
"loss": 0.2468,
"step": 34460
},
{
"epoch": 4.018773320895522,
"grad_norm": 0.5256476046275693,
"learning_rate": 9.574118957155321e-06,
"loss": 0.2653,
"step": 34465
},
{
"epoch": 4.019356343283582,
"grad_norm": 0.4948114399941902,
"learning_rate": 9.568876745351919e-06,
"loss": 0.2585,
"step": 34470
},
{
"epoch": 4.019939365671642,
"grad_norm": 0.5382404439562486,
"learning_rate": 9.563637199734744e-06,
"loss": 0.2647,
"step": 34475
},
{
"epoch": 4.020522388059701,
"grad_norm": 1.3552595252469355,
"learning_rate": 9.558400321082863e-06,
"loss": 0.2543,
"step": 34480
},
{
"epoch": 4.0211054104477615,
"grad_norm": 0.4937450707308806,
"learning_rate": 9.553166110174957e-06,
"loss": 0.2652,
"step": 34485
},
{
"epoch": 4.021688432835821,
"grad_norm": 0.5103417878634484,
"learning_rate": 9.547934567789302e-06,
"loss": 0.2631,
"step": 34490
},
{
"epoch": 4.022271455223881,
"grad_norm": 0.5142267726205416,
"learning_rate": 9.542705694703763e-06,
"loss": 0.2681,
"step": 34495
},
{
"epoch": 4.02285447761194,
"grad_norm": 0.5602783254589139,
"learning_rate": 9.537479491695845e-06,
"loss": 0.2621,
"step": 34500
},
{
"epoch": 4.0234375,
"grad_norm": 0.5185598440764575,
"learning_rate": 9.532255959542616e-06,
"loss": 0.2651,
"step": 34505
},
{
"epoch": 4.02402052238806,
"grad_norm": 0.488475073935665,
"learning_rate": 9.527035099020784e-06,
"loss": 0.2541,
"step": 34510
},
{
"epoch": 4.024603544776119,
"grad_norm": 0.516711784846163,
"learning_rate": 9.521816910906626e-06,
"loss": 0.2579,
"step": 34515
},
{
"epoch": 4.025186567164179,
"grad_norm": 0.5128036867723536,
"learning_rate": 9.516601395976038e-06,
"loss": 0.2706,
"step": 34520
},
{
"epoch": 4.0257695895522385,
"grad_norm": 0.4998739514067558,
"learning_rate": 9.511388555004523e-06,
"loss": 0.2674,
"step": 34525
},
{
"epoch": 4.026352611940299,
"grad_norm": 0.48745000252020015,
"learning_rate": 9.506178388767176e-06,
"loss": 0.2447,
"step": 34530
},
{
"epoch": 4.026935634328358,
"grad_norm": 0.5045935369935165,
"learning_rate": 9.5009708980387e-06,
"loss": 0.2447,
"step": 34535
},
{
"epoch": 4.027518656716418,
"grad_norm": 0.5237258930097034,
"learning_rate": 9.495766083593407e-06,
"loss": 0.2598,
"step": 34540
},
{
"epoch": 4.028101679104478,
"grad_norm": 0.4925259908992184,
"learning_rate": 9.490563946205183e-06,
"loss": 0.2577,
"step": 34545
},
{
"epoch": 4.028684701492537,
"grad_norm": 0.5131302100958036,
"learning_rate": 9.485364486647561e-06,
"loss": 0.2569,
"step": 34550
},
{
"epoch": 4.029267723880597,
"grad_norm": 0.5145603467207116,
"learning_rate": 9.480167705693624e-06,
"loss": 0.2496,
"step": 34555
},
{
"epoch": 4.029850746268656,
"grad_norm": 0.5158991442898166,
"learning_rate": 9.474973604116112e-06,
"loss": 0.2634,
"step": 34560
},
{
"epoch": 4.030433768656716,
"grad_norm": 0.4986558917596338,
"learning_rate": 9.469782182687317e-06,
"loss": 0.2552,
"step": 34565
},
{
"epoch": 4.0310167910447765,
"grad_norm": 0.5042877087851613,
"learning_rate": 9.464593442179162e-06,
"loss": 0.2552,
"step": 34570
},
{
"epoch": 4.031599813432836,
"grad_norm": 0.48701866303121916,
"learning_rate": 9.459407383363158e-06,
"loss": 0.3002,
"step": 34575
},
{
"epoch": 4.032182835820896,
"grad_norm": 0.5834265076023694,
"learning_rate": 9.454224007010428e-06,
"loss": 0.2663,
"step": 34580
},
{
"epoch": 4.032765858208955,
"grad_norm": 0.5239221149577419,
"learning_rate": 9.449043313891692e-06,
"loss": 0.2638,
"step": 34585
},
{
"epoch": 4.033348880597015,
"grad_norm": 0.5236996970874352,
"learning_rate": 9.443865304777266e-06,
"loss": 0.2383,
"step": 34590
},
{
"epoch": 4.033931902985074,
"grad_norm": 0.49535759671095636,
"learning_rate": 9.438689980437062e-06,
"loss": 0.2517,
"step": 34595
},
{
"epoch": 4.034514925373134,
"grad_norm": 0.5470458516533174,
"learning_rate": 9.433517341640621e-06,
"loss": 0.2567,
"step": 34600
},
{
"epoch": 4.035097947761194,
"grad_norm": 0.48917462514613874,
"learning_rate": 9.428347389157039e-06,
"loss": 0.2603,
"step": 34605
},
{
"epoch": 4.0356809701492535,
"grad_norm": 0.5428754555939528,
"learning_rate": 9.423180123755064e-06,
"loss": 0.2545,
"step": 34610
},
{
"epoch": 4.036263992537314,
"grad_norm": 0.5234581583690686,
"learning_rate": 9.418015546203002e-06,
"loss": 0.2468,
"step": 34615
},
{
"epoch": 4.036847014925373,
"grad_norm": 0.4985094927754438,
"learning_rate": 9.41285365726878e-06,
"loss": 0.2669,
"step": 34620
},
{
"epoch": 4.037430037313433,
"grad_norm": 0.5568416674224604,
"learning_rate": 9.407694457719925e-06,
"loss": 0.2747,
"step": 34625
},
{
"epoch": 4.038013059701493,
"grad_norm": 0.5069542080613687,
"learning_rate": 9.40253794832356e-06,
"loss": 0.2582,
"step": 34630
},
{
"epoch": 4.038596082089552,
"grad_norm": 0.5699199513363726,
"learning_rate": 9.397384129846404e-06,
"loss": 0.2633,
"step": 34635
},
{
"epoch": 4.039179104477612,
"grad_norm": 0.5138557167585569,
"learning_rate": 9.39223300305479e-06,
"loss": 0.2449,
"step": 34640
},
{
"epoch": 4.039762126865671,
"grad_norm": 0.5114334898282349,
"learning_rate": 9.387084568714628e-06,
"loss": 0.2666,
"step": 34645
},
{
"epoch": 4.0403451492537314,
"grad_norm": 0.5581905434816621,
"learning_rate": 9.381938827591447e-06,
"loss": 0.2644,
"step": 34650
},
{
"epoch": 4.040928171641791,
"grad_norm": 0.5073140461448422,
"learning_rate": 9.376795780450373e-06,
"loss": 0.2624,
"step": 34655
},
{
"epoch": 4.041511194029851,
"grad_norm": 0.5021790408826153,
"learning_rate": 9.371655428056122e-06,
"loss": 0.2499,
"step": 34660
},
{
"epoch": 4.042094216417911,
"grad_norm": 0.5024641361719697,
"learning_rate": 9.36651777117302e-06,
"loss": 0.2525,
"step": 34665
},
{
"epoch": 4.04267723880597,
"grad_norm": 0.5461884943720894,
"learning_rate": 9.361382810564984e-06,
"loss": 0.263,
"step": 34670
},
{
"epoch": 4.04326026119403,
"grad_norm": 0.5027030776687684,
"learning_rate": 9.35625054699554e-06,
"loss": 0.2728,
"step": 34675
},
{
"epoch": 4.043843283582089,
"grad_norm": 0.5268213655634778,
"learning_rate": 9.351120981227788e-06,
"loss": 0.254,
"step": 34680
},
{
"epoch": 4.044426305970149,
"grad_norm": 0.49028374930341145,
"learning_rate": 9.345994114024472e-06,
"loss": 0.2585,
"step": 34685
},
{
"epoch": 4.045009328358209,
"grad_norm": 0.5218721168480488,
"learning_rate": 9.34086994614789e-06,
"loss": 0.2575,
"step": 34690
},
{
"epoch": 4.0455923507462686,
"grad_norm": 0.5264320449452657,
"learning_rate": 9.33574847835996e-06,
"loss": 0.262,
"step": 34695
},
{
"epoch": 4.046175373134329,
"grad_norm": 0.5478511395750972,
"learning_rate": 9.330629711422196e-06,
"loss": 0.2543,
"step": 34700
},
{
"epoch": 4.046758395522388,
"grad_norm": 0.5732543892508573,
"learning_rate": 9.325513646095707e-06,
"loss": 0.2595,
"step": 34705
},
{
"epoch": 4.047341417910448,
"grad_norm": 0.5335244097695253,
"learning_rate": 9.320400283141208e-06,
"loss": 0.2703,
"step": 34710
},
{
"epoch": 4.047924440298507,
"grad_norm": 0.5026202484787695,
"learning_rate": 9.315289623319012e-06,
"loss": 0.2642,
"step": 34715
},
{
"epoch": 4.048507462686567,
"grad_norm": 0.5292062634479631,
"learning_rate": 9.310181667389003e-06,
"loss": 0.2591,
"step": 34720
},
{
"epoch": 4.049090485074627,
"grad_norm": 0.5535990342992809,
"learning_rate": 9.305076416110715e-06,
"loss": 0.2603,
"step": 34725
},
{
"epoch": 4.049673507462686,
"grad_norm": 0.5069663116362183,
"learning_rate": 9.299973870243222e-06,
"loss": 0.245,
"step": 34730
},
{
"epoch": 4.0502565298507465,
"grad_norm": 0.5701850619343685,
"learning_rate": 9.294874030545247e-06,
"loss": 0.2581,
"step": 34735
},
{
"epoch": 4.050839552238806,
"grad_norm": 0.5031707224572886,
"learning_rate": 9.289776897775074e-06,
"loss": 0.2623,
"step": 34740
},
{
"epoch": 4.051422574626866,
"grad_norm": 0.5023734778381901,
"learning_rate": 9.284682472690599e-06,
"loss": 0.2477,
"step": 34745
},
{
"epoch": 4.052005597014926,
"grad_norm": 0.515147751103087,
"learning_rate": 9.279590756049316e-06,
"loss": 0.2675,
"step": 34750
},
{
"epoch": 4.052588619402985,
"grad_norm": 0.5423363249426748,
"learning_rate": 9.274501748608314e-06,
"loss": 0.2649,
"step": 34755
},
{
"epoch": 4.053171641791045,
"grad_norm": 0.5474114394899581,
"learning_rate": 9.269415451124283e-06,
"loss": 0.2551,
"step": 34760
},
{
"epoch": 4.053754664179104,
"grad_norm": 0.5277244774771899,
"learning_rate": 9.2643318643535e-06,
"loss": 0.2452,
"step": 34765
},
{
"epoch": 4.054337686567164,
"grad_norm": 0.5352283483668832,
"learning_rate": 9.25925098905185e-06,
"loss": 0.2596,
"step": 34770
},
{
"epoch": 4.0549207089552235,
"grad_norm": 0.5429303684414251,
"learning_rate": 9.254172825974823e-06,
"loss": 0.2599,
"step": 34775
},
{
"epoch": 4.055503731343284,
"grad_norm": 0.5139700143368325,
"learning_rate": 9.249097375877458e-06,
"loss": 0.2434,
"step": 34780
},
{
"epoch": 4.056086753731344,
"grad_norm": 0.50005271179117,
"learning_rate": 9.244024639514465e-06,
"loss": 0.252,
"step": 34785
},
{
"epoch": 4.056669776119403,
"grad_norm": 0.4997944637264341,
"learning_rate": 9.23895461764009e-06,
"loss": 0.2511,
"step": 34790
},
{
"epoch": 4.057252798507463,
"grad_norm": 0.49561392020243344,
"learning_rate": 9.233887311008197e-06,
"loss": 0.2654,
"step": 34795
},
{
"epoch": 4.057835820895522,
"grad_norm": 0.5220940410770386,
"learning_rate": 9.22882272037225e-06,
"loss": 0.2624,
"step": 34800
},
{
"epoch": 4.058418843283582,
"grad_norm": 0.49265761402501357,
"learning_rate": 9.223760846485307e-06,
"loss": 0.2577,
"step": 34805
},
{
"epoch": 4.059001865671642,
"grad_norm": 0.5295448198844314,
"learning_rate": 9.218701690100017e-06,
"loss": 0.2806,
"step": 34810
},
{
"epoch": 4.059584888059701,
"grad_norm": 0.4915080457248284,
"learning_rate": 9.21364525196863e-06,
"loss": 0.2506,
"step": 34815
},
{
"epoch": 4.0601679104477615,
"grad_norm": 0.5096665029389867,
"learning_rate": 9.208591532842995e-06,
"loss": 0.2594,
"step": 34820
},
{
"epoch": 4.060750932835821,
"grad_norm": 0.582525339114675,
"learning_rate": 9.203540533474537e-06,
"loss": 0.2667,
"step": 34825
},
{
"epoch": 4.061333955223881,
"grad_norm": 0.5878438127457113,
"learning_rate": 9.198492254614302e-06,
"loss": 0.2566,
"step": 34830
},
{
"epoch": 4.06191697761194,
"grad_norm": 0.5481312822590019,
"learning_rate": 9.193446697012921e-06,
"loss": 0.2664,
"step": 34835
},
{
"epoch": 4.0625,
"grad_norm": 0.5204472798397481,
"learning_rate": 9.188403861420615e-06,
"loss": 0.2755,
"step": 34840
},
{
"epoch": 4.06308302238806,
"grad_norm": 0.6124691704600715,
"learning_rate": 9.183363748587207e-06,
"loss": 0.2847,
"step": 34845
},
{
"epoch": 4.063666044776119,
"grad_norm": 0.5240735363343201,
"learning_rate": 9.178326359262124e-06,
"loss": 0.2408,
"step": 34850
},
{
"epoch": 4.064249067164179,
"grad_norm": 0.5162043833097014,
"learning_rate": 9.173291694194356e-06,
"loss": 0.2636,
"step": 34855
},
{
"epoch": 4.0648320895522385,
"grad_norm": 0.4776228176419194,
"learning_rate": 9.16825975413253e-06,
"loss": 0.2685,
"step": 34860
},
{
"epoch": 4.065415111940299,
"grad_norm": 0.5115555275572906,
"learning_rate": 9.163230539824829e-06,
"loss": 0.2666,
"step": 34865
},
{
"epoch": 4.065998134328359,
"grad_norm": 0.4938491423241653,
"learning_rate": 9.158204052019069e-06,
"loss": 0.2576,
"step": 34870
},
{
"epoch": 4.066581156716418,
"grad_norm": 0.5304078190670312,
"learning_rate": 9.153180291462627e-06,
"loss": 0.2563,
"step": 34875
},
{
"epoch": 4.067164179104478,
"grad_norm": 0.5088828115285172,
"learning_rate": 9.148159258902488e-06,
"loss": 0.2629,
"step": 34880
},
{
"epoch": 4.067747201492537,
"grad_norm": 0.5073729369251818,
"learning_rate": 9.143140955085239e-06,
"loss": 0.2438,
"step": 34885
},
{
"epoch": 4.068330223880597,
"grad_norm": 0.5083760831908246,
"learning_rate": 9.138125380757046e-06,
"loss": 0.2589,
"step": 34890
},
{
"epoch": 4.068913246268656,
"grad_norm": 0.4919762102173519,
"learning_rate": 9.133112536663682e-06,
"loss": 0.2452,
"step": 34895
},
{
"epoch": 4.069496268656716,
"grad_norm": 0.507306349249229,
"learning_rate": 9.128102423550511e-06,
"loss": 0.2649,
"step": 34900
},
{
"epoch": 4.0700792910447765,
"grad_norm": 0.5389362937652415,
"learning_rate": 9.123095042162477e-06,
"loss": 0.2725,
"step": 34905
},
{
"epoch": 4.070662313432836,
"grad_norm": 0.5657063443375089,
"learning_rate": 9.118090393244147e-06,
"loss": 0.2694,
"step": 34910
},
{
"epoch": 4.071245335820896,
"grad_norm": 0.5223457192265057,
"learning_rate": 9.113088477539643e-06,
"loss": 0.2585,
"step": 34915
},
{
"epoch": 4.071828358208955,
"grad_norm": 0.5630649644285136,
"learning_rate": 9.108089295792726e-06,
"loss": 0.2672,
"step": 34920
},
{
"epoch": 4.072411380597015,
"grad_norm": 0.4750214373260717,
"learning_rate": 9.10309284874671e-06,
"loss": 0.2511,
"step": 34925
},
{
"epoch": 4.072994402985074,
"grad_norm": 0.4986075941132954,
"learning_rate": 9.098099137144522e-06,
"loss": 0.2537,
"step": 34930
},
{
"epoch": 4.073577425373134,
"grad_norm": 0.5086579097697231,
"learning_rate": 9.093108161728683e-06,
"loss": 0.2789,
"step": 34935
},
{
"epoch": 4.074160447761194,
"grad_norm": 0.4987041903205375,
"learning_rate": 9.088119923241295e-06,
"loss": 0.2557,
"step": 34940
},
{
"epoch": 4.0747434701492535,
"grad_norm": 0.5132993305722506,
"learning_rate": 9.083134422424073e-06,
"loss": 0.2746,
"step": 34945
},
{
"epoch": 4.075326492537314,
"grad_norm": 0.510535415784068,
"learning_rate": 9.07815166001831e-06,
"loss": 0.256,
"step": 34950
},
{
"epoch": 4.075909514925373,
"grad_norm": 0.4978048118559493,
"learning_rate": 9.073171636764879e-06,
"loss": 0.2548,
"step": 34955
},
{
"epoch": 4.076492537313433,
"grad_norm": 0.5377280077379235,
"learning_rate": 9.068194353404288e-06,
"loss": 0.275,
"step": 34960
},
{
"epoch": 4.077075559701493,
"grad_norm": 0.5426270596291648,
"learning_rate": 9.063219810676593e-06,
"loss": 0.2599,
"step": 34965
},
{
"epoch": 4.077658582089552,
"grad_norm": 0.505525866696443,
"learning_rate": 9.058248009321464e-06,
"loss": 0.2644,
"step": 34970
},
{
"epoch": 4.078241604477612,
"grad_norm": 0.4948310624798179,
"learning_rate": 9.053278950078163e-06,
"loss": 0.2516,
"step": 34975
},
{
"epoch": 4.078824626865671,
"grad_norm": 0.5417889814619405,
"learning_rate": 9.04831263368554e-06,
"loss": 0.2514,
"step": 34980
},
{
"epoch": 4.0794076492537314,
"grad_norm": 0.5249793428135052,
"learning_rate": 9.043349060882039e-06,
"loss": 0.2631,
"step": 34985
},
{
"epoch": 4.079990671641791,
"grad_norm": 0.5414372285067178,
"learning_rate": 9.038388232405699e-06,
"loss": 0.2514,
"step": 34990
},
{
"epoch": 4.080573694029851,
"grad_norm": 0.5426317620989723,
"learning_rate": 9.033430148994148e-06,
"loss": 0.2619,
"step": 34995
},
{
"epoch": 4.081156716417911,
"grad_norm": 0.5269626295122668,
"learning_rate": 9.028474811384597e-06,
"loss": 0.2587,
"step": 35000
},
{
"epoch": 4.08173973880597,
"grad_norm": 0.5347272041813942,
"learning_rate": 9.023522220313865e-06,
"loss": 0.2422,
"step": 35005
},
{
"epoch": 4.08232276119403,
"grad_norm": 0.5424201318175285,
"learning_rate": 9.01857237651835e-06,
"loss": 0.2706,
"step": 35010
},
{
"epoch": 4.082905783582089,
"grad_norm": 0.5145158271415362,
"learning_rate": 9.013625280734047e-06,
"loss": 0.269,
"step": 35015
},
{
"epoch": 4.083488805970149,
"grad_norm": 0.5478132605753813,
"learning_rate": 9.008680933696545e-06,
"loss": 0.2797,
"step": 35020
},
{
"epoch": 4.084071828358209,
"grad_norm": 0.523611251601591,
"learning_rate": 9.003739336141025e-06,
"loss": 0.2642,
"step": 35025
},
{
"epoch": 4.0846548507462686,
"grad_norm": 0.5363592239524483,
"learning_rate": 8.998800488802239e-06,
"loss": 0.2656,
"step": 35030
},
{
"epoch": 4.085237873134329,
"grad_norm": 1.8497445741110459,
"learning_rate": 8.99386439241457e-06,
"loss": 0.2652,
"step": 35035
},
{
"epoch": 4.085820895522388,
"grad_norm": 0.5228606562217458,
"learning_rate": 8.98893104771194e-06,
"loss": 0.2652,
"step": 35040
},
{
"epoch": 4.086403917910448,
"grad_norm": 0.5169279851288193,
"learning_rate": 8.984000455427917e-06,
"loss": 0.2529,
"step": 35045
},
{
"epoch": 4.086986940298507,
"grad_norm": 0.5121517409717713,
"learning_rate": 8.979072616295616e-06,
"loss": 0.2728,
"step": 35050
},
{
"epoch": 4.087569962686567,
"grad_norm": 0.5454166426967931,
"learning_rate": 8.974147531047763e-06,
"loss": 0.2618,
"step": 35055
},
{
"epoch": 4.088152985074627,
"grad_norm": 0.5288732534324025,
"learning_rate": 8.969225200416678e-06,
"loss": 0.2529,
"step": 35060
},
{
"epoch": 4.088736007462686,
"grad_norm": 0.5261887860723182,
"learning_rate": 8.964305625134254e-06,
"loss": 0.2591,
"step": 35065
},
{
"epoch": 4.0893190298507465,
"grad_norm": 0.5270812824511006,
"learning_rate": 8.959388805931993e-06,
"loss": 0.2765,
"step": 35070
},
{
"epoch": 4.089902052238806,
"grad_norm": 0.5693926518404087,
"learning_rate": 8.954474743540979e-06,
"loss": 0.2689,
"step": 35075
},
{
"epoch": 4.090485074626866,
"grad_norm": 0.5230461666827151,
"learning_rate": 8.94956343869187e-06,
"loss": 0.2535,
"step": 35080
},
{
"epoch": 4.091068097014926,
"grad_norm": 0.5007784450998025,
"learning_rate": 8.944654892114956e-06,
"loss": 0.2444,
"step": 35085
},
{
"epoch": 4.091651119402985,
"grad_norm": 0.5004196138596179,
"learning_rate": 8.939749104540065e-06,
"loss": 0.2274,
"step": 35090
},
{
"epoch": 4.092234141791045,
"grad_norm": 0.49148938073063914,
"learning_rate": 8.934846076696665e-06,
"loss": 0.2554,
"step": 35095
},
{
"epoch": 4.092817164179104,
"grad_norm": 0.5472066988355871,
"learning_rate": 8.929945809313773e-06,
"loss": 0.2568,
"step": 35100
},
{
"epoch": 4.093400186567164,
"grad_norm": 0.4833868095424993,
"learning_rate": 8.925048303120012e-06,
"loss": 0.234,
"step": 35105
},
{
"epoch": 4.0939832089552235,
"grad_norm": 0.49536379685887266,
"learning_rate": 8.9201535588436e-06,
"loss": 0.2684,
"step": 35110
},
{
"epoch": 4.094566231343284,
"grad_norm": 0.47392014318420916,
"learning_rate": 8.915261577212337e-06,
"loss": 0.2445,
"step": 35115
},
{
"epoch": 4.095149253731344,
"grad_norm": 0.46564338813188333,
"learning_rate": 8.910372358953614e-06,
"loss": 0.2591,
"step": 35120
},
{
"epoch": 4.095732276119403,
"grad_norm": 0.49233180038956115,
"learning_rate": 8.905485904794416e-06,
"loss": 0.2489,
"step": 35125
},
{
"epoch": 4.096315298507463,
"grad_norm": 0.5479573179986573,
"learning_rate": 8.900602215461297e-06,
"loss": 0.2444,
"step": 35130
},
{
"epoch": 4.096898320895522,
"grad_norm": 0.6122752404359795,
"learning_rate": 8.895721291680433e-06,
"loss": 0.2621,
"step": 35135
},
{
"epoch": 4.097481343283582,
"grad_norm": 0.5597183955613113,
"learning_rate": 8.890843134177555e-06,
"loss": 0.2639,
"step": 35140
},
{
"epoch": 4.098064365671641,
"grad_norm": 0.5315320998410075,
"learning_rate": 8.885967743678011e-06,
"loss": 0.2656,
"step": 35145
},
{
"epoch": 4.098647388059701,
"grad_norm": 0.5353394340704639,
"learning_rate": 8.881095120906716e-06,
"loss": 0.2561,
"step": 35150
},
{
"epoch": 4.0992304104477615,
"grad_norm": 0.5291361825474706,
"learning_rate": 8.876225266588184e-06,
"loss": 0.2519,
"step": 35155
},
{
"epoch": 4.099813432835821,
"grad_norm": 0.5036639626548466,
"learning_rate": 8.871358181446519e-06,
"loss": 0.2467,
"step": 35160
},
{
"epoch": 4.100396455223881,
"grad_norm": 0.47752774448835666,
"learning_rate": 8.866493866205407e-06,
"loss": 0.2632,
"step": 35165
},
{
"epoch": 4.10097947761194,
"grad_norm": 0.5115050932562567,
"learning_rate": 8.861632321588126e-06,
"loss": 0.2542,
"step": 35170
},
{
"epoch": 4.1015625,
"grad_norm": 0.49680977551903693,
"learning_rate": 8.856773548317545e-06,
"loss": 0.2761,
"step": 35175
},
{
"epoch": 4.10214552238806,
"grad_norm": 0.5129949691060091,
"learning_rate": 8.851917547116111e-06,
"loss": 0.2524,
"step": 35180
},
{
"epoch": 4.102728544776119,
"grad_norm": 0.5144168250882896,
"learning_rate": 8.847064318705864e-06,
"loss": 0.2616,
"step": 35185
},
{
"epoch": 4.103311567164179,
"grad_norm": 0.48219095547975166,
"learning_rate": 8.842213863808439e-06,
"loss": 0.2734,
"step": 35190
},
{
"epoch": 4.1038945895522385,
"grad_norm": 0.5426833429370523,
"learning_rate": 8.837366183145044e-06,
"loss": 0.2586,
"step": 35195
},
{
"epoch": 4.104477611940299,
"grad_norm": 0.5278663639380166,
"learning_rate": 8.83252127743649e-06,
"loss": 0.2507,
"step": 35200
},
{
"epoch": 4.105060634328359,
"grad_norm": 0.5078270817437557,
"learning_rate": 8.827679147403167e-06,
"loss": 0.2771,
"step": 35205
},
{
"epoch": 4.105643656716418,
"grad_norm": 0.47929031798955013,
"learning_rate": 8.822839793765056e-06,
"loss": 0.2546,
"step": 35210
},
{
"epoch": 4.106226679104478,
"grad_norm": 0.5516138359484303,
"learning_rate": 8.818003217241707e-06,
"loss": 0.2689,
"step": 35215
},
{
"epoch": 4.106809701492537,
"grad_norm": 0.5316918362563171,
"learning_rate": 8.813169418552294e-06,
"loss": 0.2601,
"step": 35220
},
{
"epoch": 4.107392723880597,
"grad_norm": 0.5002464993683433,
"learning_rate": 8.808338398415544e-06,
"loss": 0.2669,
"step": 35225
},
{
"epoch": 4.107975746268656,
"grad_norm": 0.5432643899604875,
"learning_rate": 8.803510157549785e-06,
"loss": 0.2606,
"step": 35230
},
{
"epoch": 4.108558768656716,
"grad_norm": 0.49159574710362086,
"learning_rate": 8.79868469667293e-06,
"loss": 0.2455,
"step": 35235
},
{
"epoch": 4.1091417910447765,
"grad_norm": 0.5350433730708042,
"learning_rate": 8.793862016502477e-06,
"loss": 0.2635,
"step": 35240
},
{
"epoch": 4.109724813432836,
"grad_norm": 0.591245801575081,
"learning_rate": 8.789042117755521e-06,
"loss": 0.2566,
"step": 35245
},
{
"epoch": 4.110307835820896,
"grad_norm": 0.5652227713274541,
"learning_rate": 8.78422500114873e-06,
"loss": 0.2586,
"step": 35250
},
{
"epoch": 4.110890858208955,
"grad_norm": 0.49119906635330823,
"learning_rate": 8.779410667398352e-06,
"loss": 0.2692,
"step": 35255
},
{
"epoch": 4.111473880597015,
"grad_norm": 0.48519076071323763,
"learning_rate": 8.774599117220254e-06,
"loss": 0.2637,
"step": 35260
},
{
"epoch": 4.112056902985074,
"grad_norm": 0.4868122980881773,
"learning_rate": 8.769790351329847e-06,
"loss": 0.249,
"step": 35265
},
{
"epoch": 4.112639925373134,
"grad_norm": 0.5577564289977741,
"learning_rate": 8.764984370442166e-06,
"loss": 0.2828,
"step": 35270
},
{
"epoch": 4.113222947761194,
"grad_norm": 0.5479227529823502,
"learning_rate": 8.7601811752718e-06,
"loss": 0.2436,
"step": 35275
},
{
"epoch": 4.1138059701492535,
"grad_norm": 0.4942580182722624,
"learning_rate": 8.755380766532945e-06,
"loss": 0.2588,
"step": 35280
},
{
"epoch": 4.114388992537314,
"grad_norm": 0.5452136170015037,
"learning_rate": 8.750583144939373e-06,
"loss": 0.2747,
"step": 35285
},
{
"epoch": 4.114972014925373,
"grad_norm": 0.48304799967687756,
"learning_rate": 8.745788311204444e-06,
"loss": 0.2428,
"step": 35290
},
{
"epoch": 4.115555037313433,
"grad_norm": 0.538047653761096,
"learning_rate": 8.740996266041108e-06,
"loss": 0.2612,
"step": 35295
},
{
"epoch": 4.116138059701493,
"grad_norm": 0.5271145665435983,
"learning_rate": 8.736207010161899e-06,
"loss": 0.2672,
"step": 35300
},
{
"epoch": 4.116721082089552,
"grad_norm": 0.5181804869567275,
"learning_rate": 8.731420544278913e-06,
"loss": 0.2543,
"step": 35305
},
{
"epoch": 4.117304104477612,
"grad_norm": 0.5048284816398966,
"learning_rate": 8.726636869103884e-06,
"loss": 0.2536,
"step": 35310
},
{
"epoch": 4.117887126865671,
"grad_norm": 0.485744198870478,
"learning_rate": 8.721855985348068e-06,
"loss": 0.2475,
"step": 35315
},
{
"epoch": 4.1184701492537314,
"grad_norm": 0.5223579357266295,
"learning_rate": 8.71707789372236e-06,
"loss": 0.2602,
"step": 35320
},
{
"epoch": 4.119053171641791,
"grad_norm": 0.5097406934491621,
"learning_rate": 8.712302594937202e-06,
"loss": 0.2492,
"step": 35325
},
{
"epoch": 4.119636194029851,
"grad_norm": 0.5365099424282715,
"learning_rate": 8.70753008970264e-06,
"loss": 0.2719,
"step": 35330
},
{
"epoch": 4.120219216417911,
"grad_norm": 0.4953737631268252,
"learning_rate": 8.7027603787283e-06,
"loss": 0.2667,
"step": 35335
},
{
"epoch": 4.12080223880597,
"grad_norm": 0.47923786373376404,
"learning_rate": 8.697993462723392e-06,
"loss": 0.2663,
"step": 35340
},
{
"epoch": 4.12138526119403,
"grad_norm": 0.49940924183785756,
"learning_rate": 8.69322934239671e-06,
"loss": 0.2481,
"step": 35345
},
{
"epoch": 4.121968283582089,
"grad_norm": 0.5273696178399012,
"learning_rate": 8.688468018456639e-06,
"loss": 0.243,
"step": 35350
},
{
"epoch": 4.122551305970149,
"grad_norm": 0.5667749246880608,
"learning_rate": 8.68370949161113e-06,
"loss": 0.2579,
"step": 35355
},
{
"epoch": 4.123134328358209,
"grad_norm": 0.5426726008060032,
"learning_rate": 8.678953762567739e-06,
"loss": 0.2662,
"step": 35360
},
{
"epoch": 4.1237173507462686,
"grad_norm": 0.5864143060614948,
"learning_rate": 8.674200832033595e-06,
"loss": 0.2526,
"step": 35365
},
{
"epoch": 4.124300373134329,
"grad_norm": 0.5247040389500216,
"learning_rate": 8.669450700715414e-06,
"loss": 0.2562,
"step": 35370
},
{
"epoch": 4.124883395522388,
"grad_norm": 0.534521634852648,
"learning_rate": 8.664703369319496e-06,
"loss": 0.2436,
"step": 35375
},
{
"epoch": 4.125466417910448,
"grad_norm": 0.5334707822368108,
"learning_rate": 8.659958838551722e-06,
"loss": 0.2587,
"step": 35380
},
{
"epoch": 4.126049440298507,
"grad_norm": 0.4776951742348572,
"learning_rate": 8.655217109117564e-06,
"loss": 0.2417,
"step": 35385
},
{
"epoch": 4.126632462686567,
"grad_norm": 0.5784760840471114,
"learning_rate": 8.650478181722055e-06,
"loss": 0.2829,
"step": 35390
},
{
"epoch": 4.127215485074627,
"grad_norm": 0.5494177092041136,
"learning_rate": 8.64574205706985e-06,
"loss": 0.2563,
"step": 35395
},
{
"epoch": 4.127798507462686,
"grad_norm": 0.530356170485387,
"learning_rate": 8.641008735865153e-06,
"loss": 0.2688,
"step": 35400
},
{
"epoch": 4.1283815298507465,
"grad_norm": 0.5049060522300168,
"learning_rate": 8.63627821881176e-06,
"loss": 0.2536,
"step": 35405
},
{
"epoch": 4.128964552238806,
"grad_norm": 0.5553285155181097,
"learning_rate": 8.631550506613062e-06,
"loss": 0.2663,
"step": 35410
},
{
"epoch": 4.129547574626866,
"grad_norm": 0.52838836040884,
"learning_rate": 8.626825599972022e-06,
"loss": 0.2538,
"step": 35415
},
{
"epoch": 4.130130597014926,
"grad_norm": 0.4762703054443566,
"learning_rate": 8.62210349959119e-06,
"loss": 0.2521,
"step": 35420
},
{
"epoch": 4.130713619402985,
"grad_norm": 0.5301260185468138,
"learning_rate": 8.617384206172696e-06,
"loss": 0.267,
"step": 35425
},
{
"epoch": 4.131296641791045,
"grad_norm": 0.5584153545755062,
"learning_rate": 8.612667720418243e-06,
"loss": 0.2513,
"step": 35430
},
{
"epoch": 4.131879664179104,
"grad_norm": 0.5203538176060201,
"learning_rate": 8.60795404302915e-06,
"loss": 0.2555,
"step": 35435
},
{
"epoch": 4.132462686567164,
"grad_norm": 0.5207548491855819,
"learning_rate": 8.60324317470627e-06,
"loss": 0.2606,
"step": 35440
},
{
"epoch": 4.1330457089552235,
"grad_norm": 0.5627320403489383,
"learning_rate": 8.598535116150086e-06,
"loss": 0.255,
"step": 35445
},
{
"epoch": 4.133628731343284,
"grad_norm": 0.5160085587749553,
"learning_rate": 8.593829868060632e-06,
"loss": 0.2479,
"step": 35450
},
{
"epoch": 4.134211753731344,
"grad_norm": 0.4992461902352645,
"learning_rate": 8.589127431137527e-06,
"loss": 0.2738,
"step": 35455
},
{
"epoch": 4.134794776119403,
"grad_norm": 0.5119714860980036,
"learning_rate": 8.584427806079988e-06,
"loss": 0.2449,
"step": 35460
},
{
"epoch": 4.135377798507463,
"grad_norm": 0.5312475240753449,
"learning_rate": 8.579730993586798e-06,
"loss": 0.2657,
"step": 35465
},
{
"epoch": 4.135960820895522,
"grad_norm": 0.5619289480912323,
"learning_rate": 8.575036994356334e-06,
"loss": 0.2599,
"step": 35470
},
{
"epoch": 4.136543843283582,
"grad_norm": 0.552781408980997,
"learning_rate": 8.570345809086543e-06,
"loss": 0.2507,
"step": 35475
},
{
"epoch": 4.137126865671641,
"grad_norm": 0.533390892931154,
"learning_rate": 8.565657438474963e-06,
"loss": 0.271,
"step": 35480
},
{
"epoch": 4.137709888059701,
"grad_norm": 0.5415315416003454,
"learning_rate": 8.560971883218714e-06,
"loss": 0.261,
"step": 35485
},
{
"epoch": 4.1382929104477615,
"grad_norm": 0.5073952471708189,
"learning_rate": 8.556289144014474e-06,
"loss": 0.2433,
"step": 35490
},
{
"epoch": 4.138875932835821,
"grad_norm": 0.5165418492461755,
"learning_rate": 8.551609221558548e-06,
"loss": 0.2542,
"step": 35495
},
{
"epoch": 4.139458955223881,
"grad_norm": 0.5293658484286743,
"learning_rate": 8.546932116546775e-06,
"loss": 0.2742,
"step": 35500
},
{
"epoch": 4.14004197761194,
"grad_norm": 0.5646742393884477,
"learning_rate": 8.542257829674608e-06,
"loss": 0.2482,
"step": 35505
},
{
"epoch": 4.140625,
"grad_norm": 0.5458670461221791,
"learning_rate": 8.53758636163706e-06,
"loss": 0.2517,
"step": 35510
},
{
"epoch": 4.14120802238806,
"grad_norm": 0.533126428830021,
"learning_rate": 8.53291771312874e-06,
"loss": 0.256,
"step": 35515
},
{
"epoch": 4.141791044776119,
"grad_norm": 0.5984955716892207,
"learning_rate": 8.528251884843829e-06,
"loss": 0.2604,
"step": 35520
},
{
"epoch": 4.142374067164179,
"grad_norm": 0.5401377877161797,
"learning_rate": 8.523588877476089e-06,
"loss": 0.2781,
"step": 35525
},
{
"epoch": 4.1429570895522385,
"grad_norm": 0.5158414589426041,
"learning_rate": 8.518928691718872e-06,
"loss": 0.2731,
"step": 35530
},
{
"epoch": 4.143540111940299,
"grad_norm": 0.514799122132554,
"learning_rate": 8.514271328265094e-06,
"loss": 0.2579,
"step": 35535
},
{
"epoch": 4.144123134328359,
"grad_norm": 0.5085066014358727,
"learning_rate": 8.509616787807263e-06,
"loss": 0.2603,
"step": 35540
},
{
"epoch": 4.144706156716418,
"grad_norm": 0.5303468068209967,
"learning_rate": 8.504965071037465e-06,
"loss": 0.2718,
"step": 35545
},
{
"epoch": 4.145289179104478,
"grad_norm": 0.5138223155373116,
"learning_rate": 8.500316178647366e-06,
"loss": 0.2534,
"step": 35550
},
{
"epoch": 4.145872201492537,
"grad_norm": 0.5569196620903132,
"learning_rate": 8.495670111328214e-06,
"loss": 0.2574,
"step": 35555
},
{
"epoch": 4.146455223880597,
"grad_norm": 0.5760705410733047,
"learning_rate": 8.491026869770832e-06,
"loss": 0.2742,
"step": 35560
},
{
"epoch": 4.147038246268656,
"grad_norm": 0.5006517435205796,
"learning_rate": 8.486386454665621e-06,
"loss": 0.2541,
"step": 35565
},
{
"epoch": 4.147621268656716,
"grad_norm": 0.5181378253670358,
"learning_rate": 8.48174886670258e-06,
"loss": 0.247,
"step": 35570
},
{
"epoch": 4.1482042910447765,
"grad_norm": 0.494115163963477,
"learning_rate": 8.477114106571255e-06,
"loss": 0.2661,
"step": 35575
},
{
"epoch": 4.148787313432836,
"grad_norm": 0.5578647736221412,
"learning_rate": 8.472482174960808e-06,
"loss": 0.265,
"step": 35580
},
{
"epoch": 4.149370335820896,
"grad_norm": 0.5384815577421193,
"learning_rate": 8.467853072559953e-06,
"loss": 0.2485,
"step": 35585
},
{
"epoch": 4.149953358208955,
"grad_norm": 0.5278795431000908,
"learning_rate": 8.463226800056995e-06,
"loss": 0.2559,
"step": 35590
},
{
"epoch": 4.150536380597015,
"grad_norm": 0.5360810310239597,
"learning_rate": 8.458603358139818e-06,
"loss": 0.2653,
"step": 35595
},
{
"epoch": 4.151119402985074,
"grad_norm": 0.4977160792205137,
"learning_rate": 8.453982747495881e-06,
"loss": 0.2501,
"step": 35600
},
{
"epoch": 4.151702425373134,
"grad_norm": 0.536231841668788,
"learning_rate": 8.449364968812228e-06,
"loss": 0.2622,
"step": 35605
},
{
"epoch": 4.152285447761194,
"grad_norm": 0.5467799897844653,
"learning_rate": 8.44475002277548e-06,
"loss": 0.2565,
"step": 35610
},
{
"epoch": 4.1528684701492535,
"grad_norm": 0.5569443669655396,
"learning_rate": 8.440137910071821e-06,
"loss": 0.267,
"step": 35615
},
{
"epoch": 4.153451492537314,
"grad_norm": 0.4949963350018481,
"learning_rate": 8.435528631387052e-06,
"loss": 0.2521,
"step": 35620
},
{
"epoch": 4.154034514925373,
"grad_norm": 0.5487988694955066,
"learning_rate": 8.430922187406501e-06,
"loss": 0.2591,
"step": 35625
},
{
"epoch": 4.154617537313433,
"grad_norm": 0.6077030405410011,
"learning_rate": 8.426318578815128e-06,
"loss": 0.2613,
"step": 35630
},
{
"epoch": 4.155200559701493,
"grad_norm": 0.5748776319339398,
"learning_rate": 8.421717806297431e-06,
"loss": 0.2789,
"step": 35635
},
{
"epoch": 4.155783582089552,
"grad_norm": 0.5384423704349776,
"learning_rate": 8.417119870537503e-06,
"loss": 0.2573,
"step": 35640
},
{
"epoch": 4.156366604477612,
"grad_norm": 0.508814057748118,
"learning_rate": 8.41252477221901e-06,
"loss": 0.2522,
"step": 35645
},
{
"epoch": 4.156949626865671,
"grad_norm": 0.48773488979698165,
"learning_rate": 8.407932512025207e-06,
"loss": 0.2487,
"step": 35650
},
{
"epoch": 4.1575326492537314,
"grad_norm": 0.4687360391225597,
"learning_rate": 8.403343090638914e-06,
"loss": 0.2337,
"step": 35655
},
{
"epoch": 4.158115671641791,
"grad_norm": 0.5245902120953518,
"learning_rate": 8.398756508742536e-06,
"loss": 0.267,
"step": 35660
},
{
"epoch": 4.158698694029851,
"grad_norm": 0.7145286657079379,
"learning_rate": 8.394172767018048e-06,
"loss": 0.2605,
"step": 35665
},
{
"epoch": 4.159281716417911,
"grad_norm": 0.5005320632691611,
"learning_rate": 8.38959186614702e-06,
"loss": 0.2506,
"step": 35670
},
{
"epoch": 4.15986473880597,
"grad_norm": 0.5446403276226495,
"learning_rate": 8.385013806810577e-06,
"loss": 0.2688,
"step": 35675
},
{
"epoch": 4.16044776119403,
"grad_norm": 0.5128631372842963,
"learning_rate": 8.380438589689438e-06,
"loss": 0.2592,
"step": 35680
},
{
"epoch": 4.161030783582089,
"grad_norm": 0.5541889766125377,
"learning_rate": 8.37586621546389e-06,
"loss": 0.2552,
"step": 35685
},
{
"epoch": 4.161613805970149,
"grad_norm": 0.5238246783833841,
"learning_rate": 8.371296684813806e-06,
"loss": 0.2522,
"step": 35690
},
{
"epoch": 4.162196828358209,
"grad_norm": 0.5621821440614606,
"learning_rate": 8.36672999841863e-06,
"loss": 0.2691,
"step": 35695
},
{
"epoch": 4.1627798507462686,
"grad_norm": 0.579634596573173,
"learning_rate": 8.36216615695738e-06,
"loss": 0.2676,
"step": 35700
},
{
"epoch": 4.163362873134329,
"grad_norm": 0.528976953956014,
"learning_rate": 8.357605161108663e-06,
"loss": 0.2661,
"step": 35705
},
{
"epoch": 4.163945895522388,
"grad_norm": 0.49251543743971965,
"learning_rate": 8.353047011550654e-06,
"loss": 0.2589,
"step": 35710
},
{
"epoch": 4.164528917910448,
"grad_norm": 0.5098277825232737,
"learning_rate": 8.348491708961102e-06,
"loss": 0.2689,
"step": 35715
},
{
"epoch": 4.165111940298507,
"grad_norm": 0.5330813333767002,
"learning_rate": 8.343939254017336e-06,
"loss": 0.2522,
"step": 35720
},
{
"epoch": 4.165694962686567,
"grad_norm": 0.5957115286883581,
"learning_rate": 8.339389647396265e-06,
"loss": 0.255,
"step": 35725
},
{
"epoch": 4.166277985074627,
"grad_norm": 0.5451929472592748,
"learning_rate": 8.334842889774374e-06,
"loss": 0.2574,
"step": 35730
},
{
"epoch": 4.166861007462686,
"grad_norm": 0.5571371475507364,
"learning_rate": 8.330298981827719e-06,
"loss": 0.2503,
"step": 35735
},
{
"epoch": 4.1674440298507465,
"grad_norm": 0.5588413498691291,
"learning_rate": 8.325757924231938e-06,
"loss": 0.2682,
"step": 35740
},
{
"epoch": 4.168027052238806,
"grad_norm": 0.4987987772903924,
"learning_rate": 8.321219717662249e-06,
"loss": 0.2478,
"step": 35745
},
{
"epoch": 4.168610074626866,
"grad_norm": 0.49153653353859766,
"learning_rate": 8.31668436279342e-06,
"loss": 0.2608,
"step": 35750
},
{
"epoch": 4.169193097014926,
"grad_norm": 0.4910461507026503,
"learning_rate": 8.312151860299835e-06,
"loss": 0.2365,
"step": 35755
},
{
"epoch": 4.169776119402985,
"grad_norm": 0.5808500192652937,
"learning_rate": 8.307622210855425e-06,
"loss": 0.2564,
"step": 35760
},
{
"epoch": 4.170359141791045,
"grad_norm": 0.5116169196975757,
"learning_rate": 8.303095415133703e-06,
"loss": 0.2638,
"step": 35765
},
{
"epoch": 4.170942164179104,
"grad_norm": 0.5097325625916911,
"learning_rate": 8.298571473807767e-06,
"loss": 0.266,
"step": 35770
},
{
"epoch": 4.171525186567164,
"grad_norm": 0.5642663632160959,
"learning_rate": 8.29405038755028e-06,
"loss": 0.2666,
"step": 35775
},
{
"epoch": 4.1721082089552235,
"grad_norm": 0.5147817847658372,
"learning_rate": 8.289532157033481e-06,
"loss": 0.2669,
"step": 35780
},
{
"epoch": 4.172691231343284,
"grad_norm": 0.5371217907838081,
"learning_rate": 8.2850167829292e-06,
"loss": 0.2564,
"step": 35785
},
{
"epoch": 4.173274253731344,
"grad_norm": 0.5246730258831356,
"learning_rate": 8.28050426590881e-06,
"loss": 0.2457,
"step": 35790
},
{
"epoch": 4.173857276119403,
"grad_norm": 0.494737036273425,
"learning_rate": 8.2759946066433e-06,
"loss": 0.2542,
"step": 35795
},
{
"epoch": 4.174440298507463,
"grad_norm": 0.5091178946485959,
"learning_rate": 8.271487805803193e-06,
"loss": 0.2493,
"step": 35800
},
{
"epoch": 4.175023320895522,
"grad_norm": 0.5218100567720455,
"learning_rate": 8.26698386405863e-06,
"loss": 0.2606,
"step": 35805
},
{
"epoch": 4.175606343283582,
"grad_norm": 0.5331823430293728,
"learning_rate": 8.262482782079281e-06,
"loss": 0.2493,
"step": 35810
},
{
"epoch": 4.176189365671641,
"grad_norm": 0.5205613862996497,
"learning_rate": 8.25798456053443e-06,
"loss": 0.2517,
"step": 35815
},
{
"epoch": 4.176772388059701,
"grad_norm": 0.5626865062257371,
"learning_rate": 8.253489200092912e-06,
"loss": 0.2456,
"step": 35820
},
{
"epoch": 4.1773554104477615,
"grad_norm": 0.592312622009633,
"learning_rate": 8.248996701423141e-06,
"loss": 0.2604,
"step": 35825
},
{
"epoch": 4.177938432835821,
"grad_norm": 0.4899871092553764,
"learning_rate": 8.244507065193117e-06,
"loss": 0.253,
"step": 35830
},
{
"epoch": 4.178521455223881,
"grad_norm": 0.4930121388676626,
"learning_rate": 8.240020292070408e-06,
"loss": 0.2609,
"step": 35835
},
{
"epoch": 4.17910447761194,
"grad_norm": 0.5052795083406623,
"learning_rate": 8.235536382722133e-06,
"loss": 0.2417,
"step": 35840
},
{
"epoch": 4.1796875,
"grad_norm": 0.49997651985199454,
"learning_rate": 8.231055337815039e-06,
"loss": 0.2529,
"step": 35845
},
{
"epoch": 4.18027052238806,
"grad_norm": 0.5588037312255781,
"learning_rate": 8.226577158015383e-06,
"loss": 0.2588,
"step": 35850
},
{
"epoch": 4.180853544776119,
"grad_norm": 0.5031696482630953,
"learning_rate": 8.22210184398905e-06,
"loss": 0.2784,
"step": 35855
},
{
"epoch": 4.181436567164179,
"grad_norm": 0.5097831655399934,
"learning_rate": 8.217629396401465e-06,
"loss": 0.2656,
"step": 35860
},
{
"epoch": 4.1820195895522385,
"grad_norm": 0.5219448757506173,
"learning_rate": 8.21315981591764e-06,
"loss": 0.2604,
"step": 35865
},
{
"epoch": 4.182602611940299,
"grad_norm": 0.5800918170815204,
"learning_rate": 8.208693103202158e-06,
"loss": 0.2532,
"step": 35870
},
{
"epoch": 4.183185634328359,
"grad_norm": 0.5219678785499666,
"learning_rate": 8.20422925891918e-06,
"loss": 0.2435,
"step": 35875
},
{
"epoch": 4.183768656716418,
"grad_norm": 0.5939386831359745,
"learning_rate": 8.199768283732432e-06,
"loss": 0.2668,
"step": 35880
},
{
"epoch": 4.184351679104478,
"grad_norm": 0.4817233331041549,
"learning_rate": 8.19531017830523e-06,
"loss": 0.2689,
"step": 35885
},
{
"epoch": 4.184934701492537,
"grad_norm": 0.5186357667568724,
"learning_rate": 8.190854943300436e-06,
"loss": 0.2584,
"step": 35890
},
{
"epoch": 4.185517723880597,
"grad_norm": 0.5439076379046998,
"learning_rate": 8.18640257938051e-06,
"loss": 0.2544,
"step": 35895
},
{
"epoch": 4.186100746268656,
"grad_norm": 0.49688221461825877,
"learning_rate": 8.181953087207467e-06,
"loss": 0.2483,
"step": 35900
},
{
"epoch": 4.186683768656716,
"grad_norm": 0.49769815620246266,
"learning_rate": 8.177506467442915e-06,
"loss": 0.2428,
"step": 35905
},
{
"epoch": 4.1872667910447765,
"grad_norm": 0.5374612550401273,
"learning_rate": 8.17306272074802e-06,
"loss": 0.2597,
"step": 35910
},
{
"epoch": 4.187849813432836,
"grad_norm": 0.48935843101678395,
"learning_rate": 8.16862184778352e-06,
"loss": 0.2486,
"step": 35915
},
{
"epoch": 4.188432835820896,
"grad_norm": 0.5140564850219497,
"learning_rate": 8.164183849209741e-06,
"loss": 0.2554,
"step": 35920
},
{
"epoch": 4.189015858208955,
"grad_norm": 0.5513480145355107,
"learning_rate": 8.159748725686554e-06,
"loss": 0.2741,
"step": 35925
},
{
"epoch": 4.189598880597015,
"grad_norm": 0.5193608240444869,
"learning_rate": 8.155316477873438e-06,
"loss": 0.2587,
"step": 35930
},
{
"epoch": 4.190181902985074,
"grad_norm": 0.5942447434682527,
"learning_rate": 8.150887106429412e-06,
"loss": 0.2578,
"step": 35935
},
{
"epoch": 4.190764925373134,
"grad_norm": 0.501049514911053,
"learning_rate": 8.146460612013083e-06,
"loss": 0.2667,
"step": 35940
},
{
"epoch": 4.191347947761194,
"grad_norm": 0.4857907189274466,
"learning_rate": 8.142036995282633e-06,
"loss": 0.2488,
"step": 35945
},
{
"epoch": 4.1919309701492535,
"grad_norm": 0.6670315784476127,
"learning_rate": 8.137616256895811e-06,
"loss": 0.272,
"step": 35950
},
{
"epoch": 4.192513992537314,
"grad_norm": 0.5773504915296385,
"learning_rate": 8.133198397509936e-06,
"loss": 0.2598,
"step": 35955
},
{
"epoch": 4.193097014925373,
"grad_norm": 0.5206693955814928,
"learning_rate": 8.128783417781909e-06,
"loss": 0.2525,
"step": 35960
},
{
"epoch": 4.193680037313433,
"grad_norm": 0.533228174492143,
"learning_rate": 8.124371318368176e-06,
"loss": 0.2506,
"step": 35965
},
{
"epoch": 4.194263059701493,
"grad_norm": 0.5342964299565146,
"learning_rate": 8.119962099924797e-06,
"loss": 0.2609,
"step": 35970
},
{
"epoch": 4.194846082089552,
"grad_norm": 0.5392135108758016,
"learning_rate": 8.115555763107362e-06,
"loss": 0.2616,
"step": 35975
},
{
"epoch": 4.195429104477612,
"grad_norm": 0.5595098596245798,
"learning_rate": 8.111152308571065e-06,
"loss": 0.2564,
"step": 35980
},
{
"epoch": 4.196012126865671,
"grad_norm": 0.5760912824826399,
"learning_rate": 8.10675173697065e-06,
"loss": 0.2704,
"step": 35985
},
{
"epoch": 4.1965951492537314,
"grad_norm": 0.4998090745247113,
"learning_rate": 8.10235404896044e-06,
"loss": 0.2553,
"step": 35990
},
{
"epoch": 4.197178171641791,
"grad_norm": 0.5243605883965704,
"learning_rate": 8.097959245194333e-06,
"loss": 0.2643,
"step": 35995
},
{
"epoch": 4.197761194029851,
"grad_norm": 0.5270577400307964,
"learning_rate": 8.09356732632579e-06,
"loss": 0.2559,
"step": 36000
},
{
"epoch": 4.198344216417911,
"grad_norm": 0.49781530633714827,
"learning_rate": 8.089178293007848e-06,
"loss": 0.2441,
"step": 36005
},
{
"epoch": 4.19892723880597,
"grad_norm": 0.4833727758202483,
"learning_rate": 8.084792145893122e-06,
"loss": 0.2437,
"step": 36010
},
{
"epoch": 4.19951026119403,
"grad_norm": 0.48774601752300895,
"learning_rate": 8.08040888563377e-06,
"loss": 0.2614,
"step": 36015
},
{
"epoch": 4.200093283582089,
"grad_norm": 0.527381214528893,
"learning_rate": 8.07602851288157e-06,
"loss": 0.2617,
"step": 36020
},
{
"epoch": 4.200676305970149,
"grad_norm": 0.5536445360071436,
"learning_rate": 8.07165102828781e-06,
"loss": 0.2617,
"step": 36025
},
{
"epoch": 4.201259328358209,
"grad_norm": 0.5053238389327633,
"learning_rate": 8.067276432503406e-06,
"loss": 0.2589,
"step": 36030
},
{
"epoch": 4.2018423507462686,
"grad_norm": 0.4972513306891196,
"learning_rate": 8.062904726178806e-06,
"loss": 0.2523,
"step": 36035
},
{
"epoch": 4.202425373134329,
"grad_norm": 0.5571497706103212,
"learning_rate": 8.058535909964041e-06,
"loss": 0.261,
"step": 36040
},
{
"epoch": 4.203008395522388,
"grad_norm": 0.5216726436146869,
"learning_rate": 8.054169984508714e-06,
"loss": 0.2473,
"step": 36045
},
{
"epoch": 4.203591417910448,
"grad_norm": 0.5058800923696642,
"learning_rate": 8.049806950461996e-06,
"loss": 0.2727,
"step": 36050
},
{
"epoch": 4.204174440298507,
"grad_norm": 0.509577924902328,
"learning_rate": 8.045446808472628e-06,
"loss": 0.2558,
"step": 36055
},
{
"epoch": 4.204757462686567,
"grad_norm": 0.5448385570748286,
"learning_rate": 8.041089559188929e-06,
"loss": 0.2437,
"step": 36060
},
{
"epoch": 4.205340485074627,
"grad_norm": 0.5009247022073703,
"learning_rate": 8.036735203258766e-06,
"loss": 0.2547,
"step": 36065
},
{
"epoch": 4.205923507462686,
"grad_norm": 0.5114127604823394,
"learning_rate": 8.032383741329598e-06,
"loss": 0.2564,
"step": 36070
},
{
"epoch": 4.2065065298507465,
"grad_norm": 0.5215847196762179,
"learning_rate": 8.028035174048446e-06,
"loss": 0.2452,
"step": 36075
},
{
"epoch": 4.207089552238806,
"grad_norm": 0.5376417982056957,
"learning_rate": 8.023689502061897e-06,
"loss": 0.2431,
"step": 36080
},
{
"epoch": 4.207672574626866,
"grad_norm": 0.5197812530448618,
"learning_rate": 8.019346726016116e-06,
"loss": 0.2471,
"step": 36085
},
{
"epoch": 4.208255597014926,
"grad_norm": 0.5202704728429384,
"learning_rate": 8.015006846556825e-06,
"loss": 0.2503,
"step": 36090
},
{
"epoch": 4.208838619402985,
"grad_norm": 0.49235619354545446,
"learning_rate": 8.010669864329334e-06,
"loss": 0.268,
"step": 36095
},
{
"epoch": 4.209421641791045,
"grad_norm": 0.5410523126980088,
"learning_rate": 8.006335779978494e-06,
"loss": 0.2569,
"step": 36100
},
{
"epoch": 4.210004664179104,
"grad_norm": 0.5192116504706309,
"learning_rate": 8.00200459414876e-06,
"loss": 0.2681,
"step": 36105
},
{
"epoch": 4.210587686567164,
"grad_norm": 0.4638425184517283,
"learning_rate": 7.997676307484123e-06,
"loss": 0.2438,
"step": 36110
},
{
"epoch": 4.2111707089552235,
"grad_norm": 0.5130433508193969,
"learning_rate": 7.993350920628164e-06,
"loss": 0.2555,
"step": 36115
},
{
"epoch": 4.211753731343284,
"grad_norm": 0.5825557038472461,
"learning_rate": 7.989028434224028e-06,
"loss": 0.2724,
"step": 36120
},
{
"epoch": 4.212336753731344,
"grad_norm": 0.5518343328605017,
"learning_rate": 7.984708848914426e-06,
"loss": 0.2479,
"step": 36125
},
{
"epoch": 4.212919776119403,
"grad_norm": 0.5050534075937266,
"learning_rate": 7.980392165341636e-06,
"loss": 0.2692,
"step": 36130
},
{
"epoch": 4.213502798507463,
"grad_norm": 0.5382208455206013,
"learning_rate": 7.976078384147515e-06,
"loss": 0.2651,
"step": 36135
},
{
"epoch": 4.214085820895522,
"grad_norm": 0.5552846334550404,
"learning_rate": 7.971767505973468e-06,
"loss": 0.2654,
"step": 36140
},
{
"epoch": 4.214668843283582,
"grad_norm": 0.5018471496633139,
"learning_rate": 7.9674595314605e-06,
"loss": 0.2588,
"step": 36145
},
{
"epoch": 4.215251865671641,
"grad_norm": 0.5182138992083678,
"learning_rate": 7.963154461249143e-06,
"loss": 0.2656,
"step": 36150
},
{
"epoch": 4.215834888059701,
"grad_norm": 0.48346843633341396,
"learning_rate": 7.958852295979542e-06,
"loss": 0.2574,
"step": 36155
},
{
"epoch": 4.2164179104477615,
"grad_norm": 0.46271003948790973,
"learning_rate": 7.95455303629137e-06,
"loss": 0.2633,
"step": 36160
},
{
"epoch": 4.217000932835821,
"grad_norm": 0.5246992653702127,
"learning_rate": 7.950256682823895e-06,
"loss": 0.2587,
"step": 36165
},
{
"epoch": 4.217583955223881,
"grad_norm": 0.46752826909368617,
"learning_rate": 7.945963236215944e-06,
"loss": 0.2526,
"step": 36170
},
{
"epoch": 4.21816697761194,
"grad_norm": 0.49985760934816387,
"learning_rate": 7.941672697105905e-06,
"loss": 0.2756,
"step": 36175
},
{
"epoch": 4.21875,
"grad_norm": 0.5480026483849813,
"learning_rate": 7.937385066131745e-06,
"loss": 0.2779,
"step": 36180
},
{
"epoch": 4.21933302238806,
"grad_norm": 0.5205461784668023,
"learning_rate": 7.933100343930995e-06,
"loss": 0.2547,
"step": 36185
},
{
"epoch": 4.219916044776119,
"grad_norm": 0.5359242257528088,
"learning_rate": 7.928818531140748e-06,
"loss": 0.2614,
"step": 36190
},
{
"epoch": 4.220499067164179,
"grad_norm": 0.5172170281092153,
"learning_rate": 7.924539628397675e-06,
"loss": 0.2572,
"step": 36195
},
{
"epoch": 4.2210820895522385,
"grad_norm": 0.48467656439608553,
"learning_rate": 7.920263636337994e-06,
"loss": 0.2537,
"step": 36200
},
{
"epoch": 4.221665111940299,
"grad_norm": 0.4963780572434781,
"learning_rate": 7.915990555597522e-06,
"loss": 0.2616,
"step": 36205
},
{
"epoch": 4.222248134328359,
"grad_norm": 0.5338221364458754,
"learning_rate": 7.911720386811613e-06,
"loss": 0.2627,
"step": 36210
},
{
"epoch": 4.222831156716418,
"grad_norm": 0.49942605853754896,
"learning_rate": 7.907453130615203e-06,
"loss": 0.2568,
"step": 36215
},
{
"epoch": 4.223414179104478,
"grad_norm": 0.5328164426898738,
"learning_rate": 7.90318878764279e-06,
"loss": 0.2603,
"step": 36220
},
{
"epoch": 4.223997201492537,
"grad_norm": 0.503178550188616,
"learning_rate": 7.898927358528447e-06,
"loss": 0.2568,
"step": 36225
},
{
"epoch": 4.224580223880597,
"grad_norm": 0.5251883305365175,
"learning_rate": 7.894668843905803e-06,
"loss": 0.2684,
"step": 36230
},
{
"epoch": 4.225163246268656,
"grad_norm": 0.5260399225353416,
"learning_rate": 7.890413244408059e-06,
"loss": 0.2634,
"step": 36235
},
{
"epoch": 4.225746268656716,
"grad_norm": 0.5101823924393167,
"learning_rate": 7.886160560667984e-06,
"loss": 0.2741,
"step": 36240
},
{
"epoch": 4.2263292910447765,
"grad_norm": 0.5553837194000492,
"learning_rate": 7.881910793317915e-06,
"loss": 0.2478,
"step": 36245
},
{
"epoch": 4.226912313432836,
"grad_norm": 0.5605173196582268,
"learning_rate": 7.87766394298974e-06,
"loss": 0.2704,
"step": 36250
},
{
"epoch": 4.227495335820896,
"grad_norm": 0.5432033405633592,
"learning_rate": 7.873420010314933e-06,
"loss": 0.2542,
"step": 36255
},
{
"epoch": 4.228078358208955,
"grad_norm": 0.48881589685516663,
"learning_rate": 7.869178995924525e-06,
"loss": 0.2591,
"step": 36260
},
{
"epoch": 4.228661380597015,
"grad_norm": 0.5909026498093377,
"learning_rate": 7.864940900449109e-06,
"loss": 0.2609,
"step": 36265
},
{
"epoch": 4.229244402985074,
"grad_norm": 0.5149006012989761,
"learning_rate": 7.860705724518857e-06,
"loss": 0.2587,
"step": 36270
},
{
"epoch": 4.229827425373134,
"grad_norm": 0.5347576530594315,
"learning_rate": 7.8564734687635e-06,
"loss": 0.2529,
"step": 36275
},
{
"epoch": 4.230410447761194,
"grad_norm": 0.5373956815286194,
"learning_rate": 7.852244133812332e-06,
"loss": 0.2681,
"step": 36280
},
{
"epoch": 4.2309934701492535,
"grad_norm": 0.5963324877668421,
"learning_rate": 7.8480177202942e-06,
"loss": 0.279,
"step": 36285
},
{
"epoch": 4.231576492537314,
"grad_norm": 0.5333019181317546,
"learning_rate": 7.843794228837556e-06,
"loss": 0.267,
"step": 36290
},
{
"epoch": 4.232159514925373,
"grad_norm": 0.5076514033378163,
"learning_rate": 7.839573660070373e-06,
"loss": 0.242,
"step": 36295
},
{
"epoch": 4.232742537313433,
"grad_norm": 0.5521856639246142,
"learning_rate": 7.83535601462022e-06,
"loss": 0.2633,
"step": 36300
},
{
"epoch": 4.233325559701493,
"grad_norm": 0.518017595350679,
"learning_rate": 7.831141293114216e-06,
"loss": 0.267,
"step": 36305
},
{
"epoch": 4.233908582089552,
"grad_norm": 0.5082093658424798,
"learning_rate": 7.82692949617905e-06,
"loss": 0.2749,
"step": 36310
},
{
"epoch": 4.234491604477612,
"grad_norm": 0.5159511799153876,
"learning_rate": 7.822720624440978e-06,
"loss": 0.2662,
"step": 36315
},
{
"epoch": 4.235074626865671,
"grad_norm": 0.562555628318104,
"learning_rate": 7.818514678525822e-06,
"loss": 0.2481,
"step": 36320
},
{
"epoch": 4.2356576492537314,
"grad_norm": 0.5499192977993825,
"learning_rate": 7.814311659058951e-06,
"loss": 0.2635,
"step": 36325
},
{
"epoch": 4.236240671641791,
"grad_norm": 0.5075712256569114,
"learning_rate": 7.810111566665333e-06,
"loss": 0.2609,
"step": 36330
},
{
"epoch": 4.236823694029851,
"grad_norm": 0.538459860137172,
"learning_rate": 7.805914401969466e-06,
"loss": 0.2648,
"step": 36335
},
{
"epoch": 4.237406716417911,
"grad_norm": 0.5356031378837667,
"learning_rate": 7.80172016559544e-06,
"loss": 0.2527,
"step": 36340
},
{
"epoch": 4.23798973880597,
"grad_norm": 0.4841147045807619,
"learning_rate": 7.797528858166891e-06,
"loss": 0.2402,
"step": 36345
},
{
"epoch": 4.23857276119403,
"grad_norm": 0.5339756626276096,
"learning_rate": 7.793340480307027e-06,
"loss": 0.2574,
"step": 36350
},
{
"epoch": 4.239155783582089,
"grad_norm": 0.5409122599507549,
"learning_rate": 7.789155032638619e-06,
"loss": 0.2519,
"step": 36355
},
{
"epoch": 4.239738805970149,
"grad_norm": 0.527606702037047,
"learning_rate": 7.784972515784004e-06,
"loss": 0.2667,
"step": 36360
},
{
"epoch": 4.240321828358209,
"grad_norm": 0.5457238162669897,
"learning_rate": 7.780792930365085e-06,
"loss": 0.2601,
"step": 36365
},
{
"epoch": 4.2409048507462686,
"grad_norm": 0.5322066682409444,
"learning_rate": 7.776616277003328e-06,
"loss": 0.2694,
"step": 36370
},
{
"epoch": 4.241487873134329,
"grad_norm": 0.5124879069234782,
"learning_rate": 7.772442556319747e-06,
"loss": 0.2366,
"step": 36375
},
{
"epoch": 4.242070895522388,
"grad_norm": 0.5589255865140835,
"learning_rate": 7.768271768934955e-06,
"loss": 0.2778,
"step": 36380
},
{
"epoch": 4.242653917910448,
"grad_norm": 0.5092111567258133,
"learning_rate": 7.76410391546909e-06,
"loss": 0.2603,
"step": 36385
},
{
"epoch": 4.243236940298507,
"grad_norm": 0.5717120076961497,
"learning_rate": 7.759938996541886e-06,
"loss": 0.2761,
"step": 36390
},
{
"epoch": 4.243819962686567,
"grad_norm": 0.4567926897637574,
"learning_rate": 7.755777012772615e-06,
"loss": 0.2576,
"step": 36395
},
{
"epoch": 4.244402985074627,
"grad_norm": 0.49641405661123544,
"learning_rate": 7.751617964780131e-06,
"loss": 0.2568,
"step": 36400
},
{
"epoch": 4.244986007462686,
"grad_norm": 0.5271208730438519,
"learning_rate": 7.747461853182842e-06,
"loss": 0.2649,
"step": 36405
},
{
"epoch": 4.2455690298507465,
"grad_norm": 0.5332217545849521,
"learning_rate": 7.743308678598722e-06,
"loss": 0.2712,
"step": 36410
},
{
"epoch": 4.246152052238806,
"grad_norm": 0.503291882174182,
"learning_rate": 7.73915844164531e-06,
"loss": 0.2592,
"step": 36415
},
{
"epoch": 4.246735074626866,
"grad_norm": 0.5499570386458553,
"learning_rate": 7.73501114293971e-06,
"loss": 0.2626,
"step": 36420
},
{
"epoch": 4.247318097014926,
"grad_norm": 0.5075973972142243,
"learning_rate": 7.730866783098576e-06,
"loss": 0.25,
"step": 36425
},
{
"epoch": 4.247901119402985,
"grad_norm": 0.5677800484300837,
"learning_rate": 7.726725362738141e-06,
"loss": 0.2873,
"step": 36430
},
{
"epoch": 4.248484141791045,
"grad_norm": 0.5039882471901479,
"learning_rate": 7.722586882474191e-06,
"loss": 0.2634,
"step": 36435
},
{
"epoch": 4.249067164179104,
"grad_norm": 0.4832140453251526,
"learning_rate": 7.71845134292208e-06,
"loss": 0.2545,
"step": 36440
},
{
"epoch": 4.249650186567164,
"grad_norm": 0.49668851944428616,
"learning_rate": 7.714318744696728e-06,
"loss": 0.2624,
"step": 36445
},
{
"epoch": 4.2502332089552235,
"grad_norm": 0.5196422995791229,
"learning_rate": 7.710189088412604e-06,
"loss": 0.2636,
"step": 36450
},
{
"epoch": 4.250816231343284,
"grad_norm": 0.5591389791142553,
"learning_rate": 7.706062374683757e-06,
"loss": 0.262,
"step": 36455
},
{
"epoch": 4.251399253731344,
"grad_norm": 0.5311377058182638,
"learning_rate": 7.70193860412378e-06,
"loss": 0.2682,
"step": 36460
},
{
"epoch": 4.251982276119403,
"grad_norm": 0.5239124817838773,
"learning_rate": 7.697817777345852e-06,
"loss": 0.2564,
"step": 36465
},
{
"epoch": 4.252565298507463,
"grad_norm": 0.5315985409727104,
"learning_rate": 7.693699894962686e-06,
"loss": 0.2633,
"step": 36470
},
{
"epoch": 4.253148320895522,
"grad_norm": 0.5551970215290292,
"learning_rate": 7.689584957586578e-06,
"loss": 0.2682,
"step": 36475
},
{
"epoch": 4.253731343283582,
"grad_norm": 0.5200032944295727,
"learning_rate": 7.68547296582938e-06,
"loss": 0.2473,
"step": 36480
},
{
"epoch": 4.254314365671641,
"grad_norm": 0.5082693332036062,
"learning_rate": 7.681363920302506e-06,
"loss": 0.2606,
"step": 36485
},
{
"epoch": 4.254897388059701,
"grad_norm": 0.49708545300777185,
"learning_rate": 7.67725782161693e-06,
"loss": 0.2549,
"step": 36490
},
{
"epoch": 4.2554804104477615,
"grad_norm": 0.5354157310233019,
"learning_rate": 7.673154670383195e-06,
"loss": 0.2594,
"step": 36495
},
{
"epoch": 4.256063432835821,
"grad_norm": 0.5855095665349248,
"learning_rate": 7.669054467211388e-06,
"loss": 0.25,
"step": 36500
},
{
"epoch": 4.256646455223881,
"grad_norm": 0.5123276101728429,
"learning_rate": 7.664957212711187e-06,
"loss": 0.2676,
"step": 36505
},
{
"epoch": 4.25722947761194,
"grad_norm": 0.5419371953353457,
"learning_rate": 7.660862907491795e-06,
"loss": 0.2583,
"step": 36510
},
{
"epoch": 4.2578125,
"grad_norm": 0.5071122813583242,
"learning_rate": 7.656771552162015e-06,
"loss": 0.2549,
"step": 36515
},
{
"epoch": 4.25839552238806,
"grad_norm": 0.5314706075957834,
"learning_rate": 7.652683147330177e-06,
"loss": 0.2612,
"step": 36520
},
{
"epoch": 4.258978544776119,
"grad_norm": 0.5254949390459906,
"learning_rate": 7.6485976936042e-06,
"loss": 0.2753,
"step": 36525
},
{
"epoch": 4.259561567164179,
"grad_norm": 0.518860620361999,
"learning_rate": 7.644515191591542e-06,
"loss": 0.2573,
"step": 36530
},
{
"epoch": 4.2601445895522385,
"grad_norm": 0.5139194775905636,
"learning_rate": 7.640435641899236e-06,
"loss": 0.2733,
"step": 36535
},
{
"epoch": 4.260727611940299,
"grad_norm": 0.4904973044462139,
"learning_rate": 7.636359045133873e-06,
"loss": 0.2436,
"step": 36540
},
{
"epoch": 4.261310634328359,
"grad_norm": 0.4845043649626959,
"learning_rate": 7.632285401901606e-06,
"loss": 0.2521,
"step": 36545
},
{
"epoch": 4.261893656716418,
"grad_norm": 0.5427050745701884,
"learning_rate": 7.6282147128081364e-06,
"loss": 0.2578,
"step": 36550
},
{
"epoch": 4.262476679104478,
"grad_norm": 0.4804396739028346,
"learning_rate": 7.624146978458754e-06,
"loss": 0.2494,
"step": 36555
},
{
"epoch": 4.263059701492537,
"grad_norm": 0.524789611447541,
"learning_rate": 7.620082199458269e-06,
"loss": 0.2719,
"step": 36560
},
{
"epoch": 4.263642723880597,
"grad_norm": 0.5390021497011211,
"learning_rate": 7.616020376411098e-06,
"loss": 0.2544,
"step": 36565
},
{
"epoch": 4.264225746268656,
"grad_norm": 0.5455055751894204,
"learning_rate": 7.611961509921182e-06,
"loss": 0.265,
"step": 36570
},
{
"epoch": 4.264808768656716,
"grad_norm": 0.5206793504486821,
"learning_rate": 7.6079056005920375e-06,
"loss": 0.2488,
"step": 36575
},
{
"epoch": 4.2653917910447765,
"grad_norm": 0.5502080598660063,
"learning_rate": 7.603852649026738e-06,
"loss": 0.2652,
"step": 36580
},
{
"epoch": 4.265974813432836,
"grad_norm": 0.5539548194259345,
"learning_rate": 7.599802655827924e-06,
"loss": 0.2557,
"step": 36585
},
{
"epoch": 4.266557835820896,
"grad_norm": 0.4687970332586143,
"learning_rate": 7.595755621597788e-06,
"loss": 0.2534,
"step": 36590
},
{
"epoch": 4.267140858208955,
"grad_norm": 0.676537775642717,
"learning_rate": 7.591711546938086e-06,
"loss": 0.2492,
"step": 36595
},
{
"epoch": 4.267723880597015,
"grad_norm": 0.5022555909708639,
"learning_rate": 7.587670432450131e-06,
"loss": 0.2655,
"step": 36600
},
{
"epoch": 4.268306902985074,
"grad_norm": 0.5339158540274818,
"learning_rate": 7.583632278734798e-06,
"loss": 0.2622,
"step": 36605
},
{
"epoch": 4.268889925373134,
"grad_norm": 0.5299828743541835,
"learning_rate": 7.57959708639252e-06,
"loss": 0.2605,
"step": 36610
},
{
"epoch": 4.269472947761194,
"grad_norm": 0.5067527837897201,
"learning_rate": 7.575564856023298e-06,
"loss": 0.2622,
"step": 36615
},
{
"epoch": 4.2700559701492535,
"grad_norm": 0.5871851959846127,
"learning_rate": 7.5715355882266815e-06,
"loss": 0.2635,
"step": 36620
},
{
"epoch": 4.270638992537314,
"grad_norm": 0.5452975219675954,
"learning_rate": 7.567509283601784e-06,
"loss": 0.2634,
"step": 36625
},
{
"epoch": 4.271222014925373,
"grad_norm": 0.5425556897223894,
"learning_rate": 7.5634859427472835e-06,
"loss": 0.2463,
"step": 36630
},
{
"epoch": 4.271805037313433,
"grad_norm": 0.5659494273780724,
"learning_rate": 7.5594655662613995e-06,
"loss": 0.2524,
"step": 36635
},
{
"epoch": 4.272388059701493,
"grad_norm": 0.5783645379273165,
"learning_rate": 7.5554481547419395e-06,
"loss": 0.2734,
"step": 36640
},
{
"epoch": 4.272971082089552,
"grad_norm": 0.5542099066930374,
"learning_rate": 7.551433708786243e-06,
"loss": 0.2619,
"step": 36645
},
{
"epoch": 4.273554104477612,
"grad_norm": 0.5261202362681132,
"learning_rate": 7.547422228991223e-06,
"loss": 0.2733,
"step": 36650
},
{
"epoch": 4.274137126865671,
"grad_norm": 0.5093685875923936,
"learning_rate": 7.543413715953347e-06,
"loss": 0.2824,
"step": 36655
},
{
"epoch": 4.2747201492537314,
"grad_norm": 0.5460849079015364,
"learning_rate": 7.539408170268644e-06,
"loss": 0.2553,
"step": 36660
},
{
"epoch": 4.275303171641791,
"grad_norm": 0.5560161818099602,
"learning_rate": 7.535405592532703e-06,
"loss": 0.2553,
"step": 36665
},
{
"epoch": 4.275886194029851,
"grad_norm": 0.4952917383487221,
"learning_rate": 7.531405983340668e-06,
"loss": 0.2519,
"step": 36670
},
{
"epoch": 4.276469216417911,
"grad_norm": 0.5111450651997773,
"learning_rate": 7.527409343287231e-06,
"loss": 0.2384,
"step": 36675
},
{
"epoch": 4.27705223880597,
"grad_norm": 0.4905017963777417,
"learning_rate": 7.523415672966675e-06,
"loss": 0.2559,
"step": 36680
},
{
"epoch": 4.27763526119403,
"grad_norm": 0.4983806910551259,
"learning_rate": 7.519424972972797e-06,
"loss": 0.2429,
"step": 36685
},
{
"epoch": 4.278218283582089,
"grad_norm": 0.5434853616544607,
"learning_rate": 7.515437243898998e-06,
"loss": 0.2663,
"step": 36690
},
{
"epoch": 4.278801305970149,
"grad_norm": 0.5324952085708943,
"learning_rate": 7.511452486338202e-06,
"loss": 0.2679,
"step": 36695
},
{
"epoch": 4.279384328358209,
"grad_norm": 0.5207583171148841,
"learning_rate": 7.507470700882905e-06,
"loss": 0.2495,
"step": 36700
},
{
"epoch": 4.2799673507462686,
"grad_norm": 0.5184628033405693,
"learning_rate": 7.503491888125165e-06,
"loss": 0.2523,
"step": 36705
},
{
"epoch": 4.280550373134329,
"grad_norm": 0.5457139838206316,
"learning_rate": 7.499516048656589e-06,
"loss": 0.2507,
"step": 36710
},
{
"epoch": 4.281133395522388,
"grad_norm": 0.5268301525162578,
"learning_rate": 7.495543183068349e-06,
"loss": 0.2538,
"step": 36715
},
{
"epoch": 4.281716417910448,
"grad_norm": 0.557522503250984,
"learning_rate": 7.491573291951176e-06,
"loss": 0.2472,
"step": 36720
},
{
"epoch": 4.282299440298507,
"grad_norm": 0.5443142569282987,
"learning_rate": 7.487606375895343e-06,
"loss": 0.2773,
"step": 36725
},
{
"epoch": 4.282882462686567,
"grad_norm": 0.4833465040881014,
"learning_rate": 7.483642435490706e-06,
"loss": 0.2717,
"step": 36730
},
{
"epoch": 4.283465485074627,
"grad_norm": 0.5174779099609201,
"learning_rate": 7.479681471326648e-06,
"loss": 0.2467,
"step": 36735
},
{
"epoch": 4.284048507462686,
"grad_norm": 0.5953972225493099,
"learning_rate": 7.475723483992149e-06,
"loss": 0.2769,
"step": 36740
},
{
"epoch": 4.2846315298507465,
"grad_norm": 0.48209904232999795,
"learning_rate": 7.471768474075706e-06,
"loss": 0.257,
"step": 36745
},
{
"epoch": 4.285214552238806,
"grad_norm": 0.5132628295947472,
"learning_rate": 7.467816442165397e-06,
"loss": 0.2642,
"step": 36750
},
{
"epoch": 4.285797574626866,
"grad_norm": 0.5161882336247029,
"learning_rate": 7.463867388848851e-06,
"loss": 0.2647,
"step": 36755
},
{
"epoch": 4.286380597014926,
"grad_norm": 0.5326378076947488,
"learning_rate": 7.459921314713253e-06,
"loss": 0.2708,
"step": 36760
},
{
"epoch": 4.286963619402985,
"grad_norm": 0.551228320162258,
"learning_rate": 7.4559782203453485e-06,
"loss": 0.2542,
"step": 36765
},
{
"epoch": 4.287546641791045,
"grad_norm": 0.5006053211843197,
"learning_rate": 7.452038106331442e-06,
"loss": 0.2403,
"step": 36770
},
{
"epoch": 4.288129664179104,
"grad_norm": 0.5245442263991795,
"learning_rate": 7.448100973257381e-06,
"loss": 0.2683,
"step": 36775
},
{
"epoch": 4.288712686567164,
"grad_norm": 0.5337926743044439,
"learning_rate": 7.444166821708584e-06,
"loss": 0.2503,
"step": 36780
},
{
"epoch": 4.2892957089552235,
"grad_norm": 0.4863058576843455,
"learning_rate": 7.440235652270024e-06,
"loss": 0.2552,
"step": 36785
},
{
"epoch": 4.289878731343284,
"grad_norm": 0.5028778268065867,
"learning_rate": 7.436307465526224e-06,
"loss": 0.2641,
"step": 36790
},
{
"epoch": 4.290461753731344,
"grad_norm": 0.507916076154588,
"learning_rate": 7.432382262061271e-06,
"loss": 0.2671,
"step": 36795
},
{
"epoch": 4.291044776119403,
"grad_norm": 0.5394543154021711,
"learning_rate": 7.4284600424588045e-06,
"loss": 0.2547,
"step": 36800
},
{
"epoch": 4.291627798507463,
"grad_norm": 0.4966968471220312,
"learning_rate": 7.424540807302019e-06,
"loss": 0.2512,
"step": 36805
},
{
"epoch": 4.292210820895522,
"grad_norm": 0.4970972940700219,
"learning_rate": 7.42062455717367e-06,
"loss": 0.2599,
"step": 36810
},
{
"epoch": 4.292793843283582,
"grad_norm": 0.5317555340317982,
"learning_rate": 7.4167112926560714e-06,
"loss": 0.257,
"step": 36815
},
{
"epoch": 4.293376865671641,
"grad_norm": 0.47957480433862504,
"learning_rate": 7.412801014331075e-06,
"loss": 0.2495,
"step": 36820
},
{
"epoch": 4.293959888059701,
"grad_norm": 0.5040812377491417,
"learning_rate": 7.408893722780108e-06,
"loss": 0.2669,
"step": 36825
},
{
"epoch": 4.2945429104477615,
"grad_norm": 0.5694451210267427,
"learning_rate": 7.4049894185841476e-06,
"loss": 0.2567,
"step": 36830
},
{
"epoch": 4.295125932835821,
"grad_norm": 0.5503887213795378,
"learning_rate": 7.401088102323729e-06,
"loss": 0.253,
"step": 36835
},
{
"epoch": 4.295708955223881,
"grad_norm": 0.5149263772610403,
"learning_rate": 7.397189774578939e-06,
"loss": 0.2532,
"step": 36840
},
{
"epoch": 4.29629197761194,
"grad_norm": 0.5036942272318263,
"learning_rate": 7.393294435929424e-06,
"loss": 0.2643,
"step": 36845
},
{
"epoch": 4.296875,
"grad_norm": 0.529239131044159,
"learning_rate": 7.389402086954368e-06,
"loss": 0.2518,
"step": 36850
},
{
"epoch": 4.29745802238806,
"grad_norm": 0.5311848561685814,
"learning_rate": 7.385512728232552e-06,
"loss": 0.2537,
"step": 36855
},
{
"epoch": 4.298041044776119,
"grad_norm": 0.5349998590566727,
"learning_rate": 7.38162636034226e-06,
"loss": 0.268,
"step": 36860
},
{
"epoch": 4.298624067164179,
"grad_norm": 0.5211252168730743,
"learning_rate": 7.37774298386138e-06,
"loss": 0.261,
"step": 36865
},
{
"epoch": 4.2992070895522385,
"grad_norm": 0.5270155424654448,
"learning_rate": 7.373862599367316e-06,
"loss": 0.2654,
"step": 36870
},
{
"epoch": 4.299790111940299,
"grad_norm": 0.5244230534494915,
"learning_rate": 7.3699852074370605e-06,
"loss": 0.262,
"step": 36875
},
{
"epoch": 4.300373134328359,
"grad_norm": 0.5377841525788232,
"learning_rate": 7.366110808647128e-06,
"loss": 0.2751,
"step": 36880
},
{
"epoch": 4.300956156716418,
"grad_norm": 0.4975736802719757,
"learning_rate": 7.362239403573614e-06,
"loss": 0.2518,
"step": 36885
},
{
"epoch": 4.301539179104478,
"grad_norm": 0.5036460743072355,
"learning_rate": 7.3583709927921574e-06,
"loss": 0.2587,
"step": 36890
},
{
"epoch": 4.302122201492537,
"grad_norm": 0.5484439567299549,
"learning_rate": 7.3545055768779526e-06,
"loss": 0.2598,
"step": 36895
},
{
"epoch": 4.302705223880597,
"grad_norm": 0.4981576744832116,
"learning_rate": 7.350643156405751e-06,
"loss": 0.2466,
"step": 36900
},
{
"epoch": 4.303288246268656,
"grad_norm": 0.4871133271522615,
"learning_rate": 7.346783731949864e-06,
"loss": 0.2612,
"step": 36905
},
{
"epoch": 4.303871268656716,
"grad_norm": 0.5570397093413274,
"learning_rate": 7.342927304084132e-06,
"loss": 0.2583,
"step": 36910
},
{
"epoch": 4.3044542910447765,
"grad_norm": 0.5341611951795412,
"learning_rate": 7.339073873381991e-06,
"loss": 0.2463,
"step": 36915
},
{
"epoch": 4.305037313432836,
"grad_norm": 0.5083869474138867,
"learning_rate": 7.335223440416391e-06,
"loss": 0.2592,
"step": 36920
},
{
"epoch": 4.305620335820896,
"grad_norm": 0.49069501591445763,
"learning_rate": 7.3313760057598715e-06,
"loss": 0.261,
"step": 36925
},
{
"epoch": 4.306203358208955,
"grad_norm": 0.5490578683808961,
"learning_rate": 7.327531569984497e-06,
"loss": 0.2597,
"step": 36930
},
{
"epoch": 4.306786380597015,
"grad_norm": 0.53560865433887,
"learning_rate": 7.3236901336619024e-06,
"loss": 0.2525,
"step": 36935
},
{
"epoch": 4.307369402985074,
"grad_norm": 0.5552324863820831,
"learning_rate": 7.319851697363271e-06,
"loss": 0.2585,
"step": 36940
},
{
"epoch": 4.307952425373134,
"grad_norm": 0.5362159037180242,
"learning_rate": 7.316016261659342e-06,
"loss": 0.2607,
"step": 36945
},
{
"epoch": 4.308535447761194,
"grad_norm": 0.47378728948205756,
"learning_rate": 7.31218382712041e-06,
"loss": 0.2435,
"step": 36950
},
{
"epoch": 4.3091184701492535,
"grad_norm": 0.541388164400969,
"learning_rate": 7.308354394316322e-06,
"loss": 0.2523,
"step": 36955
},
{
"epoch": 4.309701492537314,
"grad_norm": 0.5202445569550581,
"learning_rate": 7.304527963816472e-06,
"loss": 0.2588,
"step": 36960
},
{
"epoch": 4.310284514925373,
"grad_norm": 0.5339968171744611,
"learning_rate": 7.300704536189819e-06,
"loss": 0.2585,
"step": 36965
},
{
"epoch": 4.310867537313433,
"grad_norm": 0.5024294225573035,
"learning_rate": 7.2968841120048666e-06,
"loss": 0.2579,
"step": 36970
},
{
"epoch": 4.311450559701493,
"grad_norm": 0.5578889793339405,
"learning_rate": 7.293066691829676e-06,
"loss": 0.2499,
"step": 36975
},
{
"epoch": 4.312033582089552,
"grad_norm": 0.5927410826730686,
"learning_rate": 7.289252276231863e-06,
"loss": 0.2569,
"step": 36980
},
{
"epoch": 4.312616604477612,
"grad_norm": 0.5151664246773885,
"learning_rate": 7.285440865778594e-06,
"loss": 0.2432,
"step": 36985
},
{
"epoch": 4.313199626865671,
"grad_norm": 0.5322956216373098,
"learning_rate": 7.281632461036594e-06,
"loss": 0.2595,
"step": 36990
},
{
"epoch": 4.3137826492537314,
"grad_norm": 0.5074734962833999,
"learning_rate": 7.277827062572121e-06,
"loss": 0.263,
"step": 36995
},
{
"epoch": 4.314365671641791,
"grad_norm": 0.5277409660614346,
"learning_rate": 7.27402467095102e-06,
"loss": 0.2653,
"step": 37000
},
{
"epoch": 4.314948694029851,
"grad_norm": 0.548905404143263,
"learning_rate": 7.2702252867386575e-06,
"loss": 0.2684,
"step": 37005
},
{
"epoch": 4.315531716417911,
"grad_norm": 0.5514609071435579,
"learning_rate": 7.266428910499971e-06,
"loss": 0.2735,
"step": 37010
},
{
"epoch": 4.31611473880597,
"grad_norm": 0.5429758104363809,
"learning_rate": 7.262635542799444e-06,
"loss": 0.2652,
"step": 37015
},
{
"epoch": 4.31669776119403,
"grad_norm": 0.5303149317593546,
"learning_rate": 7.258845184201111e-06,
"loss": 0.264,
"step": 37020
},
{
"epoch": 4.317280783582089,
"grad_norm": 0.5060160025827667,
"learning_rate": 7.255057835268567e-06,
"loss": 0.2548,
"step": 37025
},
{
"epoch": 4.317863805970149,
"grad_norm": 0.5277354228805766,
"learning_rate": 7.251273496564957e-06,
"loss": 0.2623,
"step": 37030
},
{
"epoch": 4.318446828358209,
"grad_norm": 0.5563520398657388,
"learning_rate": 7.2474921686529625e-06,
"loss": 0.2716,
"step": 37035
},
{
"epoch": 4.3190298507462686,
"grad_norm": 0.5057708969419261,
"learning_rate": 7.243713852094848e-06,
"loss": 0.2426,
"step": 37040
},
{
"epoch": 4.319612873134329,
"grad_norm": 0.5564721538521776,
"learning_rate": 7.239938547452394e-06,
"loss": 0.2677,
"step": 37045
},
{
"epoch": 4.320195895522388,
"grad_norm": 0.5034968721136049,
"learning_rate": 7.2361662552869734e-06,
"loss": 0.2505,
"step": 37050
},
{
"epoch": 4.320778917910448,
"grad_norm": 0.5818160184561831,
"learning_rate": 7.232396976159475e-06,
"loss": 0.2579,
"step": 37055
},
{
"epoch": 4.321361940298507,
"grad_norm": 0.5601487613092481,
"learning_rate": 7.228630710630356e-06,
"loss": 0.2725,
"step": 37060
},
{
"epoch": 4.321944962686567,
"grad_norm": 0.4964651675982269,
"learning_rate": 7.224867459259628e-06,
"loss": 0.2591,
"step": 37065
},
{
"epoch": 4.322527985074627,
"grad_norm": 0.4994096964254231,
"learning_rate": 7.221107222606851e-06,
"loss": 0.2415,
"step": 37070
},
{
"epoch": 4.323111007462686,
"grad_norm": 0.503197828587518,
"learning_rate": 7.217350001231131e-06,
"loss": 0.2603,
"step": 37075
},
{
"epoch": 4.3236940298507465,
"grad_norm": 0.5240381374738424,
"learning_rate": 7.21359579569114e-06,
"loss": 0.2603,
"step": 37080
},
{
"epoch": 4.324277052238806,
"grad_norm": 0.5543165349117316,
"learning_rate": 7.2098446065450795e-06,
"loss": 0.2647,
"step": 37085
},
{
"epoch": 4.324860074626866,
"grad_norm": 0.5156631769536357,
"learning_rate": 7.206096434350728e-06,
"loss": 0.2573,
"step": 37090
},
{
"epoch": 4.325443097014926,
"grad_norm": 0.5425625059272369,
"learning_rate": 7.202351279665391e-06,
"loss": 0.2706,
"step": 37095
},
{
"epoch": 4.326026119402985,
"grad_norm": 0.5329765676788932,
"learning_rate": 7.198609143045948e-06,
"loss": 0.2468,
"step": 37100
},
{
"epoch": 4.326609141791045,
"grad_norm": 0.5254958625670477,
"learning_rate": 7.194870025048812e-06,
"loss": 0.2498,
"step": 37105
},
{
"epoch": 4.327192164179104,
"grad_norm": 0.50014405669136,
"learning_rate": 7.191133926229957e-06,
"loss": 0.2613,
"step": 37110
},
{
"epoch": 4.327775186567164,
"grad_norm": 0.5080791032131875,
"learning_rate": 7.187400847144904e-06,
"loss": 0.2636,
"step": 37115
},
{
"epoch": 4.3283582089552235,
"grad_norm": 0.4947112924898851,
"learning_rate": 7.183670788348726e-06,
"loss": 0.241,
"step": 37120
},
{
"epoch": 4.328941231343284,
"grad_norm": 0.5120615479347017,
"learning_rate": 7.1799437503960465e-06,
"loss": 0.2366,
"step": 37125
},
{
"epoch": 4.329524253731344,
"grad_norm": 0.495302330072865,
"learning_rate": 7.176219733841047e-06,
"loss": 0.2547,
"step": 37130
},
{
"epoch": 4.330107276119403,
"grad_norm": 0.5100176383568668,
"learning_rate": 7.17249873923744e-06,
"loss": 0.2394,
"step": 37135
},
{
"epoch": 4.330690298507463,
"grad_norm": 0.5075922533880023,
"learning_rate": 7.168780767138512e-06,
"loss": 0.2528,
"step": 37140
},
{
"epoch": 4.331273320895522,
"grad_norm": 0.5235847227785626,
"learning_rate": 7.165065818097086e-06,
"loss": 0.2542,
"step": 37145
},
{
"epoch": 4.331856343283582,
"grad_norm": 0.5491017386277993,
"learning_rate": 7.161353892665538e-06,
"loss": 0.2592,
"step": 37150
},
{
"epoch": 4.332439365671641,
"grad_norm": 0.5430027835984829,
"learning_rate": 7.157644991395801e-06,
"loss": 0.2729,
"step": 37155
},
{
"epoch": 4.333022388059701,
"grad_norm": 0.4914792423545939,
"learning_rate": 7.1539391148393474e-06,
"loss": 0.2511,
"step": 37160
},
{
"epoch": 4.3336054104477615,
"grad_norm": 0.551495047944463,
"learning_rate": 7.1502362635472135e-06,
"loss": 0.271,
"step": 37165
},
{
"epoch": 4.334188432835821,
"grad_norm": 0.5804205968353219,
"learning_rate": 7.146536438069963e-06,
"loss": 0.2753,
"step": 37170
},
{
"epoch": 4.334771455223881,
"grad_norm": 0.5042503043643048,
"learning_rate": 7.142839638957743e-06,
"loss": 0.2556,
"step": 37175
},
{
"epoch": 4.33535447761194,
"grad_norm": 0.5813491586359381,
"learning_rate": 7.139145866760217e-06,
"loss": 0.2858,
"step": 37180
},
{
"epoch": 4.3359375,
"grad_norm": 0.485487703216082,
"learning_rate": 7.1354551220266216e-06,
"loss": 0.2598,
"step": 37185
},
{
"epoch": 4.33652052238806,
"grad_norm": 0.48698592173326216,
"learning_rate": 7.1317674053057335e-06,
"loss": 0.2445,
"step": 37190
},
{
"epoch": 4.337103544776119,
"grad_norm": 0.5703412584081826,
"learning_rate": 7.128082717145881e-06,
"loss": 0.2636,
"step": 37195
},
{
"epoch": 4.337686567164179,
"grad_norm": 0.5710686569373065,
"learning_rate": 7.124401058094938e-06,
"loss": 0.2658,
"step": 37200
},
{
"epoch": 4.3382695895522385,
"grad_norm": 0.5424660992459078,
"learning_rate": 7.120722428700342e-06,
"loss": 0.2525,
"step": 37205
},
{
"epoch": 4.338852611940299,
"grad_norm": 0.5074965430480484,
"learning_rate": 7.117046829509057e-06,
"loss": 0.2566,
"step": 37210
},
{
"epoch": 4.339435634328359,
"grad_norm": 0.5156586706368018,
"learning_rate": 7.1133742610676196e-06,
"loss": 0.2725,
"step": 37215
},
{
"epoch": 4.340018656716418,
"grad_norm": 0.5239252788403638,
"learning_rate": 7.109704723922094e-06,
"loss": 0.2687,
"step": 37220
},
{
"epoch": 4.340601679104478,
"grad_norm": 0.5231514571935841,
"learning_rate": 7.106038218618125e-06,
"loss": 0.2562,
"step": 37225
},
{
"epoch": 4.341184701492537,
"grad_norm": 0.4985416893829604,
"learning_rate": 7.102374745700866e-06,
"loss": 0.2724,
"step": 37230
},
{
"epoch": 4.341767723880597,
"grad_norm": 0.5856112213107807,
"learning_rate": 7.098714305715051e-06,
"loss": 0.2606,
"step": 37235
},
{
"epoch": 4.342350746268656,
"grad_norm": 0.5305915503730493,
"learning_rate": 7.0950568992049494e-06,
"loss": 0.2516,
"step": 37240
},
{
"epoch": 4.342933768656716,
"grad_norm": 0.506075485166218,
"learning_rate": 7.091402526714383e-06,
"loss": 0.2598,
"step": 37245
},
{
"epoch": 4.3435167910447765,
"grad_norm": 0.5338640013791274,
"learning_rate": 7.087751188786723e-06,
"loss": 0.2665,
"step": 37250
},
{
"epoch": 4.344099813432836,
"grad_norm": 0.5025960931021678,
"learning_rate": 7.084102885964892e-06,
"loss": 0.2612,
"step": 37255
},
{
"epoch": 4.344682835820896,
"grad_norm": 0.550167818814279,
"learning_rate": 7.080457618791344e-06,
"loss": 0.2632,
"step": 37260
},
{
"epoch": 4.345265858208955,
"grad_norm": 0.5413176458958009,
"learning_rate": 7.076815387808115e-06,
"loss": 0.2762,
"step": 37265
},
{
"epoch": 4.345848880597015,
"grad_norm": 0.5073742095205621,
"learning_rate": 7.0731761935567495e-06,
"loss": 0.2718,
"step": 37270
},
{
"epoch": 4.346431902985074,
"grad_norm": 0.5089404131817162,
"learning_rate": 7.0695400365783784e-06,
"loss": 0.243,
"step": 37275
},
{
"epoch": 4.347014925373134,
"grad_norm": 0.4872564680605731,
"learning_rate": 7.0659069174136544e-06,
"loss": 0.2399,
"step": 37280
},
{
"epoch": 4.347597947761194,
"grad_norm": 0.5262780654065554,
"learning_rate": 7.062276836602786e-06,
"loss": 0.257,
"step": 37285
},
{
"epoch": 4.3481809701492535,
"grad_norm": 0.5568681795604292,
"learning_rate": 7.058649794685537e-06,
"loss": 0.2523,
"step": 37290
},
{
"epoch": 4.348763992537314,
"grad_norm": 0.5185483182676475,
"learning_rate": 7.055025792201212e-06,
"loss": 0.2591,
"step": 37295
},
{
"epoch": 4.349347014925373,
"grad_norm": 0.5385127587917762,
"learning_rate": 7.051404829688663e-06,
"loss": 0.2492,
"step": 37300
},
{
"epoch": 4.349930037313433,
"grad_norm": 0.5419968278587466,
"learning_rate": 7.047786907686296e-06,
"loss": 0.2554,
"step": 37305
},
{
"epoch": 4.350513059701493,
"grad_norm": 0.5825834198515136,
"learning_rate": 7.044172026732059e-06,
"loss": 0.2765,
"step": 37310
},
{
"epoch": 4.351096082089552,
"grad_norm": 0.5198659855109098,
"learning_rate": 7.040560187363447e-06,
"loss": 0.2597,
"step": 37315
},
{
"epoch": 4.351679104477612,
"grad_norm": 0.5494992849010548,
"learning_rate": 7.036951390117512e-06,
"loss": 0.2596,
"step": 37320
},
{
"epoch": 4.352262126865671,
"grad_norm": 0.5021895742655129,
"learning_rate": 7.033345635530844e-06,
"loss": 0.2605,
"step": 37325
},
{
"epoch": 4.3528451492537314,
"grad_norm": 0.5077176016169493,
"learning_rate": 7.029742924139586e-06,
"loss": 0.2593,
"step": 37330
},
{
"epoch": 4.353428171641791,
"grad_norm": 0.5192865679477117,
"learning_rate": 7.0261432564794255e-06,
"loss": 0.2536,
"step": 37335
},
{
"epoch": 4.354011194029851,
"grad_norm": 0.4977951065713548,
"learning_rate": 7.022546633085604e-06,
"loss": 0.2628,
"step": 37340
},
{
"epoch": 4.354594216417911,
"grad_norm": 0.9155990939611681,
"learning_rate": 7.018953054492889e-06,
"loss": 0.2565,
"step": 37345
},
{
"epoch": 4.35517723880597,
"grad_norm": 0.525334257689392,
"learning_rate": 7.015362521235632e-06,
"loss": 0.2483,
"step": 37350
},
{
"epoch": 4.35576026119403,
"grad_norm": 0.5142064584916372,
"learning_rate": 7.011775033847698e-06,
"loss": 0.2713,
"step": 37355
},
{
"epoch": 4.356343283582089,
"grad_norm": 0.5175803551347481,
"learning_rate": 7.008190592862514e-06,
"loss": 0.2664,
"step": 37360
},
{
"epoch": 4.356926305970149,
"grad_norm": 0.5101006717702697,
"learning_rate": 7.004609198813053e-06,
"loss": 0.2556,
"step": 37365
},
{
"epoch": 4.357509328358209,
"grad_norm": 0.5424198649205665,
"learning_rate": 7.0010308522318355e-06,
"loss": 0.2663,
"step": 37370
},
{
"epoch": 4.3580923507462686,
"grad_norm": 0.49956238051452523,
"learning_rate": 6.997455553650924e-06,
"loss": 0.252,
"step": 37375
},
{
"epoch": 4.358675373134329,
"grad_norm": 0.5042968582061769,
"learning_rate": 6.9938833036019365e-06,
"loss": 0.2585,
"step": 37380
},
{
"epoch": 4.359258395522388,
"grad_norm": 0.5095650472894148,
"learning_rate": 6.990314102616022e-06,
"loss": 0.2693,
"step": 37385
},
{
"epoch": 4.359841417910448,
"grad_norm": 0.4672276363984919,
"learning_rate": 6.9867479512239e-06,
"loss": 0.2693,
"step": 37390
},
{
"epoch": 4.360424440298507,
"grad_norm": 0.4978536274002948,
"learning_rate": 6.98318484995581e-06,
"loss": 0.2423,
"step": 37395
},
{
"epoch": 4.361007462686567,
"grad_norm": 0.515237297087287,
"learning_rate": 6.979624799341565e-06,
"loss": 0.2673,
"step": 37400
},
{
"epoch": 4.361590485074627,
"grad_norm": 0.5222560195887122,
"learning_rate": 6.976067799910499e-06,
"loss": 0.2682,
"step": 37405
},
{
"epoch": 4.362173507462686,
"grad_norm": 0.5307135877330541,
"learning_rate": 6.972513852191508e-06,
"loss": 0.2688,
"step": 37410
},
{
"epoch": 4.3627565298507465,
"grad_norm": 0.5217438038504087,
"learning_rate": 6.968962956713028e-06,
"loss": 0.2642,
"step": 37415
},
{
"epoch": 4.363339552238806,
"grad_norm": 0.5155046445501493,
"learning_rate": 6.965415114003046e-06,
"loss": 0.2494,
"step": 37420
},
{
"epoch": 4.363922574626866,
"grad_norm": 0.5647541931193226,
"learning_rate": 6.96187032458909e-06,
"loss": 0.2619,
"step": 37425
},
{
"epoch": 4.364505597014926,
"grad_norm": 0.5424547728578071,
"learning_rate": 6.958328588998242e-06,
"loss": 0.2639,
"step": 37430
},
{
"epoch": 4.365088619402985,
"grad_norm": 0.5574838744443145,
"learning_rate": 6.954789907757112e-06,
"loss": 0.2459,
"step": 37435
},
{
"epoch": 4.365671641791045,
"grad_norm": 0.5332672524908634,
"learning_rate": 6.951254281391881e-06,
"loss": 0.2557,
"step": 37440
},
{
"epoch": 4.366254664179104,
"grad_norm": 0.512641420673527,
"learning_rate": 6.947721710428251e-06,
"loss": 0.2584,
"step": 37445
},
{
"epoch": 4.366837686567164,
"grad_norm": 0.503181365629127,
"learning_rate": 6.944192195391494e-06,
"loss": 0.2521,
"step": 37450
},
{
"epoch": 4.3674207089552235,
"grad_norm": 0.5355075621911289,
"learning_rate": 6.9406657368064055e-06,
"loss": 0.2672,
"step": 37455
},
{
"epoch": 4.368003731343284,
"grad_norm": 0.5356084017821451,
"learning_rate": 6.937142335197338e-06,
"loss": 0.2667,
"step": 37460
},
{
"epoch": 4.368586753731344,
"grad_norm": 0.5057706029927774,
"learning_rate": 6.93362199108819e-06,
"loss": 0.2513,
"step": 37465
},
{
"epoch": 4.369169776119403,
"grad_norm": 0.5140002040872937,
"learning_rate": 6.930104705002403e-06,
"loss": 0.2517,
"step": 37470
},
{
"epoch": 4.369752798507463,
"grad_norm": 0.5497836785973316,
"learning_rate": 6.9265904774629585e-06,
"loss": 0.2552,
"step": 37475
},
{
"epoch": 4.370335820895522,
"grad_norm": 0.5651483183149191,
"learning_rate": 6.9230793089924005e-06,
"loss": 0.2688,
"step": 37480
},
{
"epoch": 4.370918843283582,
"grad_norm": 0.5277319736781791,
"learning_rate": 6.919571200112787e-06,
"loss": 0.2613,
"step": 37485
},
{
"epoch": 4.371501865671641,
"grad_norm": 0.5174185756631917,
"learning_rate": 6.916066151345761e-06,
"loss": 0.2588,
"step": 37490
},
{
"epoch": 4.372084888059701,
"grad_norm": 0.5221756811171331,
"learning_rate": 6.912564163212476e-06,
"loss": 0.2617,
"step": 37495
},
{
"epoch": 4.3726679104477615,
"grad_norm": 0.5485975180262437,
"learning_rate": 6.909065236233644e-06,
"loss": 0.2635,
"step": 37500
},
{
"epoch": 4.373250932835821,
"grad_norm": 0.4977335097480914,
"learning_rate": 6.90556937092953e-06,
"loss": 0.2471,
"step": 37505
},
{
"epoch": 4.373833955223881,
"grad_norm": 0.5403404673699739,
"learning_rate": 6.90207656781993e-06,
"loss": 0.2619,
"step": 37510
},
{
"epoch": 4.37441697761194,
"grad_norm": 0.5049022953370631,
"learning_rate": 6.898586827424194e-06,
"loss": 0.2455,
"step": 37515
},
{
"epoch": 4.375,
"grad_norm": 0.5599830111419839,
"learning_rate": 6.8951001502612065e-06,
"loss": 0.2542,
"step": 37520
},
{
"epoch": 4.37558302238806,
"grad_norm": 0.48238430975932345,
"learning_rate": 6.891616536849416e-06,
"loss": 0.2606,
"step": 37525
},
{
"epoch": 4.376166044776119,
"grad_norm": 0.5316769961969444,
"learning_rate": 6.888135987706787e-06,
"loss": 0.2825,
"step": 37530
},
{
"epoch": 4.376749067164179,
"grad_norm": 0.5871819644670407,
"learning_rate": 6.884658503350851e-06,
"loss": 0.2695,
"step": 37535
},
{
"epoch": 4.3773320895522385,
"grad_norm": 0.5949800330782344,
"learning_rate": 6.881184084298675e-06,
"loss": 0.2643,
"step": 37540
},
{
"epoch": 4.377915111940299,
"grad_norm": 0.5479774258194761,
"learning_rate": 6.877712731066875e-06,
"loss": 0.2632,
"step": 37545
},
{
"epoch": 4.378498134328359,
"grad_norm": 0.47928653674232297,
"learning_rate": 6.874244444171607e-06,
"loss": 0.2489,
"step": 37550
},
{
"epoch": 4.379081156716418,
"grad_norm": 0.7645855332023864,
"learning_rate": 6.870779224128571e-06,
"loss": 0.2432,
"step": 37555
},
{
"epoch": 4.379664179104478,
"grad_norm": 0.4941948185524788,
"learning_rate": 6.867317071453007e-06,
"loss": 0.2599,
"step": 37560
},
{
"epoch": 4.380247201492537,
"grad_norm": 0.5260540431972204,
"learning_rate": 6.863857986659716e-06,
"loss": 0.2419,
"step": 37565
},
{
"epoch": 4.380830223880597,
"grad_norm": 0.48896765210516463,
"learning_rate": 6.860401970263017e-06,
"loss": 0.2547,
"step": 37570
},
{
"epoch": 4.381413246268656,
"grad_norm": 0.5288192628090171,
"learning_rate": 6.856949022776798e-06,
"loss": 0.2626,
"step": 37575
},
{
"epoch": 4.381996268656716,
"grad_norm": 0.48623900419051974,
"learning_rate": 6.8534991447144706e-06,
"loss": 0.2365,
"step": 37580
},
{
"epoch": 4.3825792910447765,
"grad_norm": 0.554592369507337,
"learning_rate": 6.850052336589008e-06,
"loss": 0.2523,
"step": 37585
},
{
"epoch": 4.383162313432836,
"grad_norm": 0.4765066061767422,
"learning_rate": 6.8466085989129066e-06,
"loss": 0.2388,
"step": 37590
},
{
"epoch": 4.383745335820896,
"grad_norm": 0.49015261367610646,
"learning_rate": 6.843167932198227e-06,
"loss": 0.2608,
"step": 37595
},
{
"epoch": 4.384328358208955,
"grad_norm": 0.5079185723432356,
"learning_rate": 6.839730336956554e-06,
"loss": 0.2529,
"step": 37600
},
{
"epoch": 4.384911380597015,
"grad_norm": 0.5183434635482761,
"learning_rate": 6.836295813699031e-06,
"loss": 0.2556,
"step": 37605
},
{
"epoch": 4.385494402985074,
"grad_norm": 0.5254416040626212,
"learning_rate": 6.83286436293634e-06,
"loss": 0.269,
"step": 37610
},
{
"epoch": 4.386077425373134,
"grad_norm": 0.5626541633596372,
"learning_rate": 6.829435985178708e-06,
"loss": 0.253,
"step": 37615
},
{
"epoch": 4.386660447761194,
"grad_norm": 0.5549683727914495,
"learning_rate": 6.826010680935886e-06,
"loss": 0.2691,
"step": 37620
},
{
"epoch": 4.3872434701492535,
"grad_norm": 0.49828536853764854,
"learning_rate": 6.8225884507172005e-06,
"loss": 0.25,
"step": 37625
},
{
"epoch": 4.387826492537314,
"grad_norm": 0.4931098129488875,
"learning_rate": 6.819169295031493e-06,
"loss": 0.2566,
"step": 37630
},
{
"epoch": 4.388409514925373,
"grad_norm": 0.5532586822967704,
"learning_rate": 6.815753214387172e-06,
"loss": 0.2801,
"step": 37635
},
{
"epoch": 4.388992537313433,
"grad_norm": 0.5817928904723288,
"learning_rate": 6.812340209292164e-06,
"loss": 0.2634,
"step": 37640
},
{
"epoch": 4.389575559701493,
"grad_norm": 0.496738795223984,
"learning_rate": 6.808930280253956e-06,
"loss": 0.2596,
"step": 37645
},
{
"epoch": 4.390158582089552,
"grad_norm": 0.5210410477638936,
"learning_rate": 6.80552342777957e-06,
"loss": 0.2594,
"step": 37650
},
{
"epoch": 4.390741604477612,
"grad_norm": 0.5606455872288548,
"learning_rate": 6.80211965237557e-06,
"loss": 0.2729,
"step": 37655
},
{
"epoch": 4.391324626865671,
"grad_norm": 0.5408706260037408,
"learning_rate": 6.79871895454807e-06,
"loss": 0.2674,
"step": 37660
},
{
"epoch": 4.3919076492537314,
"grad_norm": 0.6566004981267262,
"learning_rate": 6.7953213348027235e-06,
"loss": 0.2562,
"step": 37665
},
{
"epoch": 4.392490671641791,
"grad_norm": 0.4786957306654934,
"learning_rate": 6.791926793644713e-06,
"loss": 0.2554,
"step": 37670
},
{
"epoch": 4.393073694029851,
"grad_norm": 0.5420558106985854,
"learning_rate": 6.7885353315787825e-06,
"loss": 0.2707,
"step": 37675
},
{
"epoch": 4.393656716417911,
"grad_norm": 0.5141114362646728,
"learning_rate": 6.785146949109206e-06,
"loss": 0.2483,
"step": 37680
},
{
"epoch": 4.39423973880597,
"grad_norm": 0.49279316333047224,
"learning_rate": 6.781761646739805e-06,
"loss": 0.253,
"step": 37685
},
{
"epoch": 4.39482276119403,
"grad_norm": 0.5237432022683991,
"learning_rate": 6.778379424973943e-06,
"loss": 0.2467,
"step": 37690
},
{
"epoch": 4.395405783582089,
"grad_norm": 0.528539998651686,
"learning_rate": 6.775000284314523e-06,
"loss": 0.2415,
"step": 37695
},
{
"epoch": 4.395988805970149,
"grad_norm": 0.5355518181400558,
"learning_rate": 6.771624225263994e-06,
"loss": 0.2586,
"step": 37700
},
{
"epoch": 4.396571828358209,
"grad_norm": 0.5007886297156655,
"learning_rate": 6.768251248324333e-06,
"loss": 0.26,
"step": 37705
},
{
"epoch": 4.3971548507462686,
"grad_norm": 0.5096148207366898,
"learning_rate": 6.764881353997082e-06,
"loss": 0.2583,
"step": 37710
},
{
"epoch": 4.397737873134329,
"grad_norm": 0.5326093420108741,
"learning_rate": 6.761514542783308e-06,
"loss": 0.2721,
"step": 37715
},
{
"epoch": 4.398320895522388,
"grad_norm": 0.5451094943829453,
"learning_rate": 6.758150815183618e-06,
"loss": 0.2727,
"step": 37720
},
{
"epoch": 4.398903917910448,
"grad_norm": 0.51444907180675,
"learning_rate": 6.7547901716981704e-06,
"loss": 0.2681,
"step": 37725
},
{
"epoch": 4.399486940298507,
"grad_norm": 0.5171071872102907,
"learning_rate": 6.751432612826664e-06,
"loss": 0.2643,
"step": 37730
},
{
"epoch": 4.400069962686567,
"grad_norm": 0.5994903914611903,
"learning_rate": 6.748078139068327e-06,
"loss": 0.2643,
"step": 37735
},
{
"epoch": 4.400652985074627,
"grad_norm": 0.5187717746664459,
"learning_rate": 6.7447267509219494e-06,
"loss": 0.2472,
"step": 37740
},
{
"epoch": 4.401236007462686,
"grad_norm": 0.5098504560103423,
"learning_rate": 6.741378448885838e-06,
"loss": 0.2722,
"step": 37745
},
{
"epoch": 4.4018190298507465,
"grad_norm": 0.5292380796106622,
"learning_rate": 6.738033233457863e-06,
"loss": 0.2529,
"step": 37750
},
{
"epoch": 4.402402052238806,
"grad_norm": 0.5311520117308983,
"learning_rate": 6.734691105135417e-06,
"loss": 0.2552,
"step": 37755
},
{
"epoch": 4.402985074626866,
"grad_norm": 0.5129788185616609,
"learning_rate": 6.7313520644154555e-06,
"loss": 0.248,
"step": 37760
},
{
"epoch": 4.403568097014926,
"grad_norm": 0.48753517716534006,
"learning_rate": 6.7280161117944495e-06,
"loss": 0.2562,
"step": 37765
},
{
"epoch": 4.404151119402985,
"grad_norm": 0.4782181838575392,
"learning_rate": 6.724683247768427e-06,
"loss": 0.2416,
"step": 37770
},
{
"epoch": 4.404734141791045,
"grad_norm": 0.5240340461399262,
"learning_rate": 6.721353472832953e-06,
"loss": 0.2448,
"step": 37775
},
{
"epoch": 4.405317164179104,
"grad_norm": 0.545465169069081,
"learning_rate": 6.718026787483131e-06,
"loss": 0.2622,
"step": 37780
},
{
"epoch": 4.405900186567164,
"grad_norm": 0.5545305276793467,
"learning_rate": 6.714703192213614e-06,
"loss": 0.2803,
"step": 37785
},
{
"epoch": 4.4064832089552235,
"grad_norm": 0.5516414717751895,
"learning_rate": 6.7113826875185885e-06,
"loss": 0.2627,
"step": 37790
},
{
"epoch": 4.407066231343284,
"grad_norm": 0.5450123974641333,
"learning_rate": 6.7080652738917655e-06,
"loss": 0.2617,
"step": 37795
},
{
"epoch": 4.407649253731344,
"grad_norm": 0.503846458792622,
"learning_rate": 6.704750951826438e-06,
"loss": 0.2571,
"step": 37800
},
{
"epoch": 4.408232276119403,
"grad_norm": 0.5109976775523121,
"learning_rate": 6.701439721815391e-06,
"loss": 0.2755,
"step": 37805
},
{
"epoch": 4.408815298507463,
"grad_norm": 0.5148988499513758,
"learning_rate": 6.698131584350989e-06,
"loss": 0.2503,
"step": 37810
},
{
"epoch": 4.409398320895522,
"grad_norm": 0.5485123842856282,
"learning_rate": 6.69482653992511e-06,
"loss": 0.2774,
"step": 37815
},
{
"epoch": 4.409981343283582,
"grad_norm": 0.5346591042205974,
"learning_rate": 6.691524589029188e-06,
"loss": 0.2549,
"step": 37820
},
{
"epoch": 4.410564365671641,
"grad_norm": 0.4997498350650287,
"learning_rate": 6.688225732154189e-06,
"loss": 0.2543,
"step": 37825
},
{
"epoch": 4.411147388059701,
"grad_norm": 0.4910896586352394,
"learning_rate": 6.684929969790622e-06,
"loss": 0.2528,
"step": 37830
},
{
"epoch": 4.4117304104477615,
"grad_norm": 0.5400898482405075,
"learning_rate": 6.6816373024285365e-06,
"loss": 0.2726,
"step": 37835
},
{
"epoch": 4.412313432835821,
"grad_norm": 0.5161745050709917,
"learning_rate": 6.6783477305575215e-06,
"loss": 0.2599,
"step": 37840
},
{
"epoch": 4.412896455223881,
"grad_norm": 0.5276429255682854,
"learning_rate": 6.675061254666702e-06,
"loss": 0.2581,
"step": 37845
},
{
"epoch": 4.41347947761194,
"grad_norm": 0.4720099071014724,
"learning_rate": 6.671777875244745e-06,
"loss": 0.2557,
"step": 37850
},
{
"epoch": 4.4140625,
"grad_norm": 0.5678187682080887,
"learning_rate": 6.668497592779857e-06,
"loss": 0.2711,
"step": 37855
},
{
"epoch": 4.41464552238806,
"grad_norm": 0.5280209993381787,
"learning_rate": 6.665220407759788e-06,
"loss": 0.2647,
"step": 37860
},
{
"epoch": 4.415228544776119,
"grad_norm": 0.5225296458493709,
"learning_rate": 6.661946320671822e-06,
"loss": 0.2525,
"step": 37865
},
{
"epoch": 4.415811567164179,
"grad_norm": 0.5453569617633768,
"learning_rate": 6.658675332002787e-06,
"loss": 0.2592,
"step": 37870
},
{
"epoch": 4.4163945895522385,
"grad_norm": 0.5199782382147519,
"learning_rate": 6.655407442239047e-06,
"loss": 0.2426,
"step": 37875
},
{
"epoch": 4.416977611940299,
"grad_norm": 0.5335249567633386,
"learning_rate": 6.652142651866497e-06,
"loss": 0.2642,
"step": 37880
},
{
"epoch": 4.417560634328359,
"grad_norm": 0.5715476011975159,
"learning_rate": 6.648880961370593e-06,
"loss": 0.261,
"step": 37885
},
{
"epoch": 4.418143656716418,
"grad_norm": 0.4740952002785889,
"learning_rate": 6.645622371236314e-06,
"loss": 0.254,
"step": 37890
},
{
"epoch": 4.418726679104478,
"grad_norm": 0.5514382405706211,
"learning_rate": 6.642366881948173e-06,
"loss": 0.2622,
"step": 37895
},
{
"epoch": 4.419309701492537,
"grad_norm": 0.4822002285688902,
"learning_rate": 6.639114493990238e-06,
"loss": 0.2498,
"step": 37900
},
{
"epoch": 4.419892723880597,
"grad_norm": 0.5564968018417733,
"learning_rate": 6.635865207846106e-06,
"loss": 0.263,
"step": 37905
},
{
"epoch": 4.420475746268656,
"grad_norm": 0.5677505103823726,
"learning_rate": 6.6326190239989135e-06,
"loss": 0.2576,
"step": 37910
},
{
"epoch": 4.421058768656716,
"grad_norm": 0.5207445653359911,
"learning_rate": 6.629375942931345e-06,
"loss": 0.2482,
"step": 37915
},
{
"epoch": 4.4216417910447765,
"grad_norm": 0.5403295097755192,
"learning_rate": 6.626135965125597e-06,
"loss": 0.2512,
"step": 37920
},
{
"epoch": 4.422224813432836,
"grad_norm": 0.6004710947281864,
"learning_rate": 6.622899091063442e-06,
"loss": 0.2824,
"step": 37925
},
{
"epoch": 4.422807835820896,
"grad_norm": 0.49440557059003526,
"learning_rate": 6.61966532122616e-06,
"loss": 0.2607,
"step": 37930
},
{
"epoch": 4.423390858208955,
"grad_norm": 0.5045026705601957,
"learning_rate": 6.6164346560945935e-06,
"loss": 0.2319,
"step": 37935
},
{
"epoch": 4.423973880597015,
"grad_norm": 0.5545947136972534,
"learning_rate": 6.613207096149099e-06,
"loss": 0.2706,
"step": 37940
},
{
"epoch": 4.424556902985074,
"grad_norm": 0.5364397445538137,
"learning_rate": 6.609982641869591e-06,
"loss": 0.2636,
"step": 37945
},
{
"epoch": 4.425139925373134,
"grad_norm": 0.49659530270108676,
"learning_rate": 6.606761293735513e-06,
"loss": 0.2552,
"step": 37950
},
{
"epoch": 4.425722947761194,
"grad_norm": 0.46371638977704815,
"learning_rate": 6.6035430522258455e-06,
"loss": 0.2522,
"step": 37955
},
{
"epoch": 4.4263059701492535,
"grad_norm": 0.5048728294319823,
"learning_rate": 6.600327917819114e-06,
"loss": 0.2493,
"step": 37960
},
{
"epoch": 4.426888992537314,
"grad_norm": 0.5910748256828204,
"learning_rate": 6.597115890993383e-06,
"loss": 0.2545,
"step": 37965
},
{
"epoch": 4.427472014925373,
"grad_norm": 0.5140618458048315,
"learning_rate": 6.593906972226238e-06,
"loss": 0.2373,
"step": 37970
},
{
"epoch": 4.428055037313433,
"grad_norm": 0.5607261448709014,
"learning_rate": 6.590701161994828e-06,
"loss": 0.2636,
"step": 37975
},
{
"epoch": 4.428638059701493,
"grad_norm": 0.5190995583520444,
"learning_rate": 6.587498460775811e-06,
"loss": 0.2669,
"step": 37980
},
{
"epoch": 4.429221082089552,
"grad_norm": 0.5015700867310398,
"learning_rate": 6.5842988690454135e-06,
"loss": 0.2624,
"step": 37985
},
{
"epoch": 4.429804104477612,
"grad_norm": 0.5565323088619271,
"learning_rate": 6.581102387279374e-06,
"loss": 0.2737,
"step": 37990
},
{
"epoch": 4.430387126865671,
"grad_norm": 0.5543711186726025,
"learning_rate": 6.577909015952982e-06,
"loss": 0.2716,
"step": 37995
},
{
"epoch": 4.4309701492537314,
"grad_norm": 0.5544970341913353,
"learning_rate": 6.574718755541061e-06,
"loss": 0.2605,
"step": 38000
},
{
"epoch": 4.431553171641791,
"grad_norm": 0.554823940207235,
"learning_rate": 6.571531606517972e-06,
"loss": 0.2634,
"step": 38005
},
{
"epoch": 4.432136194029851,
"grad_norm": 0.5510808119909901,
"learning_rate": 6.568347569357611e-06,
"loss": 0.246,
"step": 38010
},
{
"epoch": 4.432719216417911,
"grad_norm": 0.5381041837346422,
"learning_rate": 6.565166644533424e-06,
"loss": 0.2614,
"step": 38015
},
{
"epoch": 4.43330223880597,
"grad_norm": 0.4887734043557818,
"learning_rate": 6.561988832518367e-06,
"loss": 0.2555,
"step": 38020
},
{
"epoch": 4.43388526119403,
"grad_norm": 0.5137112665327495,
"learning_rate": 6.558814133784966e-06,
"loss": 0.272,
"step": 38025
},
{
"epoch": 4.434468283582089,
"grad_norm": 0.5898989785214453,
"learning_rate": 6.555642548805262e-06,
"loss": 0.2818,
"step": 38030
},
{
"epoch": 4.435051305970149,
"grad_norm": 0.5579589351581093,
"learning_rate": 6.552474078050835e-06,
"loss": 0.2596,
"step": 38035
},
{
"epoch": 4.435634328358209,
"grad_norm": 0.5401232346234285,
"learning_rate": 6.5493087219928114e-06,
"loss": 0.2668,
"step": 38040
},
{
"epoch": 4.4362173507462686,
"grad_norm": 0.49770926007981614,
"learning_rate": 6.54614648110185e-06,
"loss": 0.2446,
"step": 38045
},
{
"epoch": 4.436800373134329,
"grad_norm": 0.49130703990350055,
"learning_rate": 6.542987355848144e-06,
"loss": 0.2533,
"step": 38050
},
{
"epoch": 4.437383395522388,
"grad_norm": 0.5101533520889702,
"learning_rate": 6.539831346701426e-06,
"loss": 0.255,
"step": 38055
},
{
"epoch": 4.437966417910448,
"grad_norm": 0.5144750523870821,
"learning_rate": 6.536678454130965e-06,
"loss": 0.2804,
"step": 38060
},
{
"epoch": 4.438549440298507,
"grad_norm": 0.5594827325821637,
"learning_rate": 6.53352867860556e-06,
"loss": 0.2564,
"step": 38065
},
{
"epoch": 4.439132462686567,
"grad_norm": 0.5343688389768886,
"learning_rate": 6.530382020593559e-06,
"loss": 0.2604,
"step": 38070
},
{
"epoch": 4.439715485074627,
"grad_norm": 0.5539208089956309,
"learning_rate": 6.527238480562838e-06,
"loss": 0.2551,
"step": 38075
},
{
"epoch": 4.440298507462686,
"grad_norm": 0.49755846486433397,
"learning_rate": 6.52409805898081e-06,
"loss": 0.2571,
"step": 38080
},
{
"epoch": 4.4408815298507465,
"grad_norm": 0.5458675830435409,
"learning_rate": 6.520960756314427e-06,
"loss": 0.2634,
"step": 38085
},
{
"epoch": 4.441464552238806,
"grad_norm": 0.5535817242089792,
"learning_rate": 6.517826573030178e-06,
"loss": 0.2733,
"step": 38090
},
{
"epoch": 4.442047574626866,
"grad_norm": 0.5784905661614941,
"learning_rate": 6.514695509594076e-06,
"loss": 0.2654,
"step": 38095
},
{
"epoch": 4.442630597014926,
"grad_norm": 0.5430423361040847,
"learning_rate": 6.511567566471697e-06,
"loss": 0.2588,
"step": 38100
},
{
"epoch": 4.443213619402985,
"grad_norm": 0.5339731848505547,
"learning_rate": 6.50844274412812e-06,
"loss": 0.2606,
"step": 38105
},
{
"epoch": 4.443796641791045,
"grad_norm": 0.5265808645716251,
"learning_rate": 6.50532104302799e-06,
"loss": 0.2495,
"step": 38110
},
{
"epoch": 4.444379664179104,
"grad_norm": 0.5436833957811948,
"learning_rate": 6.5022024636354605e-06,
"loss": 0.2536,
"step": 38115
},
{
"epoch": 4.444962686567164,
"grad_norm": 0.47978330218350496,
"learning_rate": 6.499087006414245e-06,
"loss": 0.2654,
"step": 38120
},
{
"epoch": 4.4455457089552235,
"grad_norm": 0.5125491666211928,
"learning_rate": 6.495974671827574e-06,
"loss": 0.2461,
"step": 38125
},
{
"epoch": 4.446128731343284,
"grad_norm": 0.536205379662219,
"learning_rate": 6.492865460338228e-06,
"loss": 0.2439,
"step": 38130
},
{
"epoch": 4.446711753731344,
"grad_norm": 0.5100336962710121,
"learning_rate": 6.489759372408514e-06,
"loss": 0.2605,
"step": 38135
},
{
"epoch": 4.447294776119403,
"grad_norm": 0.5500325924585305,
"learning_rate": 6.4866564085002826e-06,
"loss": 0.262,
"step": 38140
},
{
"epoch": 4.447877798507463,
"grad_norm": 0.5329048459627673,
"learning_rate": 6.483556569074904e-06,
"loss": 0.2682,
"step": 38145
},
{
"epoch": 4.448460820895522,
"grad_norm": 0.5663551121334356,
"learning_rate": 6.480459854593305e-06,
"loss": 0.2626,
"step": 38150
},
{
"epoch": 4.449043843283582,
"grad_norm": 0.5232106945335262,
"learning_rate": 6.477366265515931e-06,
"loss": 0.2641,
"step": 38155
},
{
"epoch": 4.449626865671641,
"grad_norm": 0.5389140609005297,
"learning_rate": 6.474275802302776e-06,
"loss": 0.2713,
"step": 38160
},
{
"epoch": 4.450209888059701,
"grad_norm": 0.5200787546279895,
"learning_rate": 6.471188465413355e-06,
"loss": 0.2444,
"step": 38165
},
{
"epoch": 4.4507929104477615,
"grad_norm": 0.5309435731380026,
"learning_rate": 6.468104255306728e-06,
"loss": 0.2638,
"step": 38170
},
{
"epoch": 4.451375932835821,
"grad_norm": 0.49511464563093877,
"learning_rate": 6.465023172441489e-06,
"loss": 0.2556,
"step": 38175
},
{
"epoch": 4.451958955223881,
"grad_norm": 0.5245319387742695,
"learning_rate": 6.461945217275761e-06,
"loss": 0.2643,
"step": 38180
},
{
"epoch": 4.45254197761194,
"grad_norm": 0.48381562244638704,
"learning_rate": 6.458870390267213e-06,
"loss": 0.2367,
"step": 38185
},
{
"epoch": 4.453125,
"grad_norm": 0.4945572418335977,
"learning_rate": 6.455798691873042e-06,
"loss": 0.2516,
"step": 38190
},
{
"epoch": 4.45370802238806,
"grad_norm": 0.5119796477199425,
"learning_rate": 6.45273012254997e-06,
"loss": 0.2451,
"step": 38195
},
{
"epoch": 4.454291044776119,
"grad_norm": 0.5380360042826006,
"learning_rate": 6.449664682754278e-06,
"loss": 0.272,
"step": 38200
},
{
"epoch": 4.454874067164179,
"grad_norm": 0.5197699392470599,
"learning_rate": 6.446602372941756e-06,
"loss": 0.277,
"step": 38205
},
{
"epoch": 4.4554570895522385,
"grad_norm": 0.5280041655845327,
"learning_rate": 6.443543193567745e-06,
"loss": 0.2538,
"step": 38210
},
{
"epoch": 4.456040111940299,
"grad_norm": 0.5303693518474754,
"learning_rate": 6.440487145087116e-06,
"loss": 0.2748,
"step": 38215
},
{
"epoch": 4.456623134328359,
"grad_norm": 0.5591970954489205,
"learning_rate": 6.4374342279542726e-06,
"loss": 0.2637,
"step": 38220
},
{
"epoch": 4.457206156716418,
"grad_norm": 0.530588507584021,
"learning_rate": 6.434384442623156e-06,
"loss": 0.2499,
"step": 38225
},
{
"epoch": 4.457789179104478,
"grad_norm": 0.5370039562897317,
"learning_rate": 6.431337789547239e-06,
"loss": 0.2626,
"step": 38230
},
{
"epoch": 4.458372201492537,
"grad_norm": 0.5046364169965343,
"learning_rate": 6.428294269179531e-06,
"loss": 0.255,
"step": 38235
},
{
"epoch": 4.458955223880597,
"grad_norm": 0.561326171816721,
"learning_rate": 6.425253881972573e-06,
"loss": 0.242,
"step": 38240
},
{
"epoch": 4.459538246268656,
"grad_norm": 0.5725584683557882,
"learning_rate": 6.4222166283784415e-06,
"loss": 0.2591,
"step": 38245
},
{
"epoch": 4.460121268656716,
"grad_norm": 0.5328326967447862,
"learning_rate": 6.419182508848745e-06,
"loss": 0.2544,
"step": 38250
},
{
"epoch": 4.4607042910447765,
"grad_norm": 0.5443908173188287,
"learning_rate": 6.416151523834632e-06,
"loss": 0.2626,
"step": 38255
},
{
"epoch": 4.461287313432836,
"grad_norm": 0.5190674201133977,
"learning_rate": 6.4131236737867795e-06,
"loss": 0.2438,
"step": 38260
},
{
"epoch": 4.461870335820896,
"grad_norm": 0.5566166517489323,
"learning_rate": 6.4100989591554026e-06,
"loss": 0.2647,
"step": 38265
},
{
"epoch": 4.462453358208955,
"grad_norm": 0.5444030570437454,
"learning_rate": 6.407077380390236e-06,
"loss": 0.2546,
"step": 38270
},
{
"epoch": 4.463036380597015,
"grad_norm": 0.5236342301254023,
"learning_rate": 6.4040589379405765e-06,
"loss": 0.2515,
"step": 38275
},
{
"epoch": 4.463619402985074,
"grad_norm": 0.5036287820632196,
"learning_rate": 6.4010436322552204e-06,
"loss": 0.2482,
"step": 38280
},
{
"epoch": 4.464202425373134,
"grad_norm": 0.5342037509395847,
"learning_rate": 6.39803146378253e-06,
"loss": 0.2547,
"step": 38285
},
{
"epoch": 4.464785447761194,
"grad_norm": 0.5877368789124545,
"learning_rate": 6.395022432970375e-06,
"loss": 0.2455,
"step": 38290
},
{
"epoch": 4.4653684701492535,
"grad_norm": 0.5203001059160115,
"learning_rate": 6.392016540266179e-06,
"loss": 0.2627,
"step": 38295
},
{
"epoch": 4.465951492537314,
"grad_norm": 0.5434466886587166,
"learning_rate": 6.389013786116878e-06,
"loss": 0.2708,
"step": 38300
},
{
"epoch": 4.466534514925373,
"grad_norm": 0.5172868659600586,
"learning_rate": 6.3860141709689615e-06,
"loss": 0.259,
"step": 38305
},
{
"epoch": 4.467117537313433,
"grad_norm": 0.5048021760692184,
"learning_rate": 6.383017695268441e-06,
"loss": 0.2486,
"step": 38310
},
{
"epoch": 4.467700559701493,
"grad_norm": 0.5377071496318871,
"learning_rate": 6.380024359460862e-06,
"loss": 0.2563,
"step": 38315
},
{
"epoch": 4.468283582089552,
"grad_norm": 0.49424572546955253,
"learning_rate": 6.377034163991308e-06,
"loss": 0.2696,
"step": 38320
},
{
"epoch": 4.468866604477612,
"grad_norm": 0.5026516637006107,
"learning_rate": 6.3740471093043914e-06,
"loss": 0.2521,
"step": 38325
},
{
"epoch": 4.469449626865671,
"grad_norm": 0.5236120140353857,
"learning_rate": 6.3710631958442524e-06,
"loss": 0.2438,
"step": 38330
},
{
"epoch": 4.4700326492537314,
"grad_norm": 0.5368892754182085,
"learning_rate": 6.3680824240545835e-06,
"loss": 0.2617,
"step": 38335
},
{
"epoch": 4.470615671641791,
"grad_norm": 0.5631534410705222,
"learning_rate": 6.365104794378582e-06,
"loss": 0.257,
"step": 38340
},
{
"epoch": 4.471198694029851,
"grad_norm": 0.4969788859972526,
"learning_rate": 6.362130307259008e-06,
"loss": 0.2465,
"step": 38345
},
{
"epoch": 4.471781716417911,
"grad_norm": 0.5802544230889424,
"learning_rate": 6.3591589631381286e-06,
"loss": 0.2641,
"step": 38350
},
{
"epoch": 4.47236473880597,
"grad_norm": 0.5536864995144586,
"learning_rate": 6.356190762457753e-06,
"loss": 0.2657,
"step": 38355
},
{
"epoch": 4.47294776119403,
"grad_norm": 0.5275125308796202,
"learning_rate": 6.353225705659234e-06,
"loss": 0.2542,
"step": 38360
},
{
"epoch": 4.473530783582089,
"grad_norm": 0.5219923673534463,
"learning_rate": 6.350263793183439e-06,
"loss": 0.2687,
"step": 38365
},
{
"epoch": 4.474113805970149,
"grad_norm": 0.5082496370709878,
"learning_rate": 6.347305025470776e-06,
"loss": 0.2518,
"step": 38370
},
{
"epoch": 4.474696828358209,
"grad_norm": 0.5135128368336545,
"learning_rate": 6.344349402961194e-06,
"loss": 0.2694,
"step": 38375
},
{
"epoch": 4.4752798507462686,
"grad_norm": 0.5466314389638472,
"learning_rate": 6.341396926094155e-06,
"loss": 0.2733,
"step": 38380
},
{
"epoch": 4.475862873134329,
"grad_norm": 0.5845324971003246,
"learning_rate": 6.338447595308671e-06,
"loss": 0.2811,
"step": 38385
},
{
"epoch": 4.476445895522388,
"grad_norm": 0.5562813248321179,
"learning_rate": 6.335501411043274e-06,
"loss": 0.2784,
"step": 38390
},
{
"epoch": 4.477028917910448,
"grad_norm": 0.5183979516181434,
"learning_rate": 6.3325583737360376e-06,
"loss": 0.2606,
"step": 38395
},
{
"epoch": 4.477611940298507,
"grad_norm": 0.5237406722626377,
"learning_rate": 6.329618483824559e-06,
"loss": 0.2719,
"step": 38400
},
{
"epoch": 4.478194962686567,
"grad_norm": 0.4957426119917759,
"learning_rate": 6.326681741745978e-06,
"loss": 0.2471,
"step": 38405
},
{
"epoch": 4.478777985074627,
"grad_norm": 0.5375130796660804,
"learning_rate": 6.323748147936959e-06,
"loss": 0.2895,
"step": 38410
},
{
"epoch": 4.479361007462686,
"grad_norm": 0.5542027303220478,
"learning_rate": 6.320817702833689e-06,
"loss": 0.2648,
"step": 38415
},
{
"epoch": 4.4799440298507465,
"grad_norm": 0.5824272515172335,
"learning_rate": 6.317890406871914e-06,
"loss": 0.273,
"step": 38420
},
{
"epoch": 4.480527052238806,
"grad_norm": 0.5603501358301488,
"learning_rate": 6.314966260486882e-06,
"loss": 0.2716,
"step": 38425
},
{
"epoch": 4.481110074626866,
"grad_norm": 0.48460259345144363,
"learning_rate": 6.312045264113388e-06,
"loss": 0.2579,
"step": 38430
},
{
"epoch": 4.481693097014926,
"grad_norm": 0.5626337980726406,
"learning_rate": 6.309127418185755e-06,
"loss": 0.267,
"step": 38435
},
{
"epoch": 4.482276119402985,
"grad_norm": 0.5715045452812639,
"learning_rate": 6.306212723137846e-06,
"loss": 0.2565,
"step": 38440
},
{
"epoch": 4.482859141791045,
"grad_norm": 0.5188720938242106,
"learning_rate": 6.3033011794030416e-06,
"loss": 0.2516,
"step": 38445
},
{
"epoch": 4.483442164179104,
"grad_norm": 0.5138320356119981,
"learning_rate": 6.300392787414265e-06,
"loss": 0.2754,
"step": 38450
},
{
"epoch": 4.484025186567164,
"grad_norm": 0.5215779000371581,
"learning_rate": 6.297487547603958e-06,
"loss": 0.2622,
"step": 38455
},
{
"epoch": 4.4846082089552235,
"grad_norm": 0.5231899814597589,
"learning_rate": 6.2945854604041135e-06,
"loss": 0.2567,
"step": 38460
},
{
"epoch": 4.485191231343284,
"grad_norm": 0.5313579085351516,
"learning_rate": 6.291686526246232e-06,
"loss": 0.2723,
"step": 38465
},
{
"epoch": 4.485774253731344,
"grad_norm": 0.4949843565279442,
"learning_rate": 6.28879074556137e-06,
"loss": 0.2521,
"step": 38470
},
{
"epoch": 4.486357276119403,
"grad_norm": 0.5064034473611037,
"learning_rate": 6.2858981187800915e-06,
"loss": 0.2578,
"step": 38475
},
{
"epoch": 4.486940298507463,
"grad_norm": 0.4877199417010325,
"learning_rate": 6.283008646332507e-06,
"loss": 0.2597,
"step": 38480
},
{
"epoch": 4.487523320895522,
"grad_norm": 0.5049309841260072,
"learning_rate": 6.280122328648254e-06,
"loss": 0.241,
"step": 38485
},
{
"epoch": 4.488106343283582,
"grad_norm": 0.5282900932518092,
"learning_rate": 6.277239166156497e-06,
"loss": 0.2613,
"step": 38490
},
{
"epoch": 4.488689365671641,
"grad_norm": 0.5384873757949727,
"learning_rate": 6.274359159285937e-06,
"loss": 0.2677,
"step": 38495
},
{
"epoch": 4.489272388059701,
"grad_norm": 0.5147663737460121,
"learning_rate": 6.271482308464807e-06,
"loss": 0.2569,
"step": 38500
},
{
"epoch": 4.4898554104477615,
"grad_norm": 0.4827779590524776,
"learning_rate": 6.268608614120858e-06,
"loss": 0.259,
"step": 38505
},
{
"epoch": 4.490438432835821,
"grad_norm": 0.5095805849321287,
"learning_rate": 6.265738076681392e-06,
"loss": 0.2493,
"step": 38510
},
{
"epoch": 4.491021455223881,
"grad_norm": 0.6266904725920435,
"learning_rate": 6.262870696573219e-06,
"loss": 0.2782,
"step": 38515
},
{
"epoch": 4.49160447761194,
"grad_norm": 0.5694132285290291,
"learning_rate": 6.2600064742227e-06,
"loss": 0.2754,
"step": 38520
},
{
"epoch": 4.4921875,
"grad_norm": 0.5567702063766407,
"learning_rate": 6.2571454100557136e-06,
"loss": 0.2632,
"step": 38525
},
{
"epoch": 4.49277052238806,
"grad_norm": 0.5085069735159217,
"learning_rate": 6.254287504497672e-06,
"loss": 0.272,
"step": 38530
},
{
"epoch": 4.493353544776119,
"grad_norm": 0.5060260712801773,
"learning_rate": 6.251432757973519e-06,
"loss": 0.2522,
"step": 38535
},
{
"epoch": 4.493936567164179,
"grad_norm": 0.5539886671173206,
"learning_rate": 6.248581170907729e-06,
"loss": 0.2592,
"step": 38540
},
{
"epoch": 4.4945195895522385,
"grad_norm": 0.5150664889160734,
"learning_rate": 6.245732743724305e-06,
"loss": 0.2703,
"step": 38545
},
{
"epoch": 4.495102611940299,
"grad_norm": 0.531805818535227,
"learning_rate": 6.242887476846785e-06,
"loss": 0.2457,
"step": 38550
},
{
"epoch": 4.495685634328359,
"grad_norm": 0.5292251713955413,
"learning_rate": 6.2400453706982216e-06,
"loss": 0.2606,
"step": 38555
},
{
"epoch": 4.496268656716418,
"grad_norm": 0.6074805083067017,
"learning_rate": 6.237206425701223e-06,
"loss": 0.2844,
"step": 38560
},
{
"epoch": 4.496851679104478,
"grad_norm": 0.537703527035354,
"learning_rate": 6.234370642277903e-06,
"loss": 0.2656,
"step": 38565
},
{
"epoch": 4.497434701492537,
"grad_norm": 0.5654128975770102,
"learning_rate": 6.231538020849919e-06,
"loss": 0.2699,
"step": 38570
},
{
"epoch": 4.498017723880597,
"grad_norm": 0.5471811973711892,
"learning_rate": 6.228708561838452e-06,
"loss": 0.276,
"step": 38575
},
{
"epoch": 4.498600746268656,
"grad_norm": 0.5438608880087065,
"learning_rate": 6.225882265664218e-06,
"loss": 0.2602,
"step": 38580
},
{
"epoch": 4.499183768656716,
"grad_norm": 0.5218914053297654,
"learning_rate": 6.223059132747463e-06,
"loss": 0.2618,
"step": 38585
},
{
"epoch": 4.4997667910447765,
"grad_norm": 0.5101625333495837,
"learning_rate": 6.220239163507955e-06,
"loss": 0.2639,
"step": 38590
},
{
"epoch": 4.500349813432836,
"grad_norm": 0.5687057578995803,
"learning_rate": 6.217422358364999e-06,
"loss": 0.2683,
"step": 38595
},
{
"epoch": 4.500932835820896,
"grad_norm": 0.5443420946317586,
"learning_rate": 6.214608717737426e-06,
"loss": 0.2725,
"step": 38600
},
{
"epoch": 4.501515858208955,
"grad_norm": 0.5070065451751753,
"learning_rate": 6.211798242043596e-06,
"loss": 0.2487,
"step": 38605
},
{
"epoch": 4.502098880597015,
"grad_norm": 0.50271908357633,
"learning_rate": 6.2089909317014e-06,
"loss": 0.259,
"step": 38610
},
{
"epoch": 4.502681902985074,
"grad_norm": 0.5320041945977301,
"learning_rate": 6.206186787128262e-06,
"loss": 0.26,
"step": 38615
},
{
"epoch": 4.503264925373134,
"grad_norm": 0.5167408209298835,
"learning_rate": 6.2033858087411275e-06,
"loss": 0.2349,
"step": 38620
},
{
"epoch": 4.503847947761194,
"grad_norm": 0.5417274684958774,
"learning_rate": 6.200587996956478e-06,
"loss": 0.2674,
"step": 38625
},
{
"epoch": 4.5044309701492535,
"grad_norm": 0.5170128925942264,
"learning_rate": 6.197793352190316e-06,
"loss": 0.2652,
"step": 38630
},
{
"epoch": 4.505013992537314,
"grad_norm": 0.5346000401229939,
"learning_rate": 6.1950018748581865e-06,
"loss": 0.2597,
"step": 38635
},
{
"epoch": 4.505597014925373,
"grad_norm": 0.5317872923214294,
"learning_rate": 6.192213565375147e-06,
"loss": 0.2641,
"step": 38640
},
{
"epoch": 4.506180037313433,
"grad_norm": 0.5248799865586731,
"learning_rate": 6.189428424155801e-06,
"loss": 0.2574,
"step": 38645
},
{
"epoch": 4.506763059701493,
"grad_norm": 0.5159466862846975,
"learning_rate": 6.186646451614265e-06,
"loss": 0.256,
"step": 38650
},
{
"epoch": 4.507346082089552,
"grad_norm": 0.5355305657805994,
"learning_rate": 6.1838676481641945e-06,
"loss": 0.26,
"step": 38655
},
{
"epoch": 4.507929104477612,
"grad_norm": 0.5311840720051096,
"learning_rate": 6.1810920142187726e-06,
"loss": 0.2661,
"step": 38660
},
{
"epoch": 4.508512126865671,
"grad_norm": 0.5523166105701923,
"learning_rate": 6.178319550190705e-06,
"loss": 0.2805,
"step": 38665
},
{
"epoch": 4.5090951492537314,
"grad_norm": 0.5460956837272631,
"learning_rate": 6.175550256492235e-06,
"loss": 0.2661,
"step": 38670
},
{
"epoch": 4.5096781716417915,
"grad_norm": 0.5221584225934912,
"learning_rate": 6.172784133535133e-06,
"loss": 0.2671,
"step": 38675
},
{
"epoch": 4.510261194029851,
"grad_norm": 0.5289324027768694,
"learning_rate": 6.170021181730681e-06,
"loss": 0.2608,
"step": 38680
},
{
"epoch": 4.510844216417911,
"grad_norm": 0.5025094205111148,
"learning_rate": 6.167261401489721e-06,
"loss": 0.2634,
"step": 38685
},
{
"epoch": 4.51142723880597,
"grad_norm": 0.5486400780433371,
"learning_rate": 6.164504793222589e-06,
"loss": 0.2594,
"step": 38690
},
{
"epoch": 4.51201026119403,
"grad_norm": 0.5323475647368213,
"learning_rate": 6.161751357339184e-06,
"loss": 0.2835,
"step": 38695
},
{
"epoch": 4.512593283582089,
"grad_norm": 0.5173297707671094,
"learning_rate": 6.159001094248904e-06,
"loss": 0.2598,
"step": 38700
},
{
"epoch": 4.513176305970149,
"grad_norm": 0.514495731933102,
"learning_rate": 6.156254004360687e-06,
"loss": 0.2523,
"step": 38705
},
{
"epoch": 4.5137593283582085,
"grad_norm": 0.49633841240966786,
"learning_rate": 6.153510088083e-06,
"loss": 0.2605,
"step": 38710
},
{
"epoch": 4.5143423507462686,
"grad_norm": 0.5445902745091213,
"learning_rate": 6.150769345823841e-06,
"loss": 0.2502,
"step": 38715
},
{
"epoch": 4.514925373134329,
"grad_norm": 0.5397016488949635,
"learning_rate": 6.1480317779907285e-06,
"loss": 0.2627,
"step": 38720
},
{
"epoch": 4.515508395522388,
"grad_norm": 0.552039542472079,
"learning_rate": 6.145297384990715e-06,
"loss": 0.2437,
"step": 38725
},
{
"epoch": 4.516091417910448,
"grad_norm": 0.5078529363100549,
"learning_rate": 6.1425661672303735e-06,
"loss": 0.2622,
"step": 38730
},
{
"epoch": 4.516674440298507,
"grad_norm": 0.5381155791170246,
"learning_rate": 6.139838125115818e-06,
"loss": 0.2522,
"step": 38735
},
{
"epoch": 4.517257462686567,
"grad_norm": 0.5464931315692569,
"learning_rate": 6.1371132590526744e-06,
"loss": 0.2599,
"step": 38740
},
{
"epoch": 4.517840485074627,
"grad_norm": 0.49781898414356895,
"learning_rate": 6.134391569446108e-06,
"loss": 0.2629,
"step": 38745
},
{
"epoch": 4.518423507462686,
"grad_norm": 0.5566657570479573,
"learning_rate": 6.1316730567008086e-06,
"loss": 0.2667,
"step": 38750
},
{
"epoch": 4.5190065298507465,
"grad_norm": 0.5569678812886943,
"learning_rate": 6.12895772122099e-06,
"loss": 0.2726,
"step": 38755
},
{
"epoch": 4.519589552238806,
"grad_norm": 0.4969833583679429,
"learning_rate": 6.126245563410399e-06,
"loss": 0.2599,
"step": 38760
},
{
"epoch": 4.520172574626866,
"grad_norm": 0.4764080174611328,
"learning_rate": 6.1235365836723054e-06,
"loss": 0.2675,
"step": 38765
},
{
"epoch": 4.520755597014926,
"grad_norm": 0.5149988727702287,
"learning_rate": 6.120830782409515e-06,
"loss": 0.257,
"step": 38770
},
{
"epoch": 4.521338619402985,
"grad_norm": 0.5156047545093498,
"learning_rate": 6.118128160024346e-06,
"loss": 0.2623,
"step": 38775
},
{
"epoch": 4.521921641791045,
"grad_norm": 0.543937669874722,
"learning_rate": 6.115428716918657e-06,
"loss": 0.2604,
"step": 38780
},
{
"epoch": 4.522504664179104,
"grad_norm": 0.5511882106781363,
"learning_rate": 6.112732453493826e-06,
"loss": 0.2576,
"step": 38785
},
{
"epoch": 4.523087686567164,
"grad_norm": 0.5425246178117245,
"learning_rate": 6.110039370150765e-06,
"loss": 0.2508,
"step": 38790
},
{
"epoch": 4.5236707089552235,
"grad_norm": 0.5376242323411334,
"learning_rate": 6.107349467289907e-06,
"loss": 0.2485,
"step": 38795
},
{
"epoch": 4.524253731343284,
"grad_norm": 0.5620467692409846,
"learning_rate": 6.104662745311222e-06,
"loss": 0.2771,
"step": 38800
},
{
"epoch": 4.524836753731344,
"grad_norm": 0.5267031593834537,
"learning_rate": 6.1019792046141875e-06,
"loss": 0.2734,
"step": 38805
},
{
"epoch": 4.525419776119403,
"grad_norm": 0.5384371673164725,
"learning_rate": 6.099298845597832e-06,
"loss": 0.2593,
"step": 38810
},
{
"epoch": 4.526002798507463,
"grad_norm": 0.5125346701370309,
"learning_rate": 6.096621668660686e-06,
"loss": 0.267,
"step": 38815
},
{
"epoch": 4.526585820895522,
"grad_norm": 0.524323265736002,
"learning_rate": 6.093947674200838e-06,
"loss": 0.2678,
"step": 38820
},
{
"epoch": 4.527168843283582,
"grad_norm": 0.5278763593059844,
"learning_rate": 6.09127686261587e-06,
"loss": 0.246,
"step": 38825
},
{
"epoch": 4.527751865671641,
"grad_norm": 0.5702414796266169,
"learning_rate": 6.088609234302912e-06,
"loss": 0.2645,
"step": 38830
},
{
"epoch": 4.528334888059701,
"grad_norm": 0.5289656081427371,
"learning_rate": 6.085944789658615e-06,
"loss": 0.2526,
"step": 38835
},
{
"epoch": 4.5289179104477615,
"grad_norm": 0.5157784900885519,
"learning_rate": 6.083283529079157e-06,
"loss": 0.26,
"step": 38840
},
{
"epoch": 4.529500932835821,
"grad_norm": 0.5782867595801789,
"learning_rate": 6.08062545296024e-06,
"loss": 0.2778,
"step": 38845
},
{
"epoch": 4.530083955223881,
"grad_norm": 0.47423350452700835,
"learning_rate": 6.077970561697095e-06,
"loss": 0.2687,
"step": 38850
},
{
"epoch": 4.53066697761194,
"grad_norm": 0.5199150777035527,
"learning_rate": 6.075318855684477e-06,
"loss": 0.2562,
"step": 38855
},
{
"epoch": 4.53125,
"grad_norm": 0.5399124222227776,
"learning_rate": 6.072670335316676e-06,
"loss": 0.2578,
"step": 38860
},
{
"epoch": 4.53183302238806,
"grad_norm": 0.5150499155954146,
"learning_rate": 6.070025000987492e-06,
"loss": 0.2606,
"step": 38865
},
{
"epoch": 4.532416044776119,
"grad_norm": 0.5004514576096859,
"learning_rate": 6.067382853090269e-06,
"loss": 0.2607,
"step": 38870
},
{
"epoch": 4.532999067164179,
"grad_norm": 0.5304019536781089,
"learning_rate": 6.064743892017864e-06,
"loss": 0.2638,
"step": 38875
},
{
"epoch": 4.5335820895522385,
"grad_norm": 0.5084621632745274,
"learning_rate": 6.062108118162669e-06,
"loss": 0.2485,
"step": 38880
},
{
"epoch": 4.534165111940299,
"grad_norm": 0.5108568675959437,
"learning_rate": 6.059475531916595e-06,
"loss": 0.2452,
"step": 38885
},
{
"epoch": 4.534748134328359,
"grad_norm": 0.6152099512145118,
"learning_rate": 6.056846133671083e-06,
"loss": 0.2623,
"step": 38890
},
{
"epoch": 4.535331156716418,
"grad_norm": 0.4994264834604792,
"learning_rate": 6.054219923817101e-06,
"loss": 0.2628,
"step": 38895
},
{
"epoch": 4.535914179104478,
"grad_norm": 0.5113986800087381,
"learning_rate": 6.051596902745143e-06,
"loss": 0.2576,
"step": 38900
},
{
"epoch": 4.536497201492537,
"grad_norm": 0.5001538741025344,
"learning_rate": 6.048977070845219e-06,
"loss": 0.2488,
"step": 38905
},
{
"epoch": 4.537080223880597,
"grad_norm": 0.47927372428680814,
"learning_rate": 6.0463604285068834e-06,
"loss": 0.2513,
"step": 38910
},
{
"epoch": 4.537663246268656,
"grad_norm": 0.529013659568405,
"learning_rate": 6.043746976119201e-06,
"loss": 0.2773,
"step": 38915
},
{
"epoch": 4.538246268656716,
"grad_norm": 0.4906537140004885,
"learning_rate": 6.0411367140707625e-06,
"loss": 0.2416,
"step": 38920
},
{
"epoch": 4.5388292910447765,
"grad_norm": 0.5059263225489081,
"learning_rate": 6.038529642749697e-06,
"loss": 0.2428,
"step": 38925
},
{
"epoch": 4.539412313432836,
"grad_norm": 0.523676580280404,
"learning_rate": 6.035925762543644e-06,
"loss": 0.2595,
"step": 38930
},
{
"epoch": 4.539995335820896,
"grad_norm": 0.4983571917110067,
"learning_rate": 6.03332507383978e-06,
"loss": 0.2505,
"step": 38935
},
{
"epoch": 4.540578358208955,
"grad_norm": 0.5118483537725239,
"learning_rate": 6.030727577024802e-06,
"loss": 0.2593,
"step": 38940
},
{
"epoch": 4.541161380597015,
"grad_norm": 0.546423329616185,
"learning_rate": 6.028133272484936e-06,
"loss": 0.2727,
"step": 38945
},
{
"epoch": 4.541744402985074,
"grad_norm": 0.4903725648707753,
"learning_rate": 6.025542160605923e-06,
"loss": 0.2436,
"step": 38950
},
{
"epoch": 4.542327425373134,
"grad_norm": 0.5455420134323096,
"learning_rate": 6.022954241773038e-06,
"loss": 0.2761,
"step": 38955
},
{
"epoch": 4.542910447761194,
"grad_norm": 0.4585826547870705,
"learning_rate": 6.020369516371085e-06,
"loss": 0.2668,
"step": 38960
},
{
"epoch": 4.5434934701492535,
"grad_norm": 0.5265273982628189,
"learning_rate": 6.017787984784381e-06,
"loss": 0.2746,
"step": 38965
},
{
"epoch": 4.544076492537314,
"grad_norm": 0.5020942041450563,
"learning_rate": 6.015209647396781e-06,
"loss": 0.27,
"step": 38970
},
{
"epoch": 4.544659514925373,
"grad_norm": 0.5044068757198353,
"learning_rate": 6.012634504591658e-06,
"loss": 0.2688,
"step": 38975
},
{
"epoch": 4.545242537313433,
"grad_norm": 0.5721697002713411,
"learning_rate": 6.010062556751906e-06,
"loss": 0.262,
"step": 38980
},
{
"epoch": 4.545825559701493,
"grad_norm": 0.4880644698432569,
"learning_rate": 6.0074938042599574e-06,
"loss": 0.2422,
"step": 38985
},
{
"epoch": 4.546408582089552,
"grad_norm": 0.5263997706877247,
"learning_rate": 6.00492824749775e-06,
"loss": 0.271,
"step": 38990
},
{
"epoch": 4.546991604477612,
"grad_norm": 0.5517625766551889,
"learning_rate": 6.00236588684677e-06,
"loss": 0.2658,
"step": 38995
},
{
"epoch": 4.547574626865671,
"grad_norm": 0.49354683745418637,
"learning_rate": 5.999806722688007e-06,
"loss": 0.2471,
"step": 39000
},
{
"epoch": 4.5481576492537314,
"grad_norm": 0.5178997733040462,
"learning_rate": 5.9972507554019895e-06,
"loss": 0.2605,
"step": 39005
},
{
"epoch": 4.5487406716417915,
"grad_norm": 0.5001909529200346,
"learning_rate": 5.994697985368761e-06,
"loss": 0.2638,
"step": 39010
},
{
"epoch": 4.549323694029851,
"grad_norm": 0.5364586643874559,
"learning_rate": 5.992148412967895e-06,
"loss": 0.2609,
"step": 39015
},
{
"epoch": 4.549906716417911,
"grad_norm": 0.5668501260028482,
"learning_rate": 5.98960203857849e-06,
"loss": 0.2613,
"step": 39020
},
{
"epoch": 4.55048973880597,
"grad_norm": 1.7483447793114588,
"learning_rate": 5.987058862579167e-06,
"loss": 0.2594,
"step": 39025
},
{
"epoch": 4.55107276119403,
"grad_norm": 0.537973349519233,
"learning_rate": 5.98451888534807e-06,
"loss": 0.2804,
"step": 39030
},
{
"epoch": 4.551655783582089,
"grad_norm": 0.5651328937551121,
"learning_rate": 5.981982107262877e-06,
"loss": 0.2657,
"step": 39035
},
{
"epoch": 4.552238805970149,
"grad_norm": 0.5165974934499695,
"learning_rate": 5.9794485287007696e-06,
"loss": 0.2512,
"step": 39040
},
{
"epoch": 4.5528218283582085,
"grad_norm": 0.5823602307496405,
"learning_rate": 5.976918150038478e-06,
"loss": 0.2584,
"step": 39045
},
{
"epoch": 4.5534048507462686,
"grad_norm": 0.628072248282222,
"learning_rate": 5.974390971652237e-06,
"loss": 0.2774,
"step": 39050
},
{
"epoch": 4.553987873134329,
"grad_norm": 0.5251264104114984,
"learning_rate": 5.971866993917821e-06,
"loss": 0.2654,
"step": 39055
},
{
"epoch": 4.554570895522388,
"grad_norm": 0.5521170164983134,
"learning_rate": 5.9693462172105165e-06,
"loss": 0.2692,
"step": 39060
},
{
"epoch": 4.555153917910448,
"grad_norm": 0.4998549724119816,
"learning_rate": 5.966828641905142e-06,
"loss": 0.2438,
"step": 39065
},
{
"epoch": 4.555736940298507,
"grad_norm": 0.4897087808573781,
"learning_rate": 5.964314268376031e-06,
"loss": 0.2653,
"step": 39070
},
{
"epoch": 4.556319962686567,
"grad_norm": 0.5317645488175172,
"learning_rate": 5.961803096997056e-06,
"loss": 0.2691,
"step": 39075
},
{
"epoch": 4.556902985074627,
"grad_norm": 0.5072629913145804,
"learning_rate": 5.959295128141596e-06,
"loss": 0.2555,
"step": 39080
},
{
"epoch": 4.557486007462686,
"grad_norm": 0.5194778596385488,
"learning_rate": 5.956790362182567e-06,
"loss": 0.2648,
"step": 39085
},
{
"epoch": 4.5580690298507465,
"grad_norm": 0.4896791599803869,
"learning_rate": 5.9542887994923985e-06,
"loss": 0.2449,
"step": 39090
},
{
"epoch": 4.558652052238806,
"grad_norm": 0.518112465898371,
"learning_rate": 5.951790440443055e-06,
"loss": 0.2812,
"step": 39095
},
{
"epoch": 4.559235074626866,
"grad_norm": 0.5467229113899952,
"learning_rate": 5.949295285406015e-06,
"loss": 0.2664,
"step": 39100
},
{
"epoch": 4.559818097014926,
"grad_norm": 0.5920242852575323,
"learning_rate": 5.946803334752285e-06,
"loss": 0.2805,
"step": 39105
},
{
"epoch": 4.560401119402985,
"grad_norm": 0.5489787113365736,
"learning_rate": 5.944314588852393e-06,
"loss": 0.261,
"step": 39110
},
{
"epoch": 4.560984141791045,
"grad_norm": 0.5148234612456437,
"learning_rate": 5.941829048076392e-06,
"loss": 0.2573,
"step": 39115
},
{
"epoch": 4.561567164179104,
"grad_norm": 0.5057751472170834,
"learning_rate": 5.93934671279386e-06,
"loss": 0.258,
"step": 39120
},
{
"epoch": 4.562150186567164,
"grad_norm": 0.5035074594308664,
"learning_rate": 5.936867583373895e-06,
"loss": 0.263,
"step": 39125
},
{
"epoch": 4.5627332089552235,
"grad_norm": 0.5165259452147263,
"learning_rate": 5.934391660185121e-06,
"loss": 0.2554,
"step": 39130
},
{
"epoch": 4.563316231343284,
"grad_norm": 0.5311455792598649,
"learning_rate": 5.931918943595682e-06,
"loss": 0.2353,
"step": 39135
},
{
"epoch": 4.563899253731344,
"grad_norm": 0.5718147441246065,
"learning_rate": 5.929449433973249e-06,
"loss": 0.2623,
"step": 39140
},
{
"epoch": 4.564482276119403,
"grad_norm": 0.5146604972963873,
"learning_rate": 5.926983131685012e-06,
"loss": 0.2526,
"step": 39145
},
{
"epoch": 4.565065298507463,
"grad_norm": 0.5789918241569233,
"learning_rate": 5.924520037097688e-06,
"loss": 0.2751,
"step": 39150
},
{
"epoch": 4.565648320895522,
"grad_norm": 0.5570935922664527,
"learning_rate": 5.922060150577517e-06,
"loss": 0.2723,
"step": 39155
},
{
"epoch": 4.566231343283582,
"grad_norm": 0.48832190251301205,
"learning_rate": 5.919603472490263e-06,
"loss": 0.2858,
"step": 39160
},
{
"epoch": 4.566814365671641,
"grad_norm": 0.5002046914943303,
"learning_rate": 5.917150003201201e-06,
"loss": 0.2345,
"step": 39165
},
{
"epoch": 4.567397388059701,
"grad_norm": 0.5286197129348891,
"learning_rate": 5.914699743075149e-06,
"loss": 0.2592,
"step": 39170
},
{
"epoch": 4.5679804104477615,
"grad_norm": 0.5092297450544302,
"learning_rate": 5.9122526924764264e-06,
"loss": 0.2483,
"step": 39175
},
{
"epoch": 4.568563432835821,
"grad_norm": 0.5739015669452459,
"learning_rate": 5.909808851768898e-06,
"loss": 0.269,
"step": 39180
},
{
"epoch": 4.569146455223881,
"grad_norm": 0.5731974873144853,
"learning_rate": 5.9073682213159325e-06,
"loss": 0.2503,
"step": 39185
},
{
"epoch": 4.56972947761194,
"grad_norm": 0.5813716778069474,
"learning_rate": 5.904930801480427e-06,
"loss": 0.2734,
"step": 39190
},
{
"epoch": 4.5703125,
"grad_norm": 0.5598980422996925,
"learning_rate": 5.902496592624808e-06,
"loss": 0.2659,
"step": 39195
},
{
"epoch": 4.57089552238806,
"grad_norm": 0.5153256785923885,
"learning_rate": 5.900065595111014e-06,
"loss": 0.2523,
"step": 39200
},
{
"epoch": 4.571478544776119,
"grad_norm": 0.5204926567812194,
"learning_rate": 5.897637809300514e-06,
"loss": 0.2657,
"step": 39205
},
{
"epoch": 4.572061567164179,
"grad_norm": 0.5635535313547579,
"learning_rate": 5.895213235554298e-06,
"loss": 0.2768,
"step": 39210
},
{
"epoch": 4.5726445895522385,
"grad_norm": 0.5049764843459283,
"learning_rate": 5.892791874232868e-06,
"loss": 0.2458,
"step": 39215
},
{
"epoch": 4.573227611940299,
"grad_norm": 0.5021265636482644,
"learning_rate": 5.890373725696271e-06,
"loss": 0.2525,
"step": 39220
},
{
"epoch": 4.573810634328359,
"grad_norm": 0.5211571852581737,
"learning_rate": 5.887958790304047e-06,
"loss": 0.2618,
"step": 39225
},
{
"epoch": 4.574393656716418,
"grad_norm": 0.5501783296978301,
"learning_rate": 5.885547068415289e-06,
"loss": 0.2711,
"step": 39230
},
{
"epoch": 4.574976679104478,
"grad_norm": 0.47179845819744737,
"learning_rate": 5.883138560388587e-06,
"loss": 0.2447,
"step": 39235
},
{
"epoch": 4.575559701492537,
"grad_norm": 0.5209049140406725,
"learning_rate": 5.880733266582066e-06,
"loss": 0.2702,
"step": 39240
},
{
"epoch": 4.576142723880597,
"grad_norm": 0.5230158490968697,
"learning_rate": 5.878331187353371e-06,
"loss": 0.2569,
"step": 39245
},
{
"epoch": 4.576725746268656,
"grad_norm": 0.5148379583291693,
"learning_rate": 5.875932323059667e-06,
"loss": 0.2649,
"step": 39250
},
{
"epoch": 4.577308768656716,
"grad_norm": 0.5432816101988381,
"learning_rate": 5.87353667405764e-06,
"loss": 0.2542,
"step": 39255
},
{
"epoch": 4.5778917910447765,
"grad_norm": 0.5310160749752704,
"learning_rate": 5.871144240703507e-06,
"loss": 0.2482,
"step": 39260
},
{
"epoch": 4.578474813432836,
"grad_norm": 0.5192906911805769,
"learning_rate": 5.86875502335299e-06,
"loss": 0.2531,
"step": 39265
},
{
"epoch": 4.579057835820896,
"grad_norm": 0.5434345680182487,
"learning_rate": 5.866369022361354e-06,
"loss": 0.2462,
"step": 39270
},
{
"epoch": 4.579640858208955,
"grad_norm": 0.47507817563686167,
"learning_rate": 5.863986238083367e-06,
"loss": 0.2735,
"step": 39275
},
{
"epoch": 4.580223880597015,
"grad_norm": 0.5349078706462872,
"learning_rate": 5.8616066708733255e-06,
"loss": 0.2523,
"step": 39280
},
{
"epoch": 4.580806902985074,
"grad_norm": 0.47091628432142624,
"learning_rate": 5.859230321085049e-06,
"loss": 0.2529,
"step": 39285
},
{
"epoch": 4.581389925373134,
"grad_norm": 0.528743361178868,
"learning_rate": 5.856857189071884e-06,
"loss": 0.2704,
"step": 39290
},
{
"epoch": 4.581972947761194,
"grad_norm": 0.4990121507064239,
"learning_rate": 5.8544872751866845e-06,
"loss": 0.2472,
"step": 39295
},
{
"epoch": 4.5825559701492535,
"grad_norm": 0.5645313197773756,
"learning_rate": 5.852120579781838e-06,
"loss": 0.2839,
"step": 39300
},
{
"epoch": 4.583138992537314,
"grad_norm": 0.5226436084754454,
"learning_rate": 5.849757103209252e-06,
"loss": 0.2734,
"step": 39305
},
{
"epoch": 4.583722014925373,
"grad_norm": 0.5463664391906513,
"learning_rate": 5.847396845820349e-06,
"loss": 0.2855,
"step": 39310
},
{
"epoch": 4.584305037313433,
"grad_norm": 0.48099862258521753,
"learning_rate": 5.845039807966074e-06,
"loss": 0.253,
"step": 39315
},
{
"epoch": 4.584888059701493,
"grad_norm": 0.5409449082799026,
"learning_rate": 5.8426859899969034e-06,
"loss": 0.2741,
"step": 39320
},
{
"epoch": 4.585471082089552,
"grad_norm": 0.5182304216740492,
"learning_rate": 5.84033539226282e-06,
"loss": 0.2624,
"step": 39325
},
{
"epoch": 4.586054104477612,
"grad_norm": 0.5048264798103715,
"learning_rate": 5.83798801511334e-06,
"loss": 0.2601,
"step": 39330
},
{
"epoch": 4.586637126865671,
"grad_norm": 0.5511959412336976,
"learning_rate": 5.835643858897498e-06,
"loss": 0.263,
"step": 39335
},
{
"epoch": 4.5872201492537314,
"grad_norm": 0.5006116307130094,
"learning_rate": 5.833302923963837e-06,
"loss": 0.2699,
"step": 39340
},
{
"epoch": 4.5878031716417915,
"grad_norm": 0.5539465121332253,
"learning_rate": 5.830965210660445e-06,
"loss": 0.2727,
"step": 39345
},
{
"epoch": 4.588386194029851,
"grad_norm": 0.527718508349746,
"learning_rate": 5.828630719334905e-06,
"loss": 0.2554,
"step": 39350
},
{
"epoch": 4.588969216417911,
"grad_norm": 0.5229234931270784,
"learning_rate": 5.826299450334345e-06,
"loss": 0.2783,
"step": 39355
},
{
"epoch": 4.58955223880597,
"grad_norm": 0.5365676057638463,
"learning_rate": 5.8239714040053936e-06,
"loss": 0.2715,
"step": 39360
},
{
"epoch": 4.59013526119403,
"grad_norm": 0.5171971752161777,
"learning_rate": 5.821646580694214e-06,
"loss": 0.2675,
"step": 39365
},
{
"epoch": 4.590718283582089,
"grad_norm": 0.48569033045468135,
"learning_rate": 5.819324980746483e-06,
"loss": 0.2552,
"step": 39370
},
{
"epoch": 4.591301305970149,
"grad_norm": 0.5151000453096796,
"learning_rate": 5.817006604507401e-06,
"loss": 0.2567,
"step": 39375
},
{
"epoch": 4.5918843283582085,
"grad_norm": 0.49170361831576925,
"learning_rate": 5.814691452321687e-06,
"loss": 0.244,
"step": 39380
},
{
"epoch": 4.5924673507462686,
"grad_norm": 0.5706880060010474,
"learning_rate": 5.812379524533587e-06,
"loss": 0.2733,
"step": 39385
},
{
"epoch": 4.593050373134329,
"grad_norm": 0.5273620636702008,
"learning_rate": 5.810070821486854e-06,
"loss": 0.2607,
"step": 39390
},
{
"epoch": 4.593633395522388,
"grad_norm": 0.5088253491924825,
"learning_rate": 5.8077653435247774e-06,
"loss": 0.2708,
"step": 39395
},
{
"epoch": 4.594216417910448,
"grad_norm": 0.5192738221976703,
"learning_rate": 5.805463090990154e-06,
"loss": 0.2505,
"step": 39400
},
{
"epoch": 4.594799440298507,
"grad_norm": 0.5848398046537732,
"learning_rate": 5.803164064225313e-06,
"loss": 0.2686,
"step": 39405
},
{
"epoch": 4.595382462686567,
"grad_norm": 0.5026316392413409,
"learning_rate": 5.800868263572093e-06,
"loss": 0.2508,
"step": 39410
},
{
"epoch": 4.595965485074627,
"grad_norm": 0.5073754432419348,
"learning_rate": 5.7985756893718585e-06,
"loss": 0.2525,
"step": 39415
},
{
"epoch": 4.596548507462686,
"grad_norm": 0.4958416479262377,
"learning_rate": 5.796286341965492e-06,
"loss": 0.245,
"step": 39420
},
{
"epoch": 4.5971315298507465,
"grad_norm": 0.49618230305373573,
"learning_rate": 5.794000221693403e-06,
"loss": 0.257,
"step": 39425
},
{
"epoch": 4.597714552238806,
"grad_norm": 0.5003617310823768,
"learning_rate": 5.7917173288955105e-06,
"loss": 0.2569,
"step": 39430
},
{
"epoch": 4.598297574626866,
"grad_norm": 0.5744962967027848,
"learning_rate": 5.789437663911261e-06,
"loss": 0.2717,
"step": 39435
},
{
"epoch": 4.598880597014926,
"grad_norm": 0.5351010564525503,
"learning_rate": 5.787161227079613e-06,
"loss": 0.2697,
"step": 39440
},
{
"epoch": 4.599463619402985,
"grad_norm": 0.5091439653815344,
"learning_rate": 5.7848880187390615e-06,
"loss": 0.2643,
"step": 39445
},
{
"epoch": 4.600046641791045,
"grad_norm": 0.519037907834975,
"learning_rate": 5.782618039227603e-06,
"loss": 0.2481,
"step": 39450
},
{
"epoch": 4.600629664179104,
"grad_norm": 0.5497073186414385,
"learning_rate": 5.7803512888827626e-06,
"loss": 0.258,
"step": 39455
},
{
"epoch": 4.601212686567164,
"grad_norm": 0.56716755693992,
"learning_rate": 5.778087768041589e-06,
"loss": 0.2768,
"step": 39460
},
{
"epoch": 4.6017957089552235,
"grad_norm": 0.5217926690145527,
"learning_rate": 5.7758274770406375e-06,
"loss": 0.2476,
"step": 39465
},
{
"epoch": 4.602378731343284,
"grad_norm": 0.4986334629889643,
"learning_rate": 5.7735704162160005e-06,
"loss": 0.2693,
"step": 39470
},
{
"epoch": 4.602961753731344,
"grad_norm": 0.4957356408163735,
"learning_rate": 5.771316585903276e-06,
"loss": 0.2819,
"step": 39475
},
{
"epoch": 4.603544776119403,
"grad_norm": 0.5330093711736085,
"learning_rate": 5.769065986437591e-06,
"loss": 0.2672,
"step": 39480
},
{
"epoch": 4.604127798507463,
"grad_norm": 0.5645434647863821,
"learning_rate": 5.766818618153584e-06,
"loss": 0.2551,
"step": 39485
},
{
"epoch": 4.604710820895522,
"grad_norm": 0.4946595097058112,
"learning_rate": 5.764574481385419e-06,
"loss": 0.2454,
"step": 39490
},
{
"epoch": 4.605293843283582,
"grad_norm": 0.5224712282544505,
"learning_rate": 5.762333576466778e-06,
"loss": 0.2513,
"step": 39495
},
{
"epoch": 4.605876865671641,
"grad_norm": 0.563427600903738,
"learning_rate": 5.7600959037308626e-06,
"loss": 0.2733,
"step": 39500
},
{
"epoch": 4.606459888059701,
"grad_norm": 0.5579432122104788,
"learning_rate": 5.75786146351039e-06,
"loss": 0.2662,
"step": 39505
},
{
"epoch": 4.6070429104477615,
"grad_norm": 0.5590879807675503,
"learning_rate": 5.755630256137605e-06,
"loss": 0.2775,
"step": 39510
},
{
"epoch": 4.607625932835821,
"grad_norm": 0.4880211162127533,
"learning_rate": 5.753402281944261e-06,
"loss": 0.2511,
"step": 39515
},
{
"epoch": 4.608208955223881,
"grad_norm": 0.5488164496375905,
"learning_rate": 5.7511775412616415e-06,
"loss": 0.2558,
"step": 39520
},
{
"epoch": 4.60879197761194,
"grad_norm": 0.5506795924858428,
"learning_rate": 5.748956034420539e-06,
"loss": 0.2496,
"step": 39525
},
{
"epoch": 4.609375,
"grad_norm": 0.5605310164010058,
"learning_rate": 5.74673776175128e-06,
"loss": 0.2593,
"step": 39530
},
{
"epoch": 4.60995802238806,
"grad_norm": 0.5033277726183638,
"learning_rate": 5.744522723583689e-06,
"loss": 0.2426,
"step": 39535
},
{
"epoch": 4.610541044776119,
"grad_norm": 0.5027357838767988,
"learning_rate": 5.742310920247127e-06,
"loss": 0.2507,
"step": 39540
},
{
"epoch": 4.611124067164179,
"grad_norm": 0.5567561304167239,
"learning_rate": 5.740102352070463e-06,
"loss": 0.2844,
"step": 39545
},
{
"epoch": 4.6117070895522385,
"grad_norm": 0.564171089492422,
"learning_rate": 5.737897019382098e-06,
"loss": 0.2635,
"step": 39550
},
{
"epoch": 4.612290111940299,
"grad_norm": 0.5233508193926742,
"learning_rate": 5.735694922509938e-06,
"loss": 0.2519,
"step": 39555
},
{
"epoch": 4.612873134328359,
"grad_norm": 0.5339753828795404,
"learning_rate": 5.733496061781418e-06,
"loss": 0.2648,
"step": 39560
},
{
"epoch": 4.613456156716418,
"grad_norm": 0.48995155329985074,
"learning_rate": 5.73130043752348e-06,
"loss": 0.246,
"step": 39565
},
{
"epoch": 4.614039179104478,
"grad_norm": 0.5177319985740271,
"learning_rate": 5.729108050062603e-06,
"loss": 0.2659,
"step": 39570
},
{
"epoch": 4.614622201492537,
"grad_norm": 0.5445035797507868,
"learning_rate": 5.726918899724759e-06,
"loss": 0.2629,
"step": 39575
},
{
"epoch": 4.615205223880597,
"grad_norm": 0.49770199544713895,
"learning_rate": 5.7247329868354705e-06,
"loss": 0.2587,
"step": 39580
},
{
"epoch": 4.615788246268656,
"grad_norm": 0.5026847329904678,
"learning_rate": 5.722550311719753e-06,
"loss": 0.2548,
"step": 39585
},
{
"epoch": 4.616371268656716,
"grad_norm": 0.5446304167463728,
"learning_rate": 5.720370874702148e-06,
"loss": 0.2852,
"step": 39590
},
{
"epoch": 4.6169542910447765,
"grad_norm": 0.5383502217358929,
"learning_rate": 5.7181946761067205e-06,
"loss": 0.2648,
"step": 39595
},
{
"epoch": 4.617537313432836,
"grad_norm": 0.5350750319413246,
"learning_rate": 5.716021716257047e-06,
"loss": 0.2593,
"step": 39600
},
{
"epoch": 4.618120335820896,
"grad_norm": 0.5112603395958923,
"learning_rate": 5.71385199547623e-06,
"loss": 0.2599,
"step": 39605
},
{
"epoch": 4.618703358208955,
"grad_norm": 0.5400977023317569,
"learning_rate": 5.7116855140868874e-06,
"loss": 0.243,
"step": 39610
},
{
"epoch": 4.619286380597015,
"grad_norm": 0.600609893489187,
"learning_rate": 5.709522272411145e-06,
"loss": 0.2611,
"step": 39615
},
{
"epoch": 4.619869402985074,
"grad_norm": 0.5343060392091388,
"learning_rate": 5.707362270770665e-06,
"loss": 0.2549,
"step": 39620
},
{
"epoch": 4.620452425373134,
"grad_norm": 0.5099039748144819,
"learning_rate": 5.705205509486613e-06,
"loss": 0.2557,
"step": 39625
},
{
"epoch": 4.621035447761194,
"grad_norm": 0.5196756335227851,
"learning_rate": 5.703051988879689e-06,
"loss": 0.2509,
"step": 39630
},
{
"epoch": 4.6216184701492535,
"grad_norm": 0.5176297657516744,
"learning_rate": 5.700901709270088e-06,
"loss": 0.2606,
"step": 39635
},
{
"epoch": 4.622201492537314,
"grad_norm": 0.48431253013961323,
"learning_rate": 5.698754670977544e-06,
"loss": 0.2455,
"step": 39640
},
{
"epoch": 4.622784514925373,
"grad_norm": 0.5279944256153754,
"learning_rate": 5.696610874321296e-06,
"loss": 0.2685,
"step": 39645
},
{
"epoch": 4.623367537313433,
"grad_norm": 0.7043291715170641,
"learning_rate": 5.69447031962011e-06,
"loss": 0.2755,
"step": 39650
},
{
"epoch": 4.623950559701493,
"grad_norm": 0.5241944783645631,
"learning_rate": 5.6923330071922634e-06,
"loss": 0.2603,
"step": 39655
},
{
"epoch": 4.624533582089552,
"grad_norm": 0.4551091243035712,
"learning_rate": 5.690198937355561e-06,
"loss": 0.2653,
"step": 39660
},
{
"epoch": 4.625116604477612,
"grad_norm": 0.54382786301182,
"learning_rate": 5.68806811042731e-06,
"loss": 0.2636,
"step": 39665
},
{
"epoch": 4.625699626865671,
"grad_norm": 0.5312020355109183,
"learning_rate": 5.685940526724344e-06,
"loss": 0.2512,
"step": 39670
},
{
"epoch": 4.6262826492537314,
"grad_norm": 0.5030153366075172,
"learning_rate": 5.683816186563018e-06,
"loss": 0.2487,
"step": 39675
},
{
"epoch": 4.6268656716417915,
"grad_norm": 0.5052790696536368,
"learning_rate": 5.6816950902592005e-06,
"loss": 0.2615,
"step": 39680
},
{
"epoch": 4.627448694029851,
"grad_norm": 0.5152226706868136,
"learning_rate": 5.6795772381282785e-06,
"loss": 0.2535,
"step": 39685
},
{
"epoch": 4.628031716417911,
"grad_norm": 0.5340266358428221,
"learning_rate": 5.6774626304851555e-06,
"loss": 0.2528,
"step": 39690
},
{
"epoch": 4.62861473880597,
"grad_norm": 0.4777349890163476,
"learning_rate": 5.675351267644256e-06,
"loss": 0.2506,
"step": 39695
},
{
"epoch": 4.62919776119403,
"grad_norm": 0.5433246293517688,
"learning_rate": 5.673243149919512e-06,
"loss": 0.2574,
"step": 39700
},
{
"epoch": 4.629780783582089,
"grad_norm": 0.553171690141395,
"learning_rate": 5.671138277624391e-06,
"loss": 0.2646,
"step": 39705
},
{
"epoch": 4.630363805970149,
"grad_norm": 0.5170862670736807,
"learning_rate": 5.669036651071857e-06,
"loss": 0.2543,
"step": 39710
},
{
"epoch": 4.6309468283582085,
"grad_norm": 0.5189499541189913,
"learning_rate": 5.66693827057441e-06,
"loss": 0.2725,
"step": 39715
},
{
"epoch": 4.6315298507462686,
"grad_norm": 0.5240036738205557,
"learning_rate": 5.664843136444054e-06,
"loss": 0.2564,
"step": 39720
},
{
"epoch": 4.632112873134329,
"grad_norm": 0.5094303603605583,
"learning_rate": 5.662751248992315e-06,
"loss": 0.249,
"step": 39725
},
{
"epoch": 4.632695895522388,
"grad_norm": 0.5419601046635678,
"learning_rate": 5.660662608530239e-06,
"loss": 0.2534,
"step": 39730
},
{
"epoch": 4.633278917910448,
"grad_norm": 0.47543703416020394,
"learning_rate": 5.658577215368389e-06,
"loss": 0.2496,
"step": 39735
},
{
"epoch": 4.633861940298507,
"grad_norm": 0.5580631260534408,
"learning_rate": 5.6564950698168385e-06,
"loss": 0.2709,
"step": 39740
},
{
"epoch": 4.634444962686567,
"grad_norm": 0.531560243555757,
"learning_rate": 5.654416172185187e-06,
"loss": 0.2619,
"step": 39745
},
{
"epoch": 4.635027985074627,
"grad_norm": 0.5205084469010693,
"learning_rate": 5.652340522782542e-06,
"loss": 0.253,
"step": 39750
},
{
"epoch": 4.635611007462686,
"grad_norm": 0.5214763362082411,
"learning_rate": 5.6502681219175355e-06,
"loss": 0.2612,
"step": 39755
},
{
"epoch": 4.6361940298507465,
"grad_norm": 0.5504813105626072,
"learning_rate": 5.648198969898311e-06,
"loss": 0.2495,
"step": 39760
},
{
"epoch": 4.636777052238806,
"grad_norm": 0.5247833267231861,
"learning_rate": 5.646133067032536e-06,
"loss": 0.2481,
"step": 39765
},
{
"epoch": 4.637360074626866,
"grad_norm": 0.4800118967667409,
"learning_rate": 5.644070413627386e-06,
"loss": 0.2439,
"step": 39770
},
{
"epoch": 4.637943097014926,
"grad_norm": 0.5486091169962901,
"learning_rate": 5.642011009989562e-06,
"loss": 0.2735,
"step": 39775
},
{
"epoch": 4.638526119402985,
"grad_norm": 0.5304177171428159,
"learning_rate": 5.639954856425273e-06,
"loss": 0.2762,
"step": 39780
},
{
"epoch": 4.639109141791045,
"grad_norm": 0.5431082782054679,
"learning_rate": 5.6379019532402554e-06,
"loss": 0.2725,
"step": 39785
},
{
"epoch": 4.639692164179104,
"grad_norm": 0.46956555725103133,
"learning_rate": 5.6358523007397485e-06,
"loss": 0.2535,
"step": 39790
},
{
"epoch": 4.640275186567164,
"grad_norm": 0.576866891327412,
"learning_rate": 5.633805899228524e-06,
"loss": 0.2547,
"step": 39795
},
{
"epoch": 4.6408582089552235,
"grad_norm": 0.5083759259602593,
"learning_rate": 5.631762749010855e-06,
"loss": 0.2382,
"step": 39800
},
{
"epoch": 4.641441231343284,
"grad_norm": 0.582557041488828,
"learning_rate": 5.629722850390544e-06,
"loss": 0.2731,
"step": 39805
},
{
"epoch": 4.642024253731344,
"grad_norm": 0.5333557531674228,
"learning_rate": 5.6276862036709e-06,
"loss": 0.2618,
"step": 39810
},
{
"epoch": 4.642607276119403,
"grad_norm": 0.5261551531608604,
"learning_rate": 5.625652809154753e-06,
"loss": 0.2456,
"step": 39815
},
{
"epoch": 4.643190298507463,
"grad_norm": 0.57089408956915,
"learning_rate": 5.6236226671444555e-06,
"loss": 0.2758,
"step": 39820
},
{
"epoch": 4.643773320895522,
"grad_norm": 0.5279218139839572,
"learning_rate": 5.6215957779418624e-06,
"loss": 0.2547,
"step": 39825
},
{
"epoch": 4.644356343283582,
"grad_norm": 0.47708116054931277,
"learning_rate": 5.619572141848358e-06,
"loss": 0.2551,
"step": 39830
},
{
"epoch": 4.644939365671641,
"grad_norm": 0.4899332252791134,
"learning_rate": 5.617551759164836e-06,
"loss": 0.2452,
"step": 39835
},
{
"epoch": 4.645522388059701,
"grad_norm": 0.5156010775338732,
"learning_rate": 5.615534630191708e-06,
"loss": 0.2583,
"step": 39840
},
{
"epoch": 4.6461054104477615,
"grad_norm": 0.541749571599729,
"learning_rate": 5.613520755228901e-06,
"loss": 0.2509,
"step": 39845
},
{
"epoch": 4.646688432835821,
"grad_norm": 0.5306833646724142,
"learning_rate": 5.611510134575859e-06,
"loss": 0.2629,
"step": 39850
},
{
"epoch": 4.647271455223881,
"grad_norm": 0.5218121839872212,
"learning_rate": 5.609502768531541e-06,
"loss": 0.2563,
"step": 39855
},
{
"epoch": 4.64785447761194,
"grad_norm": 0.5340644359523524,
"learning_rate": 5.607498657394424e-06,
"loss": 0.2782,
"step": 39860
},
{
"epoch": 4.6484375,
"grad_norm": 0.5616557381089768,
"learning_rate": 5.605497801462503e-06,
"loss": 0.2869,
"step": 39865
},
{
"epoch": 4.64902052238806,
"grad_norm": 0.5233009605449933,
"learning_rate": 5.603500201033285e-06,
"loss": 0.2647,
"step": 39870
},
{
"epoch": 4.649603544776119,
"grad_norm": 0.5072572205369043,
"learning_rate": 5.601505856403786e-06,
"loss": 0.2506,
"step": 39875
},
{
"epoch": 4.650186567164179,
"grad_norm": 0.520608341845551,
"learning_rate": 5.59951476787056e-06,
"loss": 0.2534,
"step": 39880
},
{
"epoch": 4.6507695895522385,
"grad_norm": 0.5259777167662049,
"learning_rate": 5.59752693572965e-06,
"loss": 0.2608,
"step": 39885
},
{
"epoch": 4.651352611940299,
"grad_norm": 0.5452055480624008,
"learning_rate": 5.595542360276636e-06,
"loss": 0.2482,
"step": 39890
},
{
"epoch": 4.651935634328359,
"grad_norm": 0.5443251819133219,
"learning_rate": 5.593561041806601e-06,
"loss": 0.2603,
"step": 39895
},
{
"epoch": 4.652518656716418,
"grad_norm": 0.5457749401429179,
"learning_rate": 5.591582980614151e-06,
"loss": 0.253,
"step": 39900
},
{
"epoch": 4.653101679104478,
"grad_norm": 0.4980339071127979,
"learning_rate": 5.589608176993401e-06,
"loss": 0.2617,
"step": 39905
},
{
"epoch": 4.653684701492537,
"grad_norm": 0.5607813456754676,
"learning_rate": 5.587636631237991e-06,
"loss": 0.2512,
"step": 39910
},
{
"epoch": 4.654267723880597,
"grad_norm": 0.5211003881829674,
"learning_rate": 5.585668343641064e-06,
"loss": 0.2416,
"step": 39915
},
{
"epoch": 4.654850746268656,
"grad_norm": 0.5170232868854111,
"learning_rate": 5.583703314495294e-06,
"loss": 0.2628,
"step": 39920
},
{
"epoch": 4.655433768656716,
"grad_norm": 0.5249780373897078,
"learning_rate": 5.58174154409285e-06,
"loss": 0.2637,
"step": 39925
},
{
"epoch": 4.6560167910447765,
"grad_norm": 0.5088062574359778,
"learning_rate": 5.579783032725441e-06,
"loss": 0.2746,
"step": 39930
},
{
"epoch": 4.656599813432836,
"grad_norm": 0.4989325567462243,
"learning_rate": 5.577827780684269e-06,
"loss": 0.2541,
"step": 39935
},
{
"epoch": 4.657182835820896,
"grad_norm": 0.49775649080391965,
"learning_rate": 5.5758757882600706e-06,
"loss": 0.2568,
"step": 39940
},
{
"epoch": 4.657765858208955,
"grad_norm": 0.5303577158898439,
"learning_rate": 5.573927055743082e-06,
"loss": 0.2704,
"step": 39945
},
{
"epoch": 4.658348880597015,
"grad_norm": 0.5266344318925089,
"learning_rate": 5.57198158342306e-06,
"loss": 0.2594,
"step": 39950
},
{
"epoch": 4.658931902985074,
"grad_norm": 0.5586617486799256,
"learning_rate": 5.5700393715892815e-06,
"loss": 0.2585,
"step": 39955
},
{
"epoch": 4.659514925373134,
"grad_norm": 0.5464402018931301,
"learning_rate": 5.568100420530533e-06,
"loss": 0.2639,
"step": 39960
},
{
"epoch": 4.660097947761194,
"grad_norm": 0.49288058807736795,
"learning_rate": 5.566164730535119e-06,
"loss": 0.2647,
"step": 39965
},
{
"epoch": 4.6606809701492535,
"grad_norm": 0.5437758561165839,
"learning_rate": 5.5642323018908595e-06,
"loss": 0.2703,
"step": 39970
},
{
"epoch": 4.661263992537314,
"grad_norm": 0.5346299303623872,
"learning_rate": 5.5623031348850815e-06,
"loss": 0.2569,
"step": 39975
},
{
"epoch": 4.661847014925373,
"grad_norm": 0.533172516862136,
"learning_rate": 5.560377229804644e-06,
"loss": 0.2582,
"step": 39980
},
{
"epoch": 4.662430037313433,
"grad_norm": 0.5399634745002946,
"learning_rate": 5.558454586935901e-06,
"loss": 0.2805,
"step": 39985
},
{
"epoch": 4.663013059701493,
"grad_norm": 0.47633814068425956,
"learning_rate": 5.556535206564733e-06,
"loss": 0.255,
"step": 39990
},
{
"epoch": 4.663596082089552,
"grad_norm": 0.518734948250553,
"learning_rate": 5.554619088976538e-06,
"loss": 0.2497,
"step": 39995
},
{
"epoch": 4.664179104477612,
"grad_norm": 0.5554992653820158,
"learning_rate": 5.55270623445622e-06,
"loss": 0.2492,
"step": 40000
},
{
"epoch": 4.664762126865671,
"grad_norm": 0.5494475968014342,
"learning_rate": 5.5507966432882056e-06,
"loss": 0.2588,
"step": 40005
},
{
"epoch": 4.6653451492537314,
"grad_norm": 0.5386402157373839,
"learning_rate": 5.548890315756433e-06,
"loss": 0.2579,
"step": 40010
},
{
"epoch": 4.6659281716417915,
"grad_norm": 0.5444706039562287,
"learning_rate": 5.546987252144351e-06,
"loss": 0.2893,
"step": 40015
},
{
"epoch": 4.666511194029851,
"grad_norm": 0.4879983142995859,
"learning_rate": 5.545087452734928e-06,
"loss": 0.2458,
"step": 40020
},
{
"epoch": 4.667094216417911,
"grad_norm": 0.5557708654373923,
"learning_rate": 5.543190917810647e-06,
"loss": 0.2515,
"step": 40025
},
{
"epoch": 4.66767723880597,
"grad_norm": 0.5626302490494139,
"learning_rate": 5.541297647653505e-06,
"loss": 0.2566,
"step": 40030
},
{
"epoch": 4.66826026119403,
"grad_norm": 0.5570188393236565,
"learning_rate": 5.539407642545012e-06,
"loss": 0.2789,
"step": 40035
},
{
"epoch": 4.668843283582089,
"grad_norm": 0.5201331559099625,
"learning_rate": 5.537520902766193e-06,
"loss": 0.268,
"step": 40040
},
{
"epoch": 4.669426305970149,
"grad_norm": 0.543373508115044,
"learning_rate": 5.535637428597591e-06,
"loss": 0.269,
"step": 40045
},
{
"epoch": 4.6700093283582085,
"grad_norm": 0.5568551024166752,
"learning_rate": 5.533757220319257e-06,
"loss": 0.2549,
"step": 40050
},
{
"epoch": 4.6705923507462686,
"grad_norm": 0.5361543937930828,
"learning_rate": 5.531880278210764e-06,
"loss": 0.2474,
"step": 40055
},
{
"epoch": 4.671175373134329,
"grad_norm": 0.5332695866731167,
"learning_rate": 5.5300066025511885e-06,
"loss": 0.2569,
"step": 40060
},
{
"epoch": 4.671758395522388,
"grad_norm": 0.589542147153373,
"learning_rate": 5.528136193619137e-06,
"loss": 0.2773,
"step": 40065
},
{
"epoch": 4.672341417910448,
"grad_norm": 0.5398247393515335,
"learning_rate": 5.526269051692717e-06,
"loss": 0.2604,
"step": 40070
},
{
"epoch": 4.672924440298507,
"grad_norm": 0.5602495866514481,
"learning_rate": 5.524405177049553e-06,
"loss": 0.2557,
"step": 40075
},
{
"epoch": 4.673507462686567,
"grad_norm": 0.5057896777527263,
"learning_rate": 5.522544569966786e-06,
"loss": 0.2585,
"step": 40080
},
{
"epoch": 4.674090485074627,
"grad_norm": 0.5047447240054863,
"learning_rate": 5.520687230721073e-06,
"loss": 0.2586,
"step": 40085
},
{
"epoch": 4.674673507462686,
"grad_norm": 0.5719492302799094,
"learning_rate": 5.518833159588582e-06,
"loss": 0.2566,
"step": 40090
},
{
"epoch": 4.6752565298507465,
"grad_norm": 0.5195506542381323,
"learning_rate": 5.516982356844994e-06,
"loss": 0.2547,
"step": 40095
},
{
"epoch": 4.675839552238806,
"grad_norm": 0.4927618908976009,
"learning_rate": 5.515134822765504e-06,
"loss": 0.2472,
"step": 40100
},
{
"epoch": 4.676422574626866,
"grad_norm": 12.42300546948787,
"learning_rate": 5.513290557624827e-06,
"loss": 0.265,
"step": 40105
},
{
"epoch": 4.677005597014926,
"grad_norm": 0.5843261635336789,
"learning_rate": 5.511449561697183e-06,
"loss": 0.256,
"step": 40110
},
{
"epoch": 4.677588619402985,
"grad_norm": 0.6219886003787144,
"learning_rate": 5.509611835256317e-06,
"loss": 0.263,
"step": 40115
},
{
"epoch": 4.678171641791045,
"grad_norm": 0.563527995817142,
"learning_rate": 5.507777378575474e-06,
"loss": 0.2835,
"step": 40120
},
{
"epoch": 4.678754664179104,
"grad_norm": 0.4971100095783303,
"learning_rate": 5.505946191927424e-06,
"loss": 0.2659,
"step": 40125
},
{
"epoch": 4.679337686567164,
"grad_norm": 0.535047842222231,
"learning_rate": 5.504118275584444e-06,
"loss": 0.2419,
"step": 40130
},
{
"epoch": 4.6799207089552235,
"grad_norm": 0.5134808224822797,
"learning_rate": 5.5022936298183316e-06,
"loss": 0.2634,
"step": 40135
},
{
"epoch": 4.680503731343284,
"grad_norm": 0.5704217200750399,
"learning_rate": 5.500472254900392e-06,
"loss": 0.2861,
"step": 40140
},
{
"epoch": 4.681086753731344,
"grad_norm": 0.5068672631787788,
"learning_rate": 5.49865415110145e-06,
"loss": 0.2542,
"step": 40145
},
{
"epoch": 4.681669776119403,
"grad_norm": 0.5313202522974988,
"learning_rate": 5.49683931869183e-06,
"loss": 0.241,
"step": 40150
},
{
"epoch": 4.682252798507463,
"grad_norm": 0.5380947184806353,
"learning_rate": 5.495027757941394e-06,
"loss": 0.2525,
"step": 40155
},
{
"epoch": 4.682835820895522,
"grad_norm": 0.5295733825833271,
"learning_rate": 5.4932194691194905e-06,
"loss": 0.247,
"step": 40160
},
{
"epoch": 4.683418843283582,
"grad_norm": 0.5395528578987526,
"learning_rate": 5.491414452495006e-06,
"loss": 0.2479,
"step": 40165
},
{
"epoch": 4.684001865671641,
"grad_norm": 0.5966249657406705,
"learning_rate": 5.489612708336324e-06,
"loss": 0.274,
"step": 40170
},
{
"epoch": 4.684584888059701,
"grad_norm": 0.563176780579703,
"learning_rate": 5.487814236911344e-06,
"loss": 0.2607,
"step": 40175
},
{
"epoch": 4.6851679104477615,
"grad_norm": 0.5203259541293299,
"learning_rate": 5.486019038487483e-06,
"loss": 0.2772,
"step": 40180
},
{
"epoch": 4.685750932835821,
"grad_norm": 0.5344403281531664,
"learning_rate": 5.484227113331673e-06,
"loss": 0.2677,
"step": 40185
},
{
"epoch": 4.686333955223881,
"grad_norm": 0.5177440774023253,
"learning_rate": 5.482438461710355e-06,
"loss": 0.2613,
"step": 40190
},
{
"epoch": 4.68691697761194,
"grad_norm": 0.5305643996489512,
"learning_rate": 5.480653083889483e-06,
"loss": 0.2558,
"step": 40195
},
{
"epoch": 4.6875,
"grad_norm": 0.5786857544930206,
"learning_rate": 5.4788709801345244e-06,
"loss": 0.2625,
"step": 40200
},
{
"epoch": 4.68808302238806,
"grad_norm": 0.5069624380940012,
"learning_rate": 5.477092150710465e-06,
"loss": 0.2569,
"step": 40205
},
{
"epoch": 4.688666044776119,
"grad_norm": 0.5489895315358849,
"learning_rate": 5.475316595881796e-06,
"loss": 0.262,
"step": 40210
},
{
"epoch": 4.689249067164179,
"grad_norm": 0.535416688505873,
"learning_rate": 5.473544315912525e-06,
"loss": 0.2806,
"step": 40215
},
{
"epoch": 4.6898320895522385,
"grad_norm": 0.5196999033028088,
"learning_rate": 5.471775311066177e-06,
"loss": 0.2654,
"step": 40220
},
{
"epoch": 4.690415111940299,
"grad_norm": 0.5365107668452866,
"learning_rate": 5.470009581605784e-06,
"loss": 0.2651,
"step": 40225
},
{
"epoch": 4.690998134328359,
"grad_norm": 0.4878512621687332,
"learning_rate": 5.468247127793893e-06,
"loss": 0.2718,
"step": 40230
},
{
"epoch": 4.691581156716418,
"grad_norm": 0.5093462351218783,
"learning_rate": 5.46648794989256e-06,
"loss": 0.245,
"step": 40235
},
{
"epoch": 4.692164179104478,
"grad_norm": 0.5283044061072845,
"learning_rate": 5.464732048163365e-06,
"loss": 0.2652,
"step": 40240
},
{
"epoch": 4.692747201492537,
"grad_norm": 0.570138435529778,
"learning_rate": 5.462979422867388e-06,
"loss": 0.2958,
"step": 40245
},
{
"epoch": 4.693330223880597,
"grad_norm": 0.4825031535208165,
"learning_rate": 5.461230074265233e-06,
"loss": 0.2523,
"step": 40250
},
{
"epoch": 4.693913246268656,
"grad_norm": 0.5455881606816707,
"learning_rate": 5.459484002617008e-06,
"loss": 0.2535,
"step": 40255
},
{
"epoch": 4.694496268656716,
"grad_norm": 0.4903244304349601,
"learning_rate": 5.4577412081823355e-06,
"loss": 0.2615,
"step": 40260
},
{
"epoch": 4.6950792910447765,
"grad_norm": 0.5377195012743091,
"learning_rate": 5.456001691220357e-06,
"loss": 0.2653,
"step": 40265
},
{
"epoch": 4.695662313432836,
"grad_norm": 0.5292935130242334,
"learning_rate": 5.45426545198972e-06,
"loss": 0.2742,
"step": 40270
},
{
"epoch": 4.696245335820896,
"grad_norm": 0.5449815900092477,
"learning_rate": 5.452532490748581e-06,
"loss": 0.2791,
"step": 40275
},
{
"epoch": 4.696828358208955,
"grad_norm": 0.5402211664905027,
"learning_rate": 5.450802807754625e-06,
"loss": 0.2683,
"step": 40280
},
{
"epoch": 4.697411380597015,
"grad_norm": 0.5602627461356222,
"learning_rate": 5.449076403265029e-06,
"loss": 0.2609,
"step": 40285
},
{
"epoch": 4.697994402985074,
"grad_norm": 0.5371661097644681,
"learning_rate": 5.4473532775365026e-06,
"loss": 0.2652,
"step": 40290
},
{
"epoch": 4.698577425373134,
"grad_norm": 0.5055991793671961,
"learning_rate": 5.44563343082525e-06,
"loss": 0.2453,
"step": 40295
},
{
"epoch": 4.699160447761194,
"grad_norm": 0.5283521426803366,
"learning_rate": 5.443916863387002e-06,
"loss": 0.2695,
"step": 40300
},
{
"epoch": 4.6997434701492535,
"grad_norm": 0.5347185758970066,
"learning_rate": 5.4422035754769915e-06,
"loss": 0.2692,
"step": 40305
},
{
"epoch": 4.700326492537314,
"grad_norm": 0.5416159480228936,
"learning_rate": 5.4404935673499685e-06,
"loss": 0.2593,
"step": 40310
},
{
"epoch": 4.700909514925373,
"grad_norm": 0.511953850952864,
"learning_rate": 5.438786839260197e-06,
"loss": 0.2576,
"step": 40315
},
{
"epoch": 4.701492537313433,
"grad_norm": 0.5346523548979524,
"learning_rate": 5.437083391461452e-06,
"loss": 0.2751,
"step": 40320
},
{
"epoch": 4.702075559701493,
"grad_norm": 0.585314966757987,
"learning_rate": 5.4353832242070155e-06,
"loss": 0.257,
"step": 40325
},
{
"epoch": 4.702658582089552,
"grad_norm": 0.5304246627779913,
"learning_rate": 5.43368633774969e-06,
"loss": 0.2773,
"step": 40330
},
{
"epoch": 4.703241604477612,
"grad_norm": 0.5276605729069379,
"learning_rate": 5.43199273234178e-06,
"loss": 0.2613,
"step": 40335
},
{
"epoch": 4.703824626865671,
"grad_norm": 0.5367997118697415,
"learning_rate": 5.43030240823512e-06,
"loss": 0.2578,
"step": 40340
},
{
"epoch": 4.7044076492537314,
"grad_norm": 0.5098762450553499,
"learning_rate": 5.428615365681034e-06,
"loss": 0.2576,
"step": 40345
},
{
"epoch": 4.7049906716417915,
"grad_norm": 0.4949497556650158,
"learning_rate": 5.426931604930375e-06,
"loss": 0.2399,
"step": 40350
},
{
"epoch": 4.705573694029851,
"grad_norm": 0.5827573260925429,
"learning_rate": 5.425251126233498e-06,
"loss": 0.2609,
"step": 40355
},
{
"epoch": 4.706156716417911,
"grad_norm": 0.5425470511542679,
"learning_rate": 5.423573929840277e-06,
"loss": 0.2593,
"step": 40360
},
{
"epoch": 4.70673973880597,
"grad_norm": 0.5230301784054733,
"learning_rate": 5.421900016000093e-06,
"loss": 0.2694,
"step": 40365
},
{
"epoch": 4.70732276119403,
"grad_norm": 0.5335007537950297,
"learning_rate": 5.420229384961847e-06,
"loss": 0.2526,
"step": 40370
},
{
"epoch": 4.707905783582089,
"grad_norm": 0.5097797666119723,
"learning_rate": 5.418562036973937e-06,
"loss": 0.2686,
"step": 40375
},
{
"epoch": 4.708488805970149,
"grad_norm": 0.518227965926884,
"learning_rate": 5.416897972284287e-06,
"loss": 0.2537,
"step": 40380
},
{
"epoch": 4.7090718283582085,
"grad_norm": 0.532995801550675,
"learning_rate": 5.415237191140326e-06,
"loss": 0.2619,
"step": 40385
},
{
"epoch": 4.7096548507462686,
"grad_norm": 0.5120121487426791,
"learning_rate": 5.413579693788995e-06,
"loss": 0.2478,
"step": 40390
},
{
"epoch": 4.710237873134329,
"grad_norm": 0.5699124059926832,
"learning_rate": 5.411925480476752e-06,
"loss": 0.2814,
"step": 40395
},
{
"epoch": 4.710820895522388,
"grad_norm": 0.4817653417686525,
"learning_rate": 5.410274551449559e-06,
"loss": 0.2579,
"step": 40400
},
{
"epoch": 4.711403917910448,
"grad_norm": 0.5000307120977543,
"learning_rate": 5.408626906952895e-06,
"loss": 0.254,
"step": 40405
},
{
"epoch": 4.711986940298507,
"grad_norm": 0.5539859309501919,
"learning_rate": 5.406982547231746e-06,
"loss": 0.2586,
"step": 40410
},
{
"epoch": 4.712569962686567,
"grad_norm": 0.5030322900301132,
"learning_rate": 5.40534147253062e-06,
"loss": 0.2531,
"step": 40415
},
{
"epoch": 4.713152985074627,
"grad_norm": 0.5064228139112594,
"learning_rate": 5.403703683093517e-06,
"loss": 0.2694,
"step": 40420
},
{
"epoch": 4.713736007462686,
"grad_norm": 0.6053394100947187,
"learning_rate": 5.402069179163974e-06,
"loss": 0.2794,
"step": 40425
},
{
"epoch": 4.7143190298507465,
"grad_norm": 0.5284915173440232,
"learning_rate": 5.400437960985017e-06,
"loss": 0.2582,
"step": 40430
},
{
"epoch": 4.714902052238806,
"grad_norm": 0.5162057214643305,
"learning_rate": 5.398810028799196e-06,
"loss": 0.2695,
"step": 40435
},
{
"epoch": 4.715485074626866,
"grad_norm": 0.5218416661881635,
"learning_rate": 5.397185382848568e-06,
"loss": 0.2617,
"step": 40440
},
{
"epoch": 4.716068097014926,
"grad_norm": 0.5124408552690876,
"learning_rate": 5.3955640233747e-06,
"loss": 0.2708,
"step": 40445
},
{
"epoch": 4.716651119402985,
"grad_norm": 0.4967211274025528,
"learning_rate": 5.393945950618678e-06,
"loss": 0.2612,
"step": 40450
},
{
"epoch": 4.717234141791045,
"grad_norm": 0.5516360632320493,
"learning_rate": 5.392331164821091e-06,
"loss": 0.2588,
"step": 40455
},
{
"epoch": 4.717817164179104,
"grad_norm": 0.5251034968437159,
"learning_rate": 5.39071966622204e-06,
"loss": 0.2647,
"step": 40460
},
{
"epoch": 4.718400186567164,
"grad_norm": 0.5323528852485695,
"learning_rate": 5.3891114550611434e-06,
"loss": 0.279,
"step": 40465
},
{
"epoch": 4.7189832089552235,
"grad_norm": 0.5446225597407718,
"learning_rate": 5.387506531577523e-06,
"loss": 0.2574,
"step": 40470
},
{
"epoch": 4.719566231343284,
"grad_norm": 0.519963815275516,
"learning_rate": 5.385904896009821e-06,
"loss": 0.2644,
"step": 40475
},
{
"epoch": 4.720149253731344,
"grad_norm": 0.5711317539586047,
"learning_rate": 5.384306548596178e-06,
"loss": 0.2637,
"step": 40480
},
{
"epoch": 4.720732276119403,
"grad_norm": 0.5320934777742895,
"learning_rate": 5.382711489574259e-06,
"loss": 0.2633,
"step": 40485
},
{
"epoch": 4.721315298507463,
"grad_norm": 0.5258731490344641,
"learning_rate": 5.3811197191812296e-06,
"loss": 0.2523,
"step": 40490
},
{
"epoch": 4.721898320895522,
"grad_norm": 0.5204609157125093,
"learning_rate": 5.379531237653774e-06,
"loss": 0.2613,
"step": 40495
},
{
"epoch": 4.722481343283582,
"grad_norm": 0.49552434747541724,
"learning_rate": 5.377946045228084e-06,
"loss": 0.2448,
"step": 40500
},
{
"epoch": 4.723064365671641,
"grad_norm": 0.5685699400908784,
"learning_rate": 5.376364142139862e-06,
"loss": 0.291,
"step": 40505
},
{
"epoch": 4.723647388059701,
"grad_norm": 0.5229595114578796,
"learning_rate": 5.374785528624317e-06,
"loss": 0.2706,
"step": 40510
},
{
"epoch": 4.7242304104477615,
"grad_norm": 0.5530791731934469,
"learning_rate": 5.3732102049161845e-06,
"loss": 0.2662,
"step": 40515
},
{
"epoch": 4.724813432835821,
"grad_norm": 0.5389337714303444,
"learning_rate": 5.37163817124969e-06,
"loss": 0.2543,
"step": 40520
},
{
"epoch": 4.725396455223881,
"grad_norm": 0.5106618022840242,
"learning_rate": 5.370069427858584e-06,
"loss": 0.2566,
"step": 40525
},
{
"epoch": 4.72597947761194,
"grad_norm": 0.5233962639447487,
"learning_rate": 5.368503974976122e-06,
"loss": 0.2609,
"step": 40530
},
{
"epoch": 4.7265625,
"grad_norm": 0.5187126215016823,
"learning_rate": 5.366941812835075e-06,
"loss": 0.255,
"step": 40535
},
{
"epoch": 4.72714552238806,
"grad_norm": 0.5812894275984566,
"learning_rate": 5.36538294166772e-06,
"loss": 0.2774,
"step": 40540
},
{
"epoch": 4.727728544776119,
"grad_norm": 0.524220816828342,
"learning_rate": 5.363827361705844e-06,
"loss": 0.2556,
"step": 40545
},
{
"epoch": 4.728311567164179,
"grad_norm": 0.524510929715867,
"learning_rate": 5.362275073180749e-06,
"loss": 0.2737,
"step": 40550
},
{
"epoch": 4.7288945895522385,
"grad_norm": 0.5310585456415964,
"learning_rate": 5.360726076323246e-06,
"loss": 0.26,
"step": 40555
},
{
"epoch": 4.729477611940299,
"grad_norm": 0.5829572995701353,
"learning_rate": 5.3591803713636545e-06,
"loss": 0.2532,
"step": 40560
},
{
"epoch": 4.730060634328359,
"grad_norm": 0.49653735151619316,
"learning_rate": 5.357637958531805e-06,
"loss": 0.2677,
"step": 40565
},
{
"epoch": 4.730643656716418,
"grad_norm": 0.5483304065130487,
"learning_rate": 5.3560988380570405e-06,
"loss": 0.2671,
"step": 40570
},
{
"epoch": 4.731226679104478,
"grad_norm": 0.5614680943080873,
"learning_rate": 5.3545630101682155e-06,
"loss": 0.2667,
"step": 40575
},
{
"epoch": 4.731809701492537,
"grad_norm": 0.5035387139025418,
"learning_rate": 5.353030475093694e-06,
"loss": 0.2511,
"step": 40580
},
{
"epoch": 4.732392723880597,
"grad_norm": 0.5161985002683129,
"learning_rate": 5.351501233061343e-06,
"loss": 0.2738,
"step": 40585
},
{
"epoch": 4.732975746268656,
"grad_norm": 0.5531027710723204,
"learning_rate": 5.349975284298552e-06,
"loss": 0.2613,
"step": 40590
},
{
"epoch": 4.733558768656716,
"grad_norm": 0.5022586841054332,
"learning_rate": 5.348452629032209e-06,
"loss": 0.2692,
"step": 40595
},
{
"epoch": 4.7341417910447765,
"grad_norm": 0.5253995608443942,
"learning_rate": 5.346933267488726e-06,
"loss": 0.2648,
"step": 40600
},
{
"epoch": 4.734724813432836,
"grad_norm": 0.5104069186945384,
"learning_rate": 5.345417199894012e-06,
"loss": 0.2516,
"step": 40605
},
{
"epoch": 4.735307835820896,
"grad_norm": 0.559296865010764,
"learning_rate": 5.343904426473493e-06,
"loss": 0.2544,
"step": 40610
},
{
"epoch": 4.735890858208955,
"grad_norm": 0.5548292756496112,
"learning_rate": 5.342394947452106e-06,
"loss": 0.2689,
"step": 40615
},
{
"epoch": 4.736473880597015,
"grad_norm": 0.5239717307752404,
"learning_rate": 5.340888763054291e-06,
"loss": 0.2654,
"step": 40620
},
{
"epoch": 4.737056902985074,
"grad_norm": 0.5688670435061347,
"learning_rate": 5.3393858735040074e-06,
"loss": 0.268,
"step": 40625
},
{
"epoch": 4.737639925373134,
"grad_norm": 0.5198015841583459,
"learning_rate": 5.337886279024722e-06,
"loss": 0.2487,
"step": 40630
},
{
"epoch": 4.738222947761194,
"grad_norm": 0.5503052586607899,
"learning_rate": 5.336389979839405e-06,
"loss": 0.2556,
"step": 40635
},
{
"epoch": 4.7388059701492535,
"grad_norm": 0.4806789494119374,
"learning_rate": 5.3348969761705446e-06,
"loss": 0.2594,
"step": 40640
},
{
"epoch": 4.739388992537314,
"grad_norm": 0.49998377720445447,
"learning_rate": 5.3334072682401365e-06,
"loss": 0.2524,
"step": 40645
},
{
"epoch": 4.739972014925373,
"grad_norm": 0.5404004453849443,
"learning_rate": 5.331920856269686e-06,
"loss": 0.2525,
"step": 40650
},
{
"epoch": 4.740555037313433,
"grad_norm": 0.5412806575945461,
"learning_rate": 5.330437740480206e-06,
"loss": 0.2661,
"step": 40655
},
{
"epoch": 4.741138059701493,
"grad_norm": 0.4893659442503186,
"learning_rate": 5.328957921092224e-06,
"loss": 0.2619,
"step": 40660
},
{
"epoch": 4.741721082089552,
"grad_norm": 0.5552724233256167,
"learning_rate": 5.327481398325775e-06,
"loss": 0.265,
"step": 40665
},
{
"epoch": 4.742304104477612,
"grad_norm": 0.5489696726807467,
"learning_rate": 5.326008172400402e-06,
"loss": 0.2758,
"step": 40670
},
{
"epoch": 4.742887126865671,
"grad_norm": 0.500392286479777,
"learning_rate": 5.324538243535162e-06,
"loss": 0.2522,
"step": 40675
},
{
"epoch": 4.7434701492537314,
"grad_norm": 0.5139908114288545,
"learning_rate": 5.323071611948619e-06,
"loss": 0.2562,
"step": 40680
},
{
"epoch": 4.7440531716417915,
"grad_norm": 0.5034898497943342,
"learning_rate": 5.3216082778588426e-06,
"loss": 0.2526,
"step": 40685
},
{
"epoch": 4.744636194029851,
"grad_norm": 0.5437215997709941,
"learning_rate": 5.320148241483422e-06,
"loss": 0.2584,
"step": 40690
},
{
"epoch": 4.745219216417911,
"grad_norm": 0.5085616862564696,
"learning_rate": 5.318691503039448e-06,
"loss": 0.2821,
"step": 40695
},
{
"epoch": 4.74580223880597,
"grad_norm": 0.5749620527731026,
"learning_rate": 5.317238062743527e-06,
"loss": 0.2702,
"step": 40700
},
{
"epoch": 4.74638526119403,
"grad_norm": 0.5476980176559368,
"learning_rate": 5.315787920811766e-06,
"loss": 0.2519,
"step": 40705
},
{
"epoch": 4.746968283582089,
"grad_norm": 0.4931259630495937,
"learning_rate": 5.31434107745979e-06,
"loss": 0.245,
"step": 40710
},
{
"epoch": 4.747551305970149,
"grad_norm": 0.5631639328610191,
"learning_rate": 5.312897532902733e-06,
"loss": 0.2585,
"step": 40715
},
{
"epoch": 4.7481343283582085,
"grad_norm": 0.5407894786123406,
"learning_rate": 5.311457287355232e-06,
"loss": 0.2596,
"step": 40720
},
{
"epoch": 4.7487173507462686,
"grad_norm": 0.525482713469301,
"learning_rate": 5.310020341031439e-06,
"loss": 0.2536,
"step": 40725
},
{
"epoch": 4.749300373134329,
"grad_norm": 0.5247967627650835,
"learning_rate": 5.3085866941450185e-06,
"loss": 0.2442,
"step": 40730
},
{
"epoch": 4.749883395522388,
"grad_norm": 0.5372888627745188,
"learning_rate": 5.307156346909135e-06,
"loss": 0.2777,
"step": 40735
},
{
"epoch": 4.750466417910448,
"grad_norm": 0.509979551698758,
"learning_rate": 5.3057292995364695e-06,
"loss": 0.2643,
"step": 40740
},
{
"epoch": 4.751049440298507,
"grad_norm": 0.5673777341030805,
"learning_rate": 5.304305552239209e-06,
"loss": 0.2719,
"step": 40745
},
{
"epoch": 4.751632462686567,
"grad_norm": 0.6138984591484716,
"learning_rate": 5.302885105229052e-06,
"loss": 0.2473,
"step": 40750
},
{
"epoch": 4.752215485074627,
"grad_norm": 0.5511838873187485,
"learning_rate": 5.301467958717205e-06,
"loss": 0.2635,
"step": 40755
},
{
"epoch": 4.752798507462686,
"grad_norm": 0.5626985726020425,
"learning_rate": 5.300054112914385e-06,
"loss": 0.2637,
"step": 40760
},
{
"epoch": 4.7533815298507465,
"grad_norm": 0.5653948297545796,
"learning_rate": 5.298643568030817e-06,
"loss": 0.2602,
"step": 40765
},
{
"epoch": 4.753964552238806,
"grad_norm": 0.5995559376975006,
"learning_rate": 5.297236324276231e-06,
"loss": 0.2619,
"step": 40770
},
{
"epoch": 4.754547574626866,
"grad_norm": 0.4862563833282966,
"learning_rate": 5.295832381859881e-06,
"loss": 0.2473,
"step": 40775
},
{
"epoch": 4.755130597014926,
"grad_norm": 0.5391813031082727,
"learning_rate": 5.294431740990509e-06,
"loss": 0.251,
"step": 40780
},
{
"epoch": 4.755713619402985,
"grad_norm": 0.5059009637354451,
"learning_rate": 5.293034401876384e-06,
"loss": 0.2494,
"step": 40785
},
{
"epoch": 4.756296641791045,
"grad_norm": 0.5855608841540562,
"learning_rate": 5.291640364725272e-06,
"loss": 0.2731,
"step": 40790
},
{
"epoch": 4.756879664179104,
"grad_norm": 0.5309524178372007,
"learning_rate": 5.290249629744457e-06,
"loss": 0.2557,
"step": 40795
},
{
"epoch": 4.757462686567164,
"grad_norm": 0.5965301157496198,
"learning_rate": 5.288862197140726e-06,
"loss": 0.2693,
"step": 40800
},
{
"epoch": 4.7580457089552235,
"grad_norm": 0.5934960101275178,
"learning_rate": 5.28747806712038e-06,
"loss": 0.2659,
"step": 40805
},
{
"epoch": 4.758628731343284,
"grad_norm": 0.565589469279002,
"learning_rate": 5.286097239889219e-06,
"loss": 0.2785,
"step": 40810
},
{
"epoch": 4.759211753731344,
"grad_norm": 0.5046709999882028,
"learning_rate": 5.284719715652565e-06,
"loss": 0.2567,
"step": 40815
},
{
"epoch": 4.759794776119403,
"grad_norm": 0.5118365450693864,
"learning_rate": 5.283345494615238e-06,
"loss": 0.2557,
"step": 40820
},
{
"epoch": 4.760377798507463,
"grad_norm": 0.5205303636235599,
"learning_rate": 5.281974576981579e-06,
"loss": 0.256,
"step": 40825
},
{
"epoch": 4.760960820895522,
"grad_norm": 0.5395217583502233,
"learning_rate": 5.280606962955423e-06,
"loss": 0.2661,
"step": 40830
},
{
"epoch": 4.761543843283582,
"grad_norm": 0.5485553093877918,
"learning_rate": 5.279242652740121e-06,
"loss": 0.2776,
"step": 40835
},
{
"epoch": 4.762126865671641,
"grad_norm": 0.5254161397828818,
"learning_rate": 5.277881646538537e-06,
"loss": 0.2517,
"step": 40840
},
{
"epoch": 4.762709888059701,
"grad_norm": 0.5385952564235336,
"learning_rate": 5.276523944553039e-06,
"loss": 0.269,
"step": 40845
},
{
"epoch": 4.7632929104477615,
"grad_norm": 0.5207889980547087,
"learning_rate": 5.275169546985502e-06,
"loss": 0.249,
"step": 40850
},
{
"epoch": 4.763875932835821,
"grad_norm": 0.4855190115549899,
"learning_rate": 5.2738184540373165e-06,
"loss": 0.2468,
"step": 40855
},
{
"epoch": 4.764458955223881,
"grad_norm": 0.6091454621186038,
"learning_rate": 5.272470665909368e-06,
"loss": 0.271,
"step": 40860
},
{
"epoch": 4.76504197761194,
"grad_norm": 0.49247077988336013,
"learning_rate": 5.271126182802072e-06,
"loss": 0.2644,
"step": 40865
},
{
"epoch": 4.765625,
"grad_norm": 0.5095254342060529,
"learning_rate": 5.269785004915328e-06,
"loss": 0.2556,
"step": 40870
},
{
"epoch": 4.76620802238806,
"grad_norm": 0.5959620475722028,
"learning_rate": 5.268447132448565e-06,
"loss": 0.2763,
"step": 40875
},
{
"epoch": 4.766791044776119,
"grad_norm": 0.4929494771026853,
"learning_rate": 5.267112565600707e-06,
"loss": 0.2531,
"step": 40880
},
{
"epoch": 4.767374067164179,
"grad_norm": 0.5381638579536344,
"learning_rate": 5.265781304570194e-06,
"loss": 0.2544,
"step": 40885
},
{
"epoch": 4.7679570895522385,
"grad_norm": 0.5320593743143683,
"learning_rate": 5.26445334955497e-06,
"loss": 0.2494,
"step": 40890
},
{
"epoch": 4.768540111940299,
"grad_norm": 0.5045017763591447,
"learning_rate": 5.263128700752493e-06,
"loss": 0.2483,
"step": 40895
},
{
"epoch": 4.769123134328359,
"grad_norm": 0.5700101588266805,
"learning_rate": 5.261807358359719e-06,
"loss": 0.2723,
"step": 40900
},
{
"epoch": 4.769706156716418,
"grad_norm": 0.5635033094370631,
"learning_rate": 5.260489322573125e-06,
"loss": 0.2725,
"step": 40905
},
{
"epoch": 4.770289179104478,
"grad_norm": 0.6278980798894483,
"learning_rate": 5.259174593588688e-06,
"loss": 0.2538,
"step": 40910
},
{
"epoch": 4.770872201492537,
"grad_norm": 0.5626121390410826,
"learning_rate": 5.257863171601895e-06,
"loss": 0.2507,
"step": 40915
},
{
"epoch": 4.771455223880597,
"grad_norm": 0.5656869405961013,
"learning_rate": 5.25655505680774e-06,
"loss": 0.2576,
"step": 40920
},
{
"epoch": 4.772038246268656,
"grad_norm": 0.5341669013729338,
"learning_rate": 5.255250249400732e-06,
"loss": 0.2667,
"step": 40925
},
{
"epoch": 4.772621268656716,
"grad_norm": 0.5691003546065335,
"learning_rate": 5.253948749574879e-06,
"loss": 0.2784,
"step": 40930
},
{
"epoch": 4.7732042910447765,
"grad_norm": 0.5362236931275812,
"learning_rate": 5.252650557523707e-06,
"loss": 0.2567,
"step": 40935
},
{
"epoch": 4.773787313432836,
"grad_norm": 0.5153875571339367,
"learning_rate": 5.2513556734402384e-06,
"loss": 0.2547,
"step": 40940
},
{
"epoch": 4.774370335820896,
"grad_norm": 0.5544388593610338,
"learning_rate": 5.2500640975170116e-06,
"loss": 0.2742,
"step": 40945
},
{
"epoch": 4.774953358208955,
"grad_norm": 0.5186078722498859,
"learning_rate": 5.248775829946076e-06,
"loss": 0.26,
"step": 40950
},
{
"epoch": 4.775536380597015,
"grad_norm": 0.5039981307918615,
"learning_rate": 5.247490870918979e-06,
"loss": 0.2615,
"step": 40955
},
{
"epoch": 4.776119402985074,
"grad_norm": 0.5411918132359785,
"learning_rate": 5.2462092206267864e-06,
"loss": 0.2665,
"step": 40960
},
{
"epoch": 4.776702425373134,
"grad_norm": 0.5229118820120464,
"learning_rate": 5.244930879260062e-06,
"loss": 0.2624,
"step": 40965
},
{
"epoch": 4.777285447761194,
"grad_norm": 0.5592016814123764,
"learning_rate": 5.243655847008888e-06,
"loss": 0.2617,
"step": 40970
},
{
"epoch": 4.7778684701492535,
"grad_norm": 0.58770568209626,
"learning_rate": 5.242384124062848e-06,
"loss": 0.2651,
"step": 40975
},
{
"epoch": 4.778451492537314,
"grad_norm": 0.5386650433973039,
"learning_rate": 5.241115710611033e-06,
"loss": 0.24,
"step": 40980
},
{
"epoch": 4.779034514925373,
"grad_norm": 0.5263463685860263,
"learning_rate": 5.239850606842045e-06,
"loss": 0.2522,
"step": 40985
},
{
"epoch": 4.779617537313433,
"grad_norm": 0.49546510753620826,
"learning_rate": 5.2385888129439934e-06,
"loss": 0.2808,
"step": 40990
},
{
"epoch": 4.780200559701493,
"grad_norm": 0.5160900866158508,
"learning_rate": 5.237330329104494e-06,
"loss": 0.268,
"step": 40995
},
{
"epoch": 4.780783582089552,
"grad_norm": 0.5393541779676191,
"learning_rate": 5.236075155510675e-06,
"loss": 0.2731,
"step": 41000
},
{
"epoch": 4.781366604477612,
"grad_norm": 0.5075386549133473,
"learning_rate": 5.234823292349164e-06,
"loss": 0.2517,
"step": 41005
},
{
"epoch": 4.781949626865671,
"grad_norm": 0.5220473258091382,
"learning_rate": 5.2335747398061e-06,
"loss": 0.2504,
"step": 41010
},
{
"epoch": 4.7825326492537314,
"grad_norm": 0.5231653854748372,
"learning_rate": 5.2323294980671375e-06,
"loss": 0.2535,
"step": 41015
},
{
"epoch": 4.7831156716417915,
"grad_norm": 0.5170973159021661,
"learning_rate": 5.231087567317425e-06,
"loss": 0.2579,
"step": 41020
},
{
"epoch": 4.783698694029851,
"grad_norm": 0.4787404341100575,
"learning_rate": 5.229848947741629e-06,
"loss": 0.271,
"step": 41025
},
{
"epoch": 4.784281716417911,
"grad_norm": 0.5512414149848892,
"learning_rate": 5.228613639523922e-06,
"loss": 0.2701,
"step": 41030
},
{
"epoch": 4.78486473880597,
"grad_norm": 0.5135871981620789,
"learning_rate": 5.2273816428479785e-06,
"loss": 0.2448,
"step": 41035
},
{
"epoch": 4.78544776119403,
"grad_norm": 0.49858436633604963,
"learning_rate": 5.2261529578969905e-06,
"loss": 0.2445,
"step": 41040
},
{
"epoch": 4.786030783582089,
"grad_norm": 0.5510471618902676,
"learning_rate": 5.224927584853641e-06,
"loss": 0.2497,
"step": 41045
},
{
"epoch": 4.786613805970149,
"grad_norm": 0.8113177787151018,
"learning_rate": 5.223705523900145e-06,
"loss": 0.2595,
"step": 41050
},
{
"epoch": 4.7871968283582085,
"grad_norm": 0.5028236402233456,
"learning_rate": 5.2224867752181995e-06,
"loss": 0.2622,
"step": 41055
},
{
"epoch": 4.7877798507462686,
"grad_norm": 0.5069885429192494,
"learning_rate": 5.22127133898903e-06,
"loss": 0.2522,
"step": 41060
},
{
"epoch": 4.788362873134329,
"grad_norm": 0.5164495970262721,
"learning_rate": 5.220059215393352e-06,
"loss": 0.2786,
"step": 41065
},
{
"epoch": 4.788945895522388,
"grad_norm": 0.4950288137723687,
"learning_rate": 5.2188504046114005e-06,
"loss": 0.2586,
"step": 41070
},
{
"epoch": 4.789528917910448,
"grad_norm": 0.546256706251749,
"learning_rate": 5.217644906822914e-06,
"loss": 0.2624,
"step": 41075
},
{
"epoch": 4.790111940298507,
"grad_norm": 0.5324037872923388,
"learning_rate": 5.216442722207141e-06,
"loss": 0.2586,
"step": 41080
},
{
"epoch": 4.790694962686567,
"grad_norm": 0.5215843825027218,
"learning_rate": 5.215243850942831e-06,
"loss": 0.2544,
"step": 41085
},
{
"epoch": 4.791277985074627,
"grad_norm": 0.5247919475166944,
"learning_rate": 5.214048293208246e-06,
"loss": 0.2689,
"step": 41090
},
{
"epoch": 4.791861007462686,
"grad_norm": 0.5457803597659311,
"learning_rate": 5.212856049181154e-06,
"loss": 0.2775,
"step": 41095
},
{
"epoch": 4.7924440298507465,
"grad_norm": 0.5500037743985086,
"learning_rate": 5.211667119038829e-06,
"loss": 0.2669,
"step": 41100
},
{
"epoch": 4.793027052238806,
"grad_norm": 0.5141040864901906,
"learning_rate": 5.210481502958057e-06,
"loss": 0.2412,
"step": 41105
},
{
"epoch": 4.793610074626866,
"grad_norm": 0.5085772527750257,
"learning_rate": 5.209299201115125e-06,
"loss": 0.2677,
"step": 41110
},
{
"epoch": 4.794193097014926,
"grad_norm": 0.5400621924271326,
"learning_rate": 5.2081202136858296e-06,
"loss": 0.2602,
"step": 41115
},
{
"epoch": 4.794776119402985,
"grad_norm": 0.5204024070669308,
"learning_rate": 5.206944540845476e-06,
"loss": 0.2545,
"step": 41120
},
{
"epoch": 4.795359141791045,
"grad_norm": 0.5752393003235707,
"learning_rate": 5.205772182768876e-06,
"loss": 0.2572,
"step": 41125
},
{
"epoch": 4.795942164179104,
"grad_norm": 0.5273295672290484,
"learning_rate": 5.204603139630345e-06,
"loss": 0.2741,
"step": 41130
},
{
"epoch": 4.796525186567164,
"grad_norm": 0.5809780830806073,
"learning_rate": 5.2034374116037146e-06,
"loss": 0.2711,
"step": 41135
},
{
"epoch": 4.7971082089552235,
"grad_norm": 0.5082961951923609,
"learning_rate": 5.202274998862312e-06,
"loss": 0.2605,
"step": 41140
},
{
"epoch": 4.797691231343284,
"grad_norm": 0.5000814687993748,
"learning_rate": 5.2011159015789775e-06,
"loss": 0.2626,
"step": 41145
},
{
"epoch": 4.798274253731344,
"grad_norm": 0.5607675581199246,
"learning_rate": 5.199960119926059e-06,
"loss": 0.265,
"step": 41150
},
{
"epoch": 4.798857276119403,
"grad_norm": 0.5137968339531936,
"learning_rate": 5.19880765407541e-06,
"loss": 0.2494,
"step": 41155
},
{
"epoch": 4.799440298507463,
"grad_norm": 0.593887723657077,
"learning_rate": 5.197658504198392e-06,
"loss": 0.2768,
"step": 41160
},
{
"epoch": 4.800023320895522,
"grad_norm": 0.5074639536519121,
"learning_rate": 5.19651267046587e-06,
"loss": 0.2584,
"step": 41165
},
{
"epoch": 4.800606343283582,
"grad_norm": 0.5227648618254918,
"learning_rate": 5.1953701530482215e-06,
"loss": 0.2636,
"step": 41170
},
{
"epoch": 4.801189365671641,
"grad_norm": 0.5476621550591564,
"learning_rate": 5.194230952115327e-06,
"loss": 0.2688,
"step": 41175
},
{
"epoch": 4.801772388059701,
"grad_norm": 0.5210770379297213,
"learning_rate": 5.1930950678365715e-06,
"loss": 0.2854,
"step": 41180
},
{
"epoch": 4.8023554104477615,
"grad_norm": 0.5216575254526342,
"learning_rate": 5.191962500380854e-06,
"loss": 0.262,
"step": 41185
},
{
"epoch": 4.802938432835821,
"grad_norm": 0.5171537386595773,
"learning_rate": 5.190833249916577e-06,
"loss": 0.2733,
"step": 41190
},
{
"epoch": 4.803521455223881,
"grad_norm": 0.5813908164077521,
"learning_rate": 5.189707316611646e-06,
"loss": 0.2495,
"step": 41195
},
{
"epoch": 4.80410447761194,
"grad_norm": 0.5586557732836941,
"learning_rate": 5.188584700633478e-06,
"loss": 0.241,
"step": 41200
},
{
"epoch": 4.8046875,
"grad_norm": 0.5704331772628655,
"learning_rate": 5.187465402148996e-06,
"loss": 0.2714,
"step": 41205
},
{
"epoch": 4.80527052238806,
"grad_norm": 0.5038053712504418,
"learning_rate": 5.186349421324627e-06,
"loss": 0.2521,
"step": 41210
},
{
"epoch": 4.805853544776119,
"grad_norm": 0.48991005015762856,
"learning_rate": 5.185236758326307e-06,
"loss": 0.2465,
"step": 41215
},
{
"epoch": 4.806436567164179,
"grad_norm": 0.5590184867533937,
"learning_rate": 5.184127413319482e-06,
"loss": 0.2564,
"step": 41220
},
{
"epoch": 4.8070195895522385,
"grad_norm": 0.5390856299523609,
"learning_rate": 5.183021386469096e-06,
"loss": 0.2651,
"step": 41225
},
{
"epoch": 4.807602611940299,
"grad_norm": 0.5153705347905203,
"learning_rate": 5.181918677939608e-06,
"loss": 0.2394,
"step": 41230
},
{
"epoch": 4.808185634328359,
"grad_norm": 0.5204901147551954,
"learning_rate": 5.180819287894979e-06,
"loss": 0.2488,
"step": 41235
},
{
"epoch": 4.808768656716418,
"grad_norm": 0.5024091636388212,
"learning_rate": 5.179723216498677e-06,
"loss": 0.2562,
"step": 41240
},
{
"epoch": 4.809351679104478,
"grad_norm": 0.5366827271698726,
"learning_rate": 5.178630463913678e-06,
"loss": 0.2663,
"step": 41245
},
{
"epoch": 4.809934701492537,
"grad_norm": 0.5671695408736948,
"learning_rate": 5.177541030302462e-06,
"loss": 0.2702,
"step": 41250
},
{
"epoch": 4.810517723880597,
"grad_norm": 0.5516268347808859,
"learning_rate": 5.176454915827022e-06,
"loss": 0.2582,
"step": 41255
},
{
"epoch": 4.811100746268656,
"grad_norm": 0.5179267846099194,
"learning_rate": 5.17537212064885e-06,
"loss": 0.2809,
"step": 41260
},
{
"epoch": 4.811683768656716,
"grad_norm": 0.5722368640770885,
"learning_rate": 5.174292644928947e-06,
"loss": 0.2552,
"step": 41265
},
{
"epoch": 4.8122667910447765,
"grad_norm": 0.5146685148838908,
"learning_rate": 5.173216488827822e-06,
"loss": 0.2505,
"step": 41270
},
{
"epoch": 4.812849813432836,
"grad_norm": 0.5008876262394624,
"learning_rate": 5.172143652505484e-06,
"loss": 0.2469,
"step": 41275
},
{
"epoch": 4.813432835820896,
"grad_norm": 0.5546237872143527,
"learning_rate": 5.171074136121461e-06,
"loss": 0.2642,
"step": 41280
},
{
"epoch": 4.814015858208955,
"grad_norm": 0.5142335572112503,
"learning_rate": 5.170007939834775e-06,
"loss": 0.26,
"step": 41285
},
{
"epoch": 4.814598880597015,
"grad_norm": 0.49684680340233983,
"learning_rate": 5.168945063803962e-06,
"loss": 0.2638,
"step": 41290
},
{
"epoch": 4.815181902985074,
"grad_norm": 0.5244666706893247,
"learning_rate": 5.167885508187059e-06,
"loss": 0.2554,
"step": 41295
},
{
"epoch": 4.815764925373134,
"grad_norm": 0.6252559958215652,
"learning_rate": 5.166829273141612e-06,
"loss": 0.2836,
"step": 41300
},
{
"epoch": 4.816347947761194,
"grad_norm": 0.4941692603058814,
"learning_rate": 5.165776358824675e-06,
"loss": 0.2483,
"step": 41305
},
{
"epoch": 4.8169309701492535,
"grad_norm": 0.490494548706234,
"learning_rate": 5.164726765392805e-06,
"loss": 0.2504,
"step": 41310
},
{
"epoch": 4.817513992537314,
"grad_norm": 0.5106466529862049,
"learning_rate": 5.163680493002067e-06,
"loss": 0.25,
"step": 41315
},
{
"epoch": 4.818097014925373,
"grad_norm": 0.5760332400543562,
"learning_rate": 5.162637541808031e-06,
"loss": 0.2742,
"step": 41320
},
{
"epoch": 4.818680037313433,
"grad_norm": 0.5549388778893086,
"learning_rate": 5.161597911965775e-06,
"loss": 0.2693,
"step": 41325
},
{
"epoch": 4.819263059701493,
"grad_norm": 0.5452292126226116,
"learning_rate": 5.16056160362988e-06,
"loss": 0.2658,
"step": 41330
},
{
"epoch": 4.819846082089552,
"grad_norm": 0.49410615896944976,
"learning_rate": 5.159528616954435e-06,
"loss": 0.2529,
"step": 41335
},
{
"epoch": 4.820429104477612,
"grad_norm": 0.5279706776887558,
"learning_rate": 5.158498952093038e-06,
"loss": 0.2605,
"step": 41340
},
{
"epoch": 4.821012126865671,
"grad_norm": 0.6082692977800148,
"learning_rate": 5.157472609198789e-06,
"loss": 0.2657,
"step": 41345
},
{
"epoch": 4.8215951492537314,
"grad_norm": 0.5572523430582565,
"learning_rate": 5.156449588424295e-06,
"loss": 0.2571,
"step": 41350
},
{
"epoch": 4.8221781716417915,
"grad_norm": 0.5309507134319285,
"learning_rate": 5.155429889921669e-06,
"loss": 0.2394,
"step": 41355
},
{
"epoch": 4.822761194029851,
"grad_norm": 0.538160114032342,
"learning_rate": 5.154413513842533e-06,
"loss": 0.2489,
"step": 41360
},
{
"epoch": 4.823344216417911,
"grad_norm": 0.5687588679402671,
"learning_rate": 5.153400460338007e-06,
"loss": 0.2599,
"step": 41365
},
{
"epoch": 4.82392723880597,
"grad_norm": 0.5386766325848653,
"learning_rate": 5.152390729558727e-06,
"loss": 0.2595,
"step": 41370
},
{
"epoch": 4.82451026119403,
"grad_norm": 0.5107716584296756,
"learning_rate": 5.151384321654828e-06,
"loss": 0.2608,
"step": 41375
},
{
"epoch": 4.825093283582089,
"grad_norm": 0.5359963200171824,
"learning_rate": 5.1503812367759575e-06,
"loss": 0.2594,
"step": 41380
},
{
"epoch": 4.825676305970149,
"grad_norm": 0.53938224010963,
"learning_rate": 5.149381475071259e-06,
"loss": 0.2704,
"step": 41385
},
{
"epoch": 4.8262593283582085,
"grad_norm": 0.523481136237025,
"learning_rate": 5.148385036689391e-06,
"loss": 0.2655,
"step": 41390
},
{
"epoch": 4.8268423507462686,
"grad_norm": 0.5531309878440063,
"learning_rate": 5.14739192177851e-06,
"loss": 0.2629,
"step": 41395
},
{
"epoch": 4.827425373134329,
"grad_norm": 0.5487188548709266,
"learning_rate": 5.146402130486288e-06,
"loss": 0.2598,
"step": 41400
},
{
"epoch": 4.828008395522388,
"grad_norm": 0.483868133541718,
"learning_rate": 5.145415662959895e-06,
"loss": 0.2519,
"step": 41405
},
{
"epoch": 4.828591417910448,
"grad_norm": 0.5460956093552999,
"learning_rate": 5.144432519346011e-06,
"loss": 0.2628,
"step": 41410
},
{
"epoch": 4.829174440298507,
"grad_norm": 0.4684887677128535,
"learning_rate": 5.143452699790817e-06,
"loss": 0.2573,
"step": 41415
},
{
"epoch": 4.829757462686567,
"grad_norm": 0.5176221523453539,
"learning_rate": 5.142476204440002e-06,
"loss": 0.2541,
"step": 41420
},
{
"epoch": 4.830340485074627,
"grad_norm": 0.542325095255919,
"learning_rate": 5.141503033438769e-06,
"loss": 0.271,
"step": 41425
},
{
"epoch": 4.830923507462686,
"grad_norm": 0.5277846578558064,
"learning_rate": 5.140533186931809e-06,
"loss": 0.2643,
"step": 41430
},
{
"epoch": 4.8315065298507465,
"grad_norm": 0.5046873588315237,
"learning_rate": 5.139566665063337e-06,
"loss": 0.261,
"step": 41435
},
{
"epoch": 4.832089552238806,
"grad_norm": 0.5400667280080967,
"learning_rate": 5.138603467977062e-06,
"loss": 0.2538,
"step": 41440
},
{
"epoch": 4.832672574626866,
"grad_norm": 0.5617397499863674,
"learning_rate": 5.137643595816202e-06,
"loss": 0.2638,
"step": 41445
},
{
"epoch": 4.833255597014926,
"grad_norm": 0.5542382982549092,
"learning_rate": 5.136687048723483e-06,
"loss": 0.2605,
"step": 41450
},
{
"epoch": 4.833838619402985,
"grad_norm": 0.5600659457148531,
"learning_rate": 5.135733826841131e-06,
"loss": 0.2695,
"step": 41455
},
{
"epoch": 4.834421641791045,
"grad_norm": 0.5662640252754004,
"learning_rate": 5.134783930310883e-06,
"loss": 0.2864,
"step": 41460
},
{
"epoch": 4.835004664179104,
"grad_norm": 0.5118629372594454,
"learning_rate": 5.13383735927398e-06,
"loss": 0.251,
"step": 41465
},
{
"epoch": 4.835587686567164,
"grad_norm": 0.5466234570907842,
"learning_rate": 5.132894113871167e-06,
"loss": 0.2764,
"step": 41470
},
{
"epoch": 4.8361707089552235,
"grad_norm": 0.4682026764983122,
"learning_rate": 5.131954194242696e-06,
"loss": 0.2511,
"step": 41475
},
{
"epoch": 4.836753731343284,
"grad_norm": 0.5064540164484578,
"learning_rate": 5.131017600528324e-06,
"loss": 0.2595,
"step": 41480
},
{
"epoch": 4.837336753731344,
"grad_norm": 0.5180824294976669,
"learning_rate": 5.130084332867315e-06,
"loss": 0.2643,
"step": 41485
},
{
"epoch": 4.837919776119403,
"grad_norm": 0.5387925620436101,
"learning_rate": 5.129154391398433e-06,
"loss": 0.2577,
"step": 41490
},
{
"epoch": 4.838502798507463,
"grad_norm": 0.5343000066978911,
"learning_rate": 5.128227776259953e-06,
"loss": 0.2637,
"step": 41495
},
{
"epoch": 4.839085820895522,
"grad_norm": 0.5690264309006462,
"learning_rate": 5.127304487589658e-06,
"loss": 0.273,
"step": 41500
},
{
"epoch": 4.839668843283582,
"grad_norm": 0.5326039894970697,
"learning_rate": 5.126384525524826e-06,
"loss": 0.2596,
"step": 41505
},
{
"epoch": 4.840251865671641,
"grad_norm": 0.5213989444377917,
"learning_rate": 5.12546789020225e-06,
"loss": 0.2576,
"step": 41510
},
{
"epoch": 4.840834888059701,
"grad_norm": 0.5202532610547281,
"learning_rate": 5.124554581758225e-06,
"loss": 0.2698,
"step": 41515
},
{
"epoch": 4.8414179104477615,
"grad_norm": 0.5381840289433081,
"learning_rate": 5.123644600328549e-06,
"loss": 0.2522,
"step": 41520
},
{
"epoch": 4.842000932835821,
"grad_norm": 0.5360853614588602,
"learning_rate": 5.122737946048531e-06,
"loss": 0.2514,
"step": 41525
},
{
"epoch": 4.842583955223881,
"grad_norm": 0.5440661029265731,
"learning_rate": 5.121834619052979e-06,
"loss": 0.2667,
"step": 41530
},
{
"epoch": 4.84316697761194,
"grad_norm": 0.6222529165472798,
"learning_rate": 5.1209346194762086e-06,
"loss": 0.2767,
"step": 41535
},
{
"epoch": 4.84375,
"grad_norm": 0.5104595406737876,
"learning_rate": 5.120037947452043e-06,
"loss": 0.2536,
"step": 41540
},
{
"epoch": 4.84433302238806,
"grad_norm": 0.5454977030369529,
"learning_rate": 5.119144603113809e-06,
"loss": 0.2633,
"step": 41545
},
{
"epoch": 4.844916044776119,
"grad_norm": 0.4894074696870294,
"learning_rate": 5.118254586594335e-06,
"loss": 0.239,
"step": 41550
},
{
"epoch": 4.845499067164179,
"grad_norm": 0.5171996109115735,
"learning_rate": 5.117367898025964e-06,
"loss": 0.258,
"step": 41555
},
{
"epoch": 4.8460820895522385,
"grad_norm": 0.4713522480727568,
"learning_rate": 5.116484537540532e-06,
"loss": 0.2538,
"step": 41560
},
{
"epoch": 4.846665111940299,
"grad_norm": 0.5275016619452988,
"learning_rate": 5.115604505269388e-06,
"loss": 0.2529,
"step": 41565
},
{
"epoch": 4.847248134328359,
"grad_norm": 0.5513663378704136,
"learning_rate": 5.114727801343385e-06,
"loss": 0.2599,
"step": 41570
},
{
"epoch": 4.847831156716418,
"grad_norm": 0.4682971768023134,
"learning_rate": 5.113854425892884e-06,
"loss": 0.2617,
"step": 41575
},
{
"epoch": 4.848414179104478,
"grad_norm": 0.5358951952426247,
"learning_rate": 5.11298437904774e-06,
"loss": 0.2743,
"step": 41580
},
{
"epoch": 4.848997201492537,
"grad_norm": 0.5157316494416937,
"learning_rate": 5.112117660937328e-06,
"loss": 0.2599,
"step": 41585
},
{
"epoch": 4.849580223880597,
"grad_norm": 0.5944301217961508,
"learning_rate": 5.111254271690516e-06,
"loss": 0.2837,
"step": 41590
},
{
"epoch": 4.850163246268656,
"grad_norm": 0.5256746516596107,
"learning_rate": 5.110394211435682e-06,
"loss": 0.2594,
"step": 41595
},
{
"epoch": 4.850746268656716,
"grad_norm": 0.4762454762299095,
"learning_rate": 5.1095374803007115e-06,
"loss": 0.2433,
"step": 41600
},
{
"epoch": 4.8513292910447765,
"grad_norm": 0.5734092388377269,
"learning_rate": 5.1086840784129895e-06,
"loss": 0.257,
"step": 41605
},
{
"epoch": 4.851912313432836,
"grad_norm": 0.5445861592902486,
"learning_rate": 5.107834005899409e-06,
"loss": 0.2732,
"step": 41610
},
{
"epoch": 4.852495335820896,
"grad_norm": 0.5224875973279762,
"learning_rate": 5.106987262886371e-06,
"loss": 0.2631,
"step": 41615
},
{
"epoch": 4.853078358208955,
"grad_norm": 0.5095682730083747,
"learning_rate": 5.1061438494997726e-06,
"loss": 0.2466,
"step": 41620
},
{
"epoch": 4.853661380597015,
"grad_norm": 0.4893920407590629,
"learning_rate": 5.105303765865026e-06,
"loss": 0.267,
"step": 41625
},
{
"epoch": 4.854244402985074,
"grad_norm": 0.5288169382606015,
"learning_rate": 5.104467012107041e-06,
"loss": 0.27,
"step": 41630
},
{
"epoch": 4.854827425373134,
"grad_norm": 0.47419288178148694,
"learning_rate": 5.103633588350236e-06,
"loss": 0.2412,
"step": 41635
},
{
"epoch": 4.855410447761194,
"grad_norm": 0.5475481127412534,
"learning_rate": 5.102803494718532e-06,
"loss": 0.2803,
"step": 41640
},
{
"epoch": 4.8559934701492535,
"grad_norm": 0.5035109227563629,
"learning_rate": 5.1019767313353564e-06,
"loss": 0.254,
"step": 41645
},
{
"epoch": 4.856576492537314,
"grad_norm": 0.5848075197635482,
"learning_rate": 5.101153298323643e-06,
"loss": 0.269,
"step": 41650
},
{
"epoch": 4.857159514925373,
"grad_norm": 0.5471125756188394,
"learning_rate": 5.100333195805823e-06,
"loss": 0.2648,
"step": 41655
},
{
"epoch": 4.857742537313433,
"grad_norm": 0.5511916141410351,
"learning_rate": 5.099516423903844e-06,
"loss": 0.2685,
"step": 41660
},
{
"epoch": 4.858325559701493,
"grad_norm": 0.5121656009662893,
"learning_rate": 5.098702982739151e-06,
"loss": 0.2532,
"step": 41665
},
{
"epoch": 4.858908582089552,
"grad_norm": 0.4893711976943801,
"learning_rate": 5.097892872432691e-06,
"loss": 0.2389,
"step": 41670
},
{
"epoch": 4.859491604477612,
"grad_norm": 0.5103260050459835,
"learning_rate": 5.097086093104924e-06,
"loss": 0.2606,
"step": 41675
},
{
"epoch": 4.860074626865671,
"grad_norm": 0.5045149224836615,
"learning_rate": 5.096282644875807e-06,
"loss": 0.2411,
"step": 41680
},
{
"epoch": 4.8606576492537314,
"grad_norm": 0.546932084264033,
"learning_rate": 5.095482527864808e-06,
"loss": 0.2622,
"step": 41685
},
{
"epoch": 4.8612406716417915,
"grad_norm": 0.5905858554988268,
"learning_rate": 5.094685742190896e-06,
"loss": 0.2475,
"step": 41690
},
{
"epoch": 4.861823694029851,
"grad_norm": 0.5635884405524049,
"learning_rate": 5.09389228797254e-06,
"loss": 0.2578,
"step": 41695
},
{
"epoch": 4.862406716417911,
"grad_norm": 0.5446643924445254,
"learning_rate": 5.093102165327729e-06,
"loss": 0.2711,
"step": 41700
},
{
"epoch": 4.86298973880597,
"grad_norm": 0.5388030732595104,
"learning_rate": 5.092315374373937e-06,
"loss": 0.2605,
"step": 41705
},
{
"epoch": 4.86357276119403,
"grad_norm": 0.5104379618728769,
"learning_rate": 5.09153191522816e-06,
"loss": 0.2506,
"step": 41710
},
{
"epoch": 4.864155783582089,
"grad_norm": 0.5442195807888268,
"learning_rate": 5.090751788006885e-06,
"loss": 0.2565,
"step": 41715
},
{
"epoch": 4.864738805970149,
"grad_norm": 0.5151659097617564,
"learning_rate": 5.089974992826117e-06,
"loss": 0.2546,
"step": 41720
},
{
"epoch": 4.8653218283582085,
"grad_norm": 0.5475812818819479,
"learning_rate": 5.0892015298013485e-06,
"loss": 0.2628,
"step": 41725
},
{
"epoch": 4.8659048507462686,
"grad_norm": 0.5611781041571859,
"learning_rate": 5.08843139904759e-06,
"loss": 0.2607,
"step": 41730
},
{
"epoch": 4.866487873134329,
"grad_norm": 0.562604530991125,
"learning_rate": 5.087664600679356e-06,
"loss": 0.2703,
"step": 41735
},
{
"epoch": 4.867070895522388,
"grad_norm": 0.4998378795657582,
"learning_rate": 5.086901134810658e-06,
"loss": 0.2408,
"step": 41740
},
{
"epoch": 4.867653917910448,
"grad_norm": 0.4891612223302177,
"learning_rate": 5.086141001555016e-06,
"loss": 0.2463,
"step": 41745
},
{
"epoch": 4.868236940298507,
"grad_norm": 0.49336471996126907,
"learning_rate": 5.085384201025457e-06,
"loss": 0.2739,
"step": 41750
},
{
"epoch": 4.868819962686567,
"grad_norm": 0.5629493957193611,
"learning_rate": 5.084630733334508e-06,
"loss": 0.2798,
"step": 41755
},
{
"epoch": 4.869402985074627,
"grad_norm": 0.5516219780327767,
"learning_rate": 5.083880598594204e-06,
"loss": 0.2698,
"step": 41760
},
{
"epoch": 4.869986007462686,
"grad_norm": 0.5332212008821925,
"learning_rate": 5.083133796916081e-06,
"loss": 0.2549,
"step": 41765
},
{
"epoch": 4.8705690298507465,
"grad_norm": 0.5009266210637915,
"learning_rate": 5.082390328411184e-06,
"loss": 0.2572,
"step": 41770
},
{
"epoch": 4.871152052238806,
"grad_norm": 0.526802423608002,
"learning_rate": 5.081650193190057e-06,
"loss": 0.2671,
"step": 41775
},
{
"epoch": 4.871735074626866,
"grad_norm": 0.5137792697456302,
"learning_rate": 5.080913391362749e-06,
"loss": 0.2627,
"step": 41780
},
{
"epoch": 4.872318097014926,
"grad_norm": 0.5984306399966873,
"learning_rate": 5.080179923038822e-06,
"loss": 0.2764,
"step": 41785
},
{
"epoch": 4.872901119402985,
"grad_norm": 0.5094423400085093,
"learning_rate": 5.079449788327332e-06,
"loss": 0.2661,
"step": 41790
},
{
"epoch": 4.873484141791045,
"grad_norm": 0.5427727261012149,
"learning_rate": 5.078722987336843e-06,
"loss": 0.2755,
"step": 41795
},
{
"epoch": 4.874067164179104,
"grad_norm": 0.49115019336966714,
"learning_rate": 5.0779995201754225e-06,
"loss": 0.2501,
"step": 41800
},
{
"epoch": 4.874650186567164,
"grad_norm": 0.4980343372980356,
"learning_rate": 5.077279386950642e-06,
"loss": 0.2569,
"step": 41805
},
{
"epoch": 4.8752332089552235,
"grad_norm": 0.579200878426523,
"learning_rate": 5.076562587769584e-06,
"loss": 0.2848,
"step": 41810
},
{
"epoch": 4.875816231343284,
"grad_norm": 0.5313901159630037,
"learning_rate": 5.0758491227388235e-06,
"loss": 0.2568,
"step": 41815
},
{
"epoch": 4.876399253731344,
"grad_norm": 0.5160576663677819,
"learning_rate": 5.07513899196445e-06,
"loss": 0.2552,
"step": 41820
},
{
"epoch": 4.876982276119403,
"grad_norm": 0.5357377913236523,
"learning_rate": 5.074432195552053e-06,
"loss": 0.2615,
"step": 41825
},
{
"epoch": 4.877565298507463,
"grad_norm": 0.5228674064442037,
"learning_rate": 5.073728733606722e-06,
"loss": 0.2558,
"step": 41830
},
{
"epoch": 4.878148320895522,
"grad_norm": 0.5488209572438764,
"learning_rate": 5.073028606233059e-06,
"loss": 0.2543,
"step": 41835
},
{
"epoch": 4.878731343283582,
"grad_norm": 0.5475836927974559,
"learning_rate": 5.072331813535166e-06,
"loss": 0.2472,
"step": 41840
},
{
"epoch": 4.879314365671641,
"grad_norm": 0.5122104063146348,
"learning_rate": 5.071638355616648e-06,
"loss": 0.2588,
"step": 41845
},
{
"epoch": 4.879897388059701,
"grad_norm": 0.5236345664813228,
"learning_rate": 5.070948232580618e-06,
"loss": 0.2564,
"step": 41850
},
{
"epoch": 4.8804804104477615,
"grad_norm": 0.5460049029970603,
"learning_rate": 5.070261444529688e-06,
"loss": 0.2517,
"step": 41855
},
{
"epoch": 4.881063432835821,
"grad_norm": 0.5282890574808324,
"learning_rate": 5.069577991565977e-06,
"loss": 0.2445,
"step": 41860
},
{
"epoch": 4.881646455223881,
"grad_norm": 0.5527711274915231,
"learning_rate": 5.0688978737911085e-06,
"loss": 0.2751,
"step": 41865
},
{
"epoch": 4.88222947761194,
"grad_norm": 0.5873242083073615,
"learning_rate": 5.06822109130621e-06,
"loss": 0.2641,
"step": 41870
},
{
"epoch": 4.8828125,
"grad_norm": 0.6454795207223442,
"learning_rate": 5.067547644211914e-06,
"loss": 0.2661,
"step": 41875
},
{
"epoch": 4.88339552238806,
"grad_norm": 0.515405993205954,
"learning_rate": 5.066877532608349e-06,
"loss": 0.257,
"step": 41880
},
{
"epoch": 4.883978544776119,
"grad_norm": 0.5775533229734363,
"learning_rate": 5.066210756595164e-06,
"loss": 0.2666,
"step": 41885
},
{
"epoch": 4.884561567164179,
"grad_norm": 0.5682993188365679,
"learning_rate": 5.065547316271494e-06,
"loss": 0.2716,
"step": 41890
},
{
"epoch": 4.8851445895522385,
"grad_norm": 0.5021547582884069,
"learning_rate": 5.064887211735991e-06,
"loss": 0.247,
"step": 41895
},
{
"epoch": 4.885727611940299,
"grad_norm": 0.5351609081219216,
"learning_rate": 5.064230443086805e-06,
"loss": 0.2656,
"step": 41900
},
{
"epoch": 4.886310634328359,
"grad_norm": 0.5185797141181296,
"learning_rate": 5.0635770104215915e-06,
"loss": 0.2601,
"step": 41905
},
{
"epoch": 4.886893656716418,
"grad_norm": 0.4897415944298069,
"learning_rate": 5.062926913837507e-06,
"loss": 0.2647,
"step": 41910
},
{
"epoch": 4.887476679104478,
"grad_norm": 0.5146341622967224,
"learning_rate": 5.062280153431218e-06,
"loss": 0.2483,
"step": 41915
},
{
"epoch": 4.888059701492537,
"grad_norm": 0.5201069230866949,
"learning_rate": 5.06163672929889e-06,
"loss": 0.2639,
"step": 41920
},
{
"epoch": 4.888642723880597,
"grad_norm": 0.49058031879353137,
"learning_rate": 5.060996641536193e-06,
"loss": 0.2613,
"step": 41925
},
{
"epoch": 4.889225746268656,
"grad_norm": 0.5195891301651294,
"learning_rate": 5.060359890238305e-06,
"loss": 0.2673,
"step": 41930
},
{
"epoch": 4.889808768656716,
"grad_norm": 0.5534183022778316,
"learning_rate": 5.059726475499902e-06,
"loss": 0.2527,
"step": 41935
},
{
"epoch": 4.8903917910447765,
"grad_norm": 0.5331781774186816,
"learning_rate": 5.059096397415167e-06,
"loss": 0.2754,
"step": 41940
},
{
"epoch": 4.890974813432836,
"grad_norm": 0.5142689622820151,
"learning_rate": 5.058469656077789e-06,
"loss": 0.2557,
"step": 41945
},
{
"epoch": 4.891557835820896,
"grad_norm": 0.4939092021509841,
"learning_rate": 5.057846251580957e-06,
"loss": 0.2442,
"step": 41950
},
{
"epoch": 4.892140858208955,
"grad_norm": 0.5580170055851186,
"learning_rate": 5.057226184017362e-06,
"loss": 0.2732,
"step": 41955
},
{
"epoch": 4.892723880597015,
"grad_norm": 0.5560264363819188,
"learning_rate": 5.056609453479208e-06,
"loss": 0.2551,
"step": 41960
},
{
"epoch": 4.893306902985074,
"grad_norm": 0.5396528546006861,
"learning_rate": 5.055996060058192e-06,
"loss": 0.2646,
"step": 41965
},
{
"epoch": 4.893889925373134,
"grad_norm": 0.5068940114371213,
"learning_rate": 5.055386003845524e-06,
"loss": 0.2549,
"step": 41970
},
{
"epoch": 4.894472947761194,
"grad_norm": 0.5745579405298369,
"learning_rate": 5.054779284931909e-06,
"loss": 0.277,
"step": 41975
},
{
"epoch": 4.8950559701492535,
"grad_norm": 0.5806705260974102,
"learning_rate": 5.0541759034075645e-06,
"loss": 0.2495,
"step": 41980
},
{
"epoch": 4.895638992537314,
"grad_norm": 0.5562877907163609,
"learning_rate": 5.053575859362203e-06,
"loss": 0.2797,
"step": 41985
},
{
"epoch": 4.896222014925373,
"grad_norm": 0.50697607397542,
"learning_rate": 5.0529791528850515e-06,
"loss": 0.2529,
"step": 41990
},
{
"epoch": 4.896805037313433,
"grad_norm": 0.5483589167480247,
"learning_rate": 5.052385784064827e-06,
"loss": 0.2516,
"step": 41995
},
{
"epoch": 4.897388059701493,
"grad_norm": 0.5100587783263615,
"learning_rate": 5.051795752989764e-06,
"loss": 0.2535,
"step": 42000
},
{
"epoch": 4.897971082089552,
"grad_norm": 0.5757002834872101,
"learning_rate": 5.051209059747594e-06,
"loss": 0.2696,
"step": 42005
},
{
"epoch": 4.898554104477612,
"grad_norm": 0.5331593458553955,
"learning_rate": 5.050625704425547e-06,
"loss": 0.2623,
"step": 42010
},
{
"epoch": 4.899137126865671,
"grad_norm": 0.5283321720941034,
"learning_rate": 5.0500456871103686e-06,
"loss": 0.2446,
"step": 42015
},
{
"epoch": 4.8997201492537314,
"grad_norm": 0.5114177573958205,
"learning_rate": 5.049469007888298e-06,
"loss": 0.2493,
"step": 42020
},
{
"epoch": 4.9003031716417915,
"grad_norm": 0.5679235645395738,
"learning_rate": 5.048895666845084e-06,
"loss": 0.2668,
"step": 42025
},
{
"epoch": 4.900886194029851,
"grad_norm": 0.5652236312690536,
"learning_rate": 5.048325664065975e-06,
"loss": 0.2706,
"step": 42030
},
{
"epoch": 4.901469216417911,
"grad_norm": 0.5449377388643637,
"learning_rate": 5.047758999635728e-06,
"loss": 0.2699,
"step": 42035
},
{
"epoch": 4.90205223880597,
"grad_norm": 0.5568006658896507,
"learning_rate": 5.047195673638596e-06,
"loss": 0.277,
"step": 42040
},
{
"epoch": 4.90263526119403,
"grad_norm": 0.5024409873880541,
"learning_rate": 5.0466356861583445e-06,
"loss": 0.2498,
"step": 42045
},
{
"epoch": 4.903218283582089,
"grad_norm": 0.5619462307067028,
"learning_rate": 5.046079037278237e-06,
"loss": 0.2529,
"step": 42050
},
{
"epoch": 4.903801305970149,
"grad_norm": 0.5265679100431098,
"learning_rate": 5.0455257270810425e-06,
"loss": 0.2604,
"step": 42055
},
{
"epoch": 4.9043843283582085,
"grad_norm": 0.518928623322086,
"learning_rate": 5.044975755649028e-06,
"loss": 0.2544,
"step": 42060
},
{
"epoch": 4.9049673507462686,
"grad_norm": 0.5215674212626602,
"learning_rate": 5.044429123063977e-06,
"loss": 0.2721,
"step": 42065
},
{
"epoch": 4.905550373134329,
"grad_norm": 0.5200189509891983,
"learning_rate": 5.043885829407164e-06,
"loss": 0.259,
"step": 42070
},
{
"epoch": 4.906133395522388,
"grad_norm": 0.5527420023829943,
"learning_rate": 5.043345874759371e-06,
"loss": 0.2555,
"step": 42075
},
{
"epoch": 4.906716417910448,
"grad_norm": 0.5342382947782167,
"learning_rate": 5.042809259200885e-06,
"loss": 0.256,
"step": 42080
},
{
"epoch": 4.907299440298507,
"grad_norm": 0.5141683380541643,
"learning_rate": 5.042275982811495e-06,
"loss": 0.267,
"step": 42085
},
{
"epoch": 4.907882462686567,
"grad_norm": 0.5766851960493508,
"learning_rate": 5.041746045670495e-06,
"loss": 0.265,
"step": 42090
},
{
"epoch": 4.908465485074627,
"grad_norm": 0.5319560178381133,
"learning_rate": 5.041219447856681e-06,
"loss": 0.2586,
"step": 42095
},
{
"epoch": 4.909048507462686,
"grad_norm": 0.551495036550619,
"learning_rate": 5.040696189448356e-06,
"loss": 0.2508,
"step": 42100
},
{
"epoch": 4.9096315298507465,
"grad_norm": 0.5028715651681404,
"learning_rate": 5.040176270523318e-06,
"loss": 0.2572,
"step": 42105
},
{
"epoch": 4.910214552238806,
"grad_norm": 0.5854116582149756,
"learning_rate": 5.039659691158878e-06,
"loss": 0.2441,
"step": 42110
},
{
"epoch": 4.910797574626866,
"grad_norm": 0.5241056006342661,
"learning_rate": 5.039146451431845e-06,
"loss": 0.2485,
"step": 42115
},
{
"epoch": 4.911380597014926,
"grad_norm": 0.5299036291724838,
"learning_rate": 5.038636551418533e-06,
"loss": 0.26,
"step": 42120
},
{
"epoch": 4.911963619402985,
"grad_norm": 0.5693228608809118,
"learning_rate": 5.038129991194761e-06,
"loss": 0.2702,
"step": 42125
},
{
"epoch": 4.912546641791045,
"grad_norm": 0.540432097247365,
"learning_rate": 5.0376267708358455e-06,
"loss": 0.271,
"step": 42130
},
{
"epoch": 4.913129664179104,
"grad_norm": 0.565345592234274,
"learning_rate": 5.037126890416614e-06,
"loss": 0.2781,
"step": 42135
},
{
"epoch": 4.913712686567164,
"grad_norm": 0.5143268463409374,
"learning_rate": 5.036630350011395e-06,
"loss": 0.2669,
"step": 42140
},
{
"epoch": 4.9142957089552235,
"grad_norm": 0.5714145458883612,
"learning_rate": 5.036137149694013e-06,
"loss": 0.275,
"step": 42145
},
{
"epoch": 4.914878731343284,
"grad_norm": 0.510914157357852,
"learning_rate": 5.03564728953781e-06,
"loss": 0.2632,
"step": 42150
},
{
"epoch": 4.915461753731344,
"grad_norm": 0.5008171389831337,
"learning_rate": 5.035160769615618e-06,
"loss": 0.2508,
"step": 42155
},
{
"epoch": 4.916044776119403,
"grad_norm": 0.5749675729501972,
"learning_rate": 5.034677589999783e-06,
"loss": 0.2664,
"step": 42160
},
{
"epoch": 4.916627798507463,
"grad_norm": 0.5233900101671892,
"learning_rate": 5.034197750762141e-06,
"loss": 0.2527,
"step": 42165
},
{
"epoch": 4.917210820895522,
"grad_norm": 0.5376105637275604,
"learning_rate": 5.033721251974047e-06,
"loss": 0.2505,
"step": 42170
},
{
"epoch": 4.917793843283582,
"grad_norm": 0.5028727183182046,
"learning_rate": 5.03324809370635e-06,
"loss": 0.2587,
"step": 42175
},
{
"epoch": 4.918376865671641,
"grad_norm": 0.5708070256275176,
"learning_rate": 5.032778276029403e-06,
"loss": 0.2492,
"step": 42180
},
{
"epoch": 4.918959888059701,
"grad_norm": 0.5284780496823562,
"learning_rate": 5.032311799013064e-06,
"loss": 0.237,
"step": 42185
},
{
"epoch": 4.9195429104477615,
"grad_norm": 0.5437624915243243,
"learning_rate": 5.031848662726692e-06,
"loss": 0.2527,
"step": 42190
},
{
"epoch": 4.920125932835821,
"grad_norm": 0.5272630207823388,
"learning_rate": 5.031388867239153e-06,
"loss": 0.2516,
"step": 42195
},
{
"epoch": 4.920708955223881,
"grad_norm": 0.5609438463293948,
"learning_rate": 5.030932412618815e-06,
"loss": 0.2645,
"step": 42200
},
{
"epoch": 4.92129197761194,
"grad_norm": 0.5523835138093406,
"learning_rate": 5.030479298933544e-06,
"loss": 0.2527,
"step": 42205
},
{
"epoch": 4.921875,
"grad_norm": 0.5448438770973634,
"learning_rate": 5.030029526250719e-06,
"loss": 0.2696,
"step": 42210
},
{
"epoch": 4.92245802238806,
"grad_norm": 0.5334394635727333,
"learning_rate": 5.029583094637212e-06,
"loss": 0.2708,
"step": 42215
},
{
"epoch": 4.923041044776119,
"grad_norm": 0.5473799855053304,
"learning_rate": 5.029140004159409e-06,
"loss": 0.2665,
"step": 42220
},
{
"epoch": 4.923624067164179,
"grad_norm": 0.5088585387852517,
"learning_rate": 5.028700254883189e-06,
"loss": 0.2617,
"step": 42225
},
{
"epoch": 4.9242070895522385,
"grad_norm": 0.5526294910242032,
"learning_rate": 5.028263846873938e-06,
"loss": 0.2625,
"step": 42230
},
{
"epoch": 4.924790111940299,
"grad_norm": 0.5488602735603476,
"learning_rate": 5.027830780196549e-06,
"loss": 0.2553,
"step": 42235
},
{
"epoch": 4.925373134328359,
"grad_norm": 0.5256559656515214,
"learning_rate": 5.02740105491541e-06,
"loss": 0.2538,
"step": 42240
},
{
"epoch": 4.925956156716418,
"grad_norm": 0.49572639691711673,
"learning_rate": 5.026974671094422e-06,
"loss": 0.2777,
"step": 42245
},
{
"epoch": 4.926539179104478,
"grad_norm": 0.5130528437133531,
"learning_rate": 5.026551628796982e-06,
"loss": 0.276,
"step": 42250
},
{
"epoch": 4.927122201492537,
"grad_norm": 0.5252623718875515,
"learning_rate": 5.026131928085994e-06,
"loss": 0.2675,
"step": 42255
},
{
"epoch": 4.927705223880597,
"grad_norm": 0.5249616603820997,
"learning_rate": 5.025715569023859e-06,
"loss": 0.2512,
"step": 42260
},
{
"epoch": 4.928288246268656,
"grad_norm": 0.49308468732337574,
"learning_rate": 5.025302551672492e-06,
"loss": 0.2573,
"step": 42265
},
{
"epoch": 4.928871268656716,
"grad_norm": 0.5241375322385623,
"learning_rate": 5.024892876093299e-06,
"loss": 0.251,
"step": 42270
},
{
"epoch": 4.9294542910447765,
"grad_norm": 0.536547646668262,
"learning_rate": 5.024486542347199e-06,
"loss": 0.2529,
"step": 42275
},
{
"epoch": 4.930037313432836,
"grad_norm": 0.5432071331597679,
"learning_rate": 5.024083550494606e-06,
"loss": 0.2706,
"step": 42280
},
{
"epoch": 4.930620335820896,
"grad_norm": 0.519888884714539,
"learning_rate": 5.023683900595444e-06,
"loss": 0.2606,
"step": 42285
},
{
"epoch": 4.931203358208955,
"grad_norm": 0.5150429884830602,
"learning_rate": 5.023287592709136e-06,
"loss": 0.2551,
"step": 42290
},
{
"epoch": 4.931786380597015,
"grad_norm": 0.5381892098962242,
"learning_rate": 5.02289462689461e-06,
"loss": 0.2575,
"step": 42295
},
{
"epoch": 4.932369402985074,
"grad_norm": 0.5129812840405475,
"learning_rate": 5.0225050032102965e-06,
"loss": 0.2645,
"step": 42300
},
{
"epoch": 4.932952425373134,
"grad_norm": 0.5270017766303224,
"learning_rate": 5.022118721714127e-06,
"loss": 0.2562,
"step": 42305
},
{
"epoch": 4.933535447761194,
"grad_norm": 0.5607984557942124,
"learning_rate": 5.021735782463537e-06,
"loss": 0.2687,
"step": 42310
},
{
"epoch": 4.9341184701492535,
"grad_norm": 0.48585619759926013,
"learning_rate": 5.02135618551547e-06,
"loss": 0.2686,
"step": 42315
},
{
"epoch": 4.934701492537314,
"grad_norm": 0.5343965401543147,
"learning_rate": 5.020979930926365e-06,
"loss": 0.2575,
"step": 42320
},
{
"epoch": 4.935284514925373,
"grad_norm": 0.5572109602036827,
"learning_rate": 5.02060701875217e-06,
"loss": 0.2587,
"step": 42325
},
{
"epoch": 4.935867537313433,
"grad_norm": 0.4902847984858535,
"learning_rate": 5.020237449048333e-06,
"loss": 0.2561,
"step": 42330
},
{
"epoch": 4.936450559701493,
"grad_norm": 0.5390598769697923,
"learning_rate": 5.019871221869802e-06,
"loss": 0.2684,
"step": 42335
},
{
"epoch": 4.937033582089552,
"grad_norm": 0.5147941260934705,
"learning_rate": 5.0195083372710345e-06,
"loss": 0.2664,
"step": 42340
},
{
"epoch": 4.937616604477612,
"grad_norm": 0.548980577215786,
"learning_rate": 5.019148795305989e-06,
"loss": 0.2526,
"step": 42345
},
{
"epoch": 4.938199626865671,
"grad_norm": 0.5035908145770805,
"learning_rate": 5.018792596028123e-06,
"loss": 0.2587,
"step": 42350
},
{
"epoch": 4.9387826492537314,
"grad_norm": 0.5386835530616152,
"learning_rate": 5.018439739490402e-06,
"loss": 0.2459,
"step": 42355
},
{
"epoch": 4.9393656716417915,
"grad_norm": 0.5672610362941495,
"learning_rate": 5.018090225745291e-06,
"loss": 0.2659,
"step": 42360
},
{
"epoch": 4.939948694029851,
"grad_norm": 0.5227959253547065,
"learning_rate": 5.017744054844761e-06,
"loss": 0.2733,
"step": 42365
},
{
"epoch": 4.940531716417911,
"grad_norm": 0.5482815190946052,
"learning_rate": 5.017401226840284e-06,
"loss": 0.276,
"step": 42370
},
{
"epoch": 4.94111473880597,
"grad_norm": 0.6095589249221928,
"learning_rate": 5.017061741782833e-06,
"loss": 0.2588,
"step": 42375
},
{
"epoch": 4.94169776119403,
"grad_norm": 0.5800286959569063,
"learning_rate": 5.016725599722889e-06,
"loss": 0.2872,
"step": 42380
},
{
"epoch": 4.942280783582089,
"grad_norm": 0.5277976550400842,
"learning_rate": 5.016392800710434e-06,
"loss": 0.2592,
"step": 42385
},
{
"epoch": 4.942863805970149,
"grad_norm": 0.4926808393479529,
"learning_rate": 5.016063344794947e-06,
"loss": 0.2586,
"step": 42390
},
{
"epoch": 4.9434468283582085,
"grad_norm": 0.5243174413474881,
"learning_rate": 5.015737232025418e-06,
"loss": 0.2509,
"step": 42395
},
{
"epoch": 4.9440298507462686,
"grad_norm": 0.5444091475374001,
"learning_rate": 5.0154144624503365e-06,
"loss": 0.2645,
"step": 42400
},
{
"epoch": 4.944612873134329,
"grad_norm": 0.5396246731925429,
"learning_rate": 5.015095036117697e-06,
"loss": 0.2712,
"step": 42405
},
{
"epoch": 4.945195895522388,
"grad_norm": 0.5737818317841215,
"learning_rate": 5.014778953074992e-06,
"loss": 0.2674,
"step": 42410
},
{
"epoch": 4.945778917910448,
"grad_norm": 0.5765872960092024,
"learning_rate": 5.014466213369223e-06,
"loss": 0.2559,
"step": 42415
},
{
"epoch": 4.946361940298507,
"grad_norm": 0.5236453221462053,
"learning_rate": 5.014156817046891e-06,
"loss": 0.2715,
"step": 42420
},
{
"epoch": 4.946944962686567,
"grad_norm": 0.5116473112764134,
"learning_rate": 5.013850764153996e-06,
"loss": 0.2594,
"step": 42425
},
{
"epoch": 4.947527985074627,
"grad_norm": 0.5441658732781992,
"learning_rate": 5.013548054736049e-06,
"loss": 0.2743,
"step": 42430
},
{
"epoch": 4.948111007462686,
"grad_norm": 0.5245587783681419,
"learning_rate": 5.013248688838061e-06,
"loss": 0.253,
"step": 42435
},
{
"epoch": 4.9486940298507465,
"grad_norm": 0.5122101874152416,
"learning_rate": 5.012952666504542e-06,
"loss": 0.2705,
"step": 42440
},
{
"epoch": 4.949277052238806,
"grad_norm": 0.5570048059800029,
"learning_rate": 5.012659987779512e-06,
"loss": 0.2562,
"step": 42445
},
{
"epoch": 4.949860074626866,
"grad_norm": 0.5608195747463256,
"learning_rate": 5.012370652706484e-06,
"loss": 0.2738,
"step": 42450
},
{
"epoch": 4.950443097014926,
"grad_norm": 0.49721929201135545,
"learning_rate": 5.012084661328482e-06,
"loss": 0.2539,
"step": 42455
},
{
"epoch": 4.951026119402985,
"grad_norm": 0.5130788204130693,
"learning_rate": 5.011802013688029e-06,
"loss": 0.2817,
"step": 42460
},
{
"epoch": 4.951609141791045,
"grad_norm": 0.5395718479680132,
"learning_rate": 5.011522709827154e-06,
"loss": 0.2619,
"step": 42465
},
{
"epoch": 4.952192164179104,
"grad_norm": 0.5558579165824954,
"learning_rate": 5.011246749787385e-06,
"loss": 0.2768,
"step": 42470
},
{
"epoch": 4.952775186567164,
"grad_norm": 0.48807134518878653,
"learning_rate": 5.010974133609758e-06,
"loss": 0.2497,
"step": 42475
},
{
"epoch": 4.9533582089552235,
"grad_norm": 0.5410353509505209,
"learning_rate": 5.010704861334803e-06,
"loss": 0.2462,
"step": 42480
},
{
"epoch": 4.953941231343284,
"grad_norm": 0.5297584301230719,
"learning_rate": 5.010438933002563e-06,
"loss": 0.2772,
"step": 42485
},
{
"epoch": 4.954524253731344,
"grad_norm": 0.4885389228572895,
"learning_rate": 5.010176348652576e-06,
"loss": 0.2618,
"step": 42490
},
{
"epoch": 4.955107276119403,
"grad_norm": 0.5265683653471985,
"learning_rate": 5.009917108323885e-06,
"loss": 0.2589,
"step": 42495
},
{
"epoch": 4.955690298507463,
"grad_norm": 0.5488495005880858,
"learning_rate": 5.0096612120550436e-06,
"loss": 0.2698,
"step": 42500
},
{
"epoch": 4.956273320895522,
"grad_norm": 0.500672598203943,
"learning_rate": 5.009408659884092e-06,
"loss": 0.2725,
"step": 42505
},
{
"epoch": 4.956856343283582,
"grad_norm": 0.5649699889084325,
"learning_rate": 5.009159451848587e-06,
"loss": 0.2475,
"step": 42510
},
{
"epoch": 4.957439365671641,
"grad_norm": 0.5068588579455506,
"learning_rate": 5.008913587985581e-06,
"loss": 0.2531,
"step": 42515
},
{
"epoch": 4.958022388059701,
"grad_norm": 0.4904332591542818,
"learning_rate": 5.008671068331634e-06,
"loss": 0.2752,
"step": 42520
},
{
"epoch": 4.9586054104477615,
"grad_norm": 0.5292395690628368,
"learning_rate": 5.008431892922808e-06,
"loss": 0.2569,
"step": 42525
},
{
"epoch": 4.959188432835821,
"grad_norm": 0.5162847077380787,
"learning_rate": 5.00819606179466e-06,
"loss": 0.2517,
"step": 42530
},
{
"epoch": 4.959771455223881,
"grad_norm": 0.511489903051284,
"learning_rate": 5.007963574982264e-06,
"loss": 0.2658,
"step": 42535
},
{
"epoch": 4.96035447761194,
"grad_norm": 0.5597316656999656,
"learning_rate": 5.007734432520179e-06,
"loss": 0.2874,
"step": 42540
},
{
"epoch": 4.9609375,
"grad_norm": 0.5351076117828275,
"learning_rate": 5.0075086344424855e-06,
"loss": 0.2746,
"step": 42545
},
{
"epoch": 4.96152052238806,
"grad_norm": 0.49566378784578796,
"learning_rate": 5.0072861807827505e-06,
"loss": 0.2522,
"step": 42550
},
{
"epoch": 4.962103544776119,
"grad_norm": 0.5434957468307663,
"learning_rate": 5.007067071574053e-06,
"loss": 0.2583,
"step": 42555
},
{
"epoch": 4.962686567164179,
"grad_norm": 0.5753697977872646,
"learning_rate": 5.0068513068489765e-06,
"loss": 0.2733,
"step": 42560
},
{
"epoch": 4.9632695895522385,
"grad_norm": 0.5272924317732932,
"learning_rate": 5.006638886639597e-06,
"loss": 0.2576,
"step": 42565
},
{
"epoch": 4.963852611940299,
"grad_norm": 0.5191068335476884,
"learning_rate": 5.0064298109775035e-06,
"loss": 0.2493,
"step": 42570
},
{
"epoch": 4.964435634328359,
"grad_norm": 0.5195820566025755,
"learning_rate": 5.00622407989378e-06,
"loss": 0.2593,
"step": 42575
},
{
"epoch": 4.965018656716418,
"grad_norm": 0.5156457041364598,
"learning_rate": 5.006021693419021e-06,
"loss": 0.2662,
"step": 42580
},
{
"epoch": 4.965601679104478,
"grad_norm": 0.49984629226854865,
"learning_rate": 5.005822651583317e-06,
"loss": 0.2709,
"step": 42585
},
{
"epoch": 4.966184701492537,
"grad_norm": 0.5638317611211411,
"learning_rate": 5.0056269544162635e-06,
"loss": 0.2483,
"step": 42590
},
{
"epoch": 4.966767723880597,
"grad_norm": 0.5253048930761857,
"learning_rate": 5.005434601946959e-06,
"loss": 0.2757,
"step": 42595
},
{
"epoch": 4.967350746268656,
"grad_norm": 0.5406929590852747,
"learning_rate": 5.0052455942040045e-06,
"loss": 0.2537,
"step": 42600
},
{
"epoch": 4.967933768656716,
"grad_norm": 0.5350349817512208,
"learning_rate": 5.005059931215503e-06,
"loss": 0.2689,
"step": 42605
},
{
"epoch": 4.9685167910447765,
"grad_norm": 0.5309767938037251,
"learning_rate": 5.004877613009064e-06,
"loss": 0.2651,
"step": 42610
},
{
"epoch": 4.969099813432836,
"grad_norm": 0.5355737554112076,
"learning_rate": 5.004698639611792e-06,
"loss": 0.253,
"step": 42615
},
{
"epoch": 4.969682835820896,
"grad_norm": 0.5290801600001703,
"learning_rate": 5.0045230110503e-06,
"loss": 0.2555,
"step": 42620
},
{
"epoch": 4.970265858208955,
"grad_norm": 0.4998835620223301,
"learning_rate": 5.004350727350704e-06,
"loss": 0.2478,
"step": 42625
},
{
"epoch": 4.970848880597015,
"grad_norm": 0.5661328413618032,
"learning_rate": 5.00418178853862e-06,
"loss": 0.2726,
"step": 42630
},
{
"epoch": 4.971431902985074,
"grad_norm": 0.5681877161289965,
"learning_rate": 5.004016194639169e-06,
"loss": 0.2646,
"step": 42635
},
{
"epoch": 4.972014925373134,
"grad_norm": 0.5706374125383276,
"learning_rate": 5.003853945676969e-06,
"loss": 0.2676,
"step": 42640
},
{
"epoch": 4.972597947761194,
"grad_norm": 0.5537135888616044,
"learning_rate": 5.0036950416761485e-06,
"loss": 0.2819,
"step": 42645
},
{
"epoch": 4.9731809701492535,
"grad_norm": 0.5217565278454588,
"learning_rate": 5.0035394826603345e-06,
"loss": 0.2539,
"step": 42650
},
{
"epoch": 4.973763992537314,
"grad_norm": 0.5104889223282257,
"learning_rate": 5.003387268652657e-06,
"loss": 0.25,
"step": 42655
},
{
"epoch": 4.974347014925373,
"grad_norm": 0.535969910630488,
"learning_rate": 5.003238399675746e-06,
"loss": 0.2599,
"step": 42660
},
{
"epoch": 4.974930037313433,
"grad_norm": 0.513665049975225,
"learning_rate": 5.003092875751742e-06,
"loss": 0.2537,
"step": 42665
},
{
"epoch": 4.975513059701493,
"grad_norm": 0.5649291292140013,
"learning_rate": 5.002950696902278e-06,
"loss": 0.2632,
"step": 42670
},
{
"epoch": 4.976096082089552,
"grad_norm": 0.507554516214919,
"learning_rate": 5.0028118631485e-06,
"loss": 0.2526,
"step": 42675
},
{
"epoch": 4.976679104477612,
"grad_norm": 0.4737579168132423,
"learning_rate": 5.002676374511046e-06,
"loss": 0.251,
"step": 42680
},
{
"epoch": 4.977262126865671,
"grad_norm": 0.47485280932488766,
"learning_rate": 5.002544231010064e-06,
"loss": 0.2561,
"step": 42685
},
{
"epoch": 4.9778451492537314,
"grad_norm": 0.5233764045395329,
"learning_rate": 5.0024154326652044e-06,
"loss": 0.2726,
"step": 42690
},
{
"epoch": 4.9784281716417915,
"grad_norm": 0.5136717520620079,
"learning_rate": 5.002289979495614e-06,
"loss": 0.2532,
"step": 42695
},
{
"epoch": 4.979011194029851,
"grad_norm": 0.5438972770455104,
"learning_rate": 5.002167871519951e-06,
"loss": 0.2627,
"step": 42700
},
{
"epoch": 4.979594216417911,
"grad_norm": 0.5374919797374826,
"learning_rate": 5.00204910875637e-06,
"loss": 0.2637,
"step": 42705
},
{
"epoch": 4.98017723880597,
"grad_norm": 0.5618090151137566,
"learning_rate": 5.001933691222527e-06,
"loss": 0.2704,
"step": 42710
},
{
"epoch": 4.98076026119403,
"grad_norm": 0.5400968553646867,
"learning_rate": 5.001821618935589e-06,
"loss": 0.2656,
"step": 42715
},
{
"epoch": 4.981343283582089,
"grad_norm": 0.5356939870669652,
"learning_rate": 5.001712891912217e-06,
"loss": 0.2746,
"step": 42720
},
{
"epoch": 4.981926305970149,
"grad_norm": 0.5589525144895989,
"learning_rate": 5.001607510168576e-06,
"loss": 0.253,
"step": 42725
},
{
"epoch": 4.9825093283582085,
"grad_norm": 0.51423787731282,
"learning_rate": 5.001505473720337e-06,
"loss": 0.2588,
"step": 42730
},
{
"epoch": 4.9830923507462686,
"grad_norm": 0.5416516453895129,
"learning_rate": 5.001406782582673e-06,
"loss": 0.2767,
"step": 42735
},
{
"epoch": 4.983675373134329,
"grad_norm": 0.5138911139053113,
"learning_rate": 5.001311436770255e-06,
"loss": 0.2645,
"step": 42740
},
{
"epoch": 4.984258395522388,
"grad_norm": 0.5367203943984868,
"learning_rate": 5.001219436297262e-06,
"loss": 0.2568,
"step": 42745
},
{
"epoch": 4.984841417910448,
"grad_norm": 0.5501690738444324,
"learning_rate": 5.001130781177377e-06,
"loss": 0.2685,
"step": 42750
},
{
"epoch": 4.985424440298507,
"grad_norm": 0.5373366414628004,
"learning_rate": 5.0010454714237786e-06,
"loss": 0.2662,
"step": 42755
},
{
"epoch": 4.986007462686567,
"grad_norm": 0.5347366551370059,
"learning_rate": 5.000963507049151e-06,
"loss": 0.266,
"step": 42760
},
{
"epoch": 4.986590485074627,
"grad_norm": 0.546458556108298,
"learning_rate": 5.000884888065682e-06,
"loss": 0.2691,
"step": 42765
},
{
"epoch": 4.987173507462686,
"grad_norm": 0.5032987363405463,
"learning_rate": 5.000809614485062e-06,
"loss": 0.2651,
"step": 42770
},
{
"epoch": 4.9877565298507465,
"grad_norm": 0.5424532656947771,
"learning_rate": 5.0007376863184835e-06,
"loss": 0.2492,
"step": 42775
},
{
"epoch": 4.988339552238806,
"grad_norm": 0.5043402438100439,
"learning_rate": 5.000669103576643e-06,
"loss": 0.2589,
"step": 42780
},
{
"epoch": 4.988922574626866,
"grad_norm": 0.5343459268973346,
"learning_rate": 5.000603866269734e-06,
"loss": 0.2615,
"step": 42785
},
{
"epoch": 4.989505597014926,
"grad_norm": 0.5689999173772574,
"learning_rate": 5.000541974407462e-06,
"loss": 0.2596,
"step": 42790
},
{
"epoch": 4.990088619402985,
"grad_norm": 0.5861951582292237,
"learning_rate": 5.0004834279990245e-06,
"loss": 0.2659,
"step": 42795
},
{
"epoch": 4.990671641791045,
"grad_norm": 0.5023106712695729,
"learning_rate": 5.000428227053131e-06,
"loss": 0.26,
"step": 42800
},
{
"epoch": 4.991254664179104,
"grad_norm": 0.5257901894197187,
"learning_rate": 5.000376371577987e-06,
"loss": 0.2604,
"step": 42805
},
{
"epoch": 4.991837686567164,
"grad_norm": 0.4763993165887568,
"learning_rate": 5.000327861581302e-06,
"loss": 0.2486,
"step": 42810
},
{
"epoch": 4.9924207089552235,
"grad_norm": 0.5167447453038132,
"learning_rate": 5.000282697070291e-06,
"loss": 0.2747,
"step": 42815
},
{
"epoch": 4.993003731343284,
"grad_norm": 0.4866714281289037,
"learning_rate": 5.000240878051671e-06,
"loss": 0.2499,
"step": 42820
},
{
"epoch": 4.993586753731344,
"grad_norm": 0.5005658570867595,
"learning_rate": 5.000202404531656e-06,
"loss": 0.2503,
"step": 42825
},
{
"epoch": 4.994169776119403,
"grad_norm": 0.5380777958946619,
"learning_rate": 5.0001672765159696e-06,
"loss": 0.2529,
"step": 42830
},
{
"epoch": 4.994752798507463,
"grad_norm": 0.5187253248059627,
"learning_rate": 5.000135494009835e-06,
"loss": 0.2748,
"step": 42835
},
{
"epoch": 4.995335820895522,
"grad_norm": 0.5165834922078414,
"learning_rate": 5.000107057017976e-06,
"loss": 0.2754,
"step": 42840
},
{
"epoch": 4.995918843283582,
"grad_norm": 0.5600181004782937,
"learning_rate": 5.000081965544622e-06,
"loss": 0.2637,
"step": 42845
},
{
"epoch": 4.996501865671641,
"grad_norm": 0.5397015314691016,
"learning_rate": 5.0000602195935046e-06,
"loss": 0.2614,
"step": 42850
},
{
"epoch": 4.997084888059701,
"grad_norm": 0.4935744319945381,
"learning_rate": 5.000041819167857e-06,
"loss": 0.252,
"step": 42855
},
{
"epoch": 4.9976679104477615,
"grad_norm": 0.5591119193292369,
"learning_rate": 5.000026764270413e-06,
"loss": 0.2561,
"step": 42860
},
{
"epoch": 4.998250932835821,
"grad_norm": 0.5626430914617118,
"learning_rate": 5.000015054903415e-06,
"loss": 0.2765,
"step": 42865
},
{
"epoch": 4.998833955223881,
"grad_norm": 0.5479808963178789,
"learning_rate": 5.0000066910686e-06,
"loss": 0.2616,
"step": 42870
},
{
"epoch": 4.99941697761194,
"grad_norm": 0.4935988139893575,
"learning_rate": 5.0000016727672125e-06,
"loss": 0.2403,
"step": 42875
},
{
"epoch": 5.0,
"grad_norm": 1.2398633217504624,
"learning_rate": 5e-06,
"loss": 0.2729,
"step": 42880
},
{
"epoch": 5.0,
"step": 42880,
"total_flos": 2442892589137920.0,
"train_loss": 0.2958939010054985,
"train_runtime": 111352.539,
"train_samples_per_second": 1.54,
"train_steps_per_second": 0.385
}
],
"logging_steps": 5,
"max_steps": 42880,
"num_input_tokens_seen": 0,
"num_train_epochs": 5,
"save_steps": 100,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 2442892589137920.0,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}