{
  "best_global_step": 4000,
  "best_metric": 2.1779719650083877,
  "best_model_checkpoint": "./SALAMA_C4/checkpoint-4000",
  "epoch": 2.7229407760381212,
  "eval_steps": 2000,
  "global_step": 4000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.013614703880190605,
      "grad_norm": 6.790204048156738,
      "learning_rate": 6.333333333333334e-07,
      "loss": 0.1302,
      "step": 20
    },
    {
      "epoch": 0.02722940776038121,
      "grad_norm": 6.674442768096924,
      "learning_rate": 1.3e-06,
      "loss": 0.1286,
      "step": 40
    },
    {
      "epoch": 0.04084411164057182,
      "grad_norm": 6.946224689483643,
      "learning_rate": 1.9666666666666668e-06,
      "loss": 0.1394,
      "step": 60
    },
    {
      "epoch": 0.05445881552076242,
      "grad_norm": 8.382658004760742,
      "learning_rate": 2.6333333333333332e-06,
      "loss": 0.1267,
      "step": 80
    },
    {
      "epoch": 0.06807351940095303,
      "grad_norm": 5.505548477172852,
      "learning_rate": 3.3000000000000006e-06,
      "loss": 0.1,
      "step": 100
    },
    {
      "epoch": 0.08168822328114364,
      "grad_norm": 3.323385715484619,
      "learning_rate": 3.966666666666667e-06,
      "loss": 0.0979,
      "step": 120
    },
    {
      "epoch": 0.09530292716133425,
      "grad_norm": 3.4979546070098877,
      "learning_rate": 4.633333333333334e-06,
      "loss": 0.0934,
      "step": 140
    },
    {
      "epoch": 0.10891763104152484,
      "grad_norm": 5.154605865478516,
      "learning_rate": 5.300000000000001e-06,
      "loss": 0.1039,
      "step": 160
    },
    {
      "epoch": 0.12253233492171545,
      "grad_norm": 5.488887786865234,
      "learning_rate": 5.966666666666667e-06,
      "loss": 0.1029,
      "step": 180
    },
    {
      "epoch": 0.13614703880190607,
      "grad_norm": 3.0190138816833496,
      "learning_rate": 6.633333333333334e-06,
      "loss": 0.1151,
      "step": 200
    },
    {
      "epoch": 0.14976174268209666,
      "grad_norm": 4.113169193267822,
      "learning_rate": 7.3e-06,
      "loss": 0.0936,
      "step": 220
    },
    {
      "epoch": 0.16337644656228728,
      "grad_norm": 6.0397515296936035,
      "learning_rate": 7.966666666666668e-06,
      "loss": 0.1214,
      "step": 240
    },
    {
      "epoch": 0.17699115044247787,
      "grad_norm": 6.033300399780273,
      "learning_rate": 8.633333333333334e-06,
      "loss": 0.1153,
      "step": 260
    },
    {
      "epoch": 0.1906058543226685,
      "grad_norm": 3.8764054775238037,
      "learning_rate": 9.3e-06,
      "loss": 0.1025,
      "step": 280
    },
    {
      "epoch": 0.2042205582028591,
      "grad_norm": 5.634373188018799,
      "learning_rate": 9.966666666666667e-06,
      "loss": 0.12,
      "step": 300
    },
    {
      "epoch": 0.21783526208304968,
      "grad_norm": 4.584943771362305,
      "learning_rate": 9.973030518097942e-06,
      "loss": 0.1099,
      "step": 320
    },
    {
      "epoch": 0.2314499659632403,
      "grad_norm": 5.086080551147461,
      "learning_rate": 9.944641589779987e-06,
      "loss": 0.1267,
      "step": 340
    },
    {
      "epoch": 0.2450646698434309,
      "grad_norm": 5.2667741775512695,
      "learning_rate": 9.91625266146203e-06,
      "loss": 0.1059,
      "step": 360
    },
    {
      "epoch": 0.2586793737236215,
      "grad_norm": 5.114360809326172,
      "learning_rate": 9.887863733144075e-06,
      "loss": 0.1294,
      "step": 380
    },
    {
      "epoch": 0.27229407760381213,
      "grad_norm": 6.17408561706543,
      "learning_rate": 9.859474804826119e-06,
      "loss": 0.12,
      "step": 400
    },
    {
      "epoch": 0.2859087814840027,
      "grad_norm": 4.034195899963379,
      "learning_rate": 9.831085876508163e-06,
      "loss": 0.1082,
      "step": 420
    },
    {
      "epoch": 0.2995234853641933,
      "grad_norm": 4.708987236022949,
      "learning_rate": 9.802696948190207e-06,
      "loss": 0.1041,
      "step": 440
    },
    {
      "epoch": 0.3131381892443839,
      "grad_norm": 3.7173264026641846,
      "learning_rate": 9.774308019872251e-06,
      "loss": 0.115,
      "step": 460
    },
    {
      "epoch": 0.32675289312457456,
      "grad_norm": 4.3770647048950195,
      "learning_rate": 9.745919091554295e-06,
      "loss": 0.1202,
      "step": 480
    },
    {
      "epoch": 0.34036759700476515,
      "grad_norm": 3.7374465465545654,
      "learning_rate": 9.717530163236339e-06,
      "loss": 0.1008,
      "step": 500
    },
    {
      "epoch": 0.35398230088495575,
      "grad_norm": 3.801574945449829,
      "learning_rate": 9.689141234918381e-06,
      "loss": 0.1171,
      "step": 520
    },
    {
      "epoch": 0.36759700476514634,
      "grad_norm": 3.962735176086426,
      "learning_rate": 9.660752306600427e-06,
      "loss": 0.0997,
      "step": 540
    },
    {
      "epoch": 0.381211708645337,
      "grad_norm": 4.948540687561035,
      "learning_rate": 9.632363378282471e-06,
      "loss": 0.1304,
      "step": 560
    },
    {
      "epoch": 0.3948264125255276,
      "grad_norm": 5.182698726654053,
      "learning_rate": 9.603974449964515e-06,
      "loss": 0.1059,
      "step": 580
    },
    {
      "epoch": 0.4084411164057182,
      "grad_norm": 5.27463436126709,
      "learning_rate": 9.57558552164656e-06,
      "loss": 0.1379,
      "step": 600
    },
    {
      "epoch": 0.42205582028590877,
      "grad_norm": 4.754587650299072,
      "learning_rate": 9.547196593328602e-06,
      "loss": 0.1135,
      "step": 620
    },
    {
      "epoch": 0.43567052416609936,
      "grad_norm": 4.032769680023193,
      "learning_rate": 9.518807665010646e-06,
      "loss": 0.0925,
      "step": 640
    },
    {
      "epoch": 0.44928522804629,
      "grad_norm": Infinity,
      "learning_rate": 9.49041873669269e-06,
      "loss": 0.1764,
      "step": 660
    },
    {
      "epoch": 0.4628999319264806,
      "grad_norm": 12.64933967590332,
      "learning_rate": 9.467707594038326e-06,
      "loss": 0.5231,
      "step": 680
    },
    {
      "epoch": 0.4765146358066712,
      "grad_norm": 16.499448776245117,
      "learning_rate": 9.43931866572037e-06,
      "loss": 0.1525,
      "step": 700
    },
    {
      "epoch": 0.4901293396868618,
      "grad_norm": 2.7940587997436523,
      "learning_rate": 9.410929737402414e-06,
      "loss": 0.1635,
      "step": 720
    },
    {
      "epoch": 0.5037440435670524,
      "grad_norm": 5.6086039543151855,
      "learning_rate": 9.382540809084458e-06,
      "loss": 0.1177,
      "step": 740
    },
    {
      "epoch": 0.517358747447243,
      "grad_norm": 5.121674537658691,
      "learning_rate": 9.354151880766502e-06,
      "loss": 0.1185,
      "step": 760
    },
    {
      "epoch": 0.5309734513274337,
      "grad_norm": 5.849049091339111,
      "learning_rate": 9.325762952448546e-06,
      "loss": 0.1358,
      "step": 780
    },
    {
      "epoch": 0.5445881552076243,
      "grad_norm": 4.180930137634277,
      "learning_rate": 9.29737402413059e-06,
      "loss": 0.1192,
      "step": 800
    },
    {
      "epoch": 0.5582028590878149,
      "grad_norm": 5.0372796058654785,
      "learning_rate": 9.268985095812634e-06,
      "loss": 0.1194,
      "step": 820
    },
    {
      "epoch": 0.5718175629680055,
      "grad_norm": 3.9500577449798584,
      "learning_rate": 9.240596167494677e-06,
      "loss": 0.1026,
      "step": 840
    },
    {
      "epoch": 0.585432266848196,
      "grad_norm": 5.2215800285339355,
      "learning_rate": 9.212207239176721e-06,
      "loss": 0.1227,
      "step": 860
    },
    {
      "epoch": 0.5990469707283866,
      "grad_norm": 7.128079414367676,
      "learning_rate": 9.183818310858765e-06,
      "loss": 0.1296,
      "step": 880
    },
    {
      "epoch": 0.6126616746085772,
      "grad_norm": 4.189157485961914,
      "learning_rate": 9.15542938254081e-06,
      "loss": 0.1092,
      "step": 900
    },
    {
      "epoch": 0.6262763784887678,
      "grad_norm": 4.203508377075195,
      "learning_rate": 9.127040454222855e-06,
      "loss": 0.1067,
      "step": 920
    },
    {
      "epoch": 0.6398910823689585,
      "grad_norm": 5.561427593231201,
      "learning_rate": 9.098651525904897e-06,
      "loss": 0.1156,
      "step": 940
    },
    {
      "epoch": 0.6535057862491491,
      "grad_norm": 2.9789481163024902,
      "learning_rate": 9.070262597586941e-06,
      "loss": 0.1144,
      "step": 960
    },
    {
      "epoch": 0.6671204901293397,
      "grad_norm": 6.220657825469971,
      "learning_rate": 9.041873669268985e-06,
      "loss": 0.121,
      "step": 980
    },
    {
      "epoch": 0.6807351940095303,
      "grad_norm": 5.125210285186768,
      "learning_rate": 9.01348474095103e-06,
      "loss": 0.1073,
      "step": 1000
    },
    {
      "epoch": 0.6943498978897209,
      "grad_norm": 4.665761947631836,
      "learning_rate": 8.985095812633073e-06,
      "loss": 0.1032,
      "step": 1020
    },
    {
      "epoch": 0.7079646017699115,
      "grad_norm": 3.9673075675964355,
      "learning_rate": 8.956706884315118e-06,
      "loss": 0.1133,
      "step": 1040
    },
    {
      "epoch": 0.7215793056501021,
      "grad_norm": 4.1961140632629395,
      "learning_rate": 8.928317955997162e-06,
      "loss": 0.1008,
      "step": 1060
    },
    {
      "epoch": 0.7351940095302927,
      "grad_norm": 6.15500545501709,
      "learning_rate": 8.899929027679206e-06,
      "loss": 0.1041,
      "step": 1080
    },
    {
      "epoch": 0.7488087134104833,
      "grad_norm": 4.683866024017334,
      "learning_rate": 8.87154009936125e-06,
      "loss": 0.1086,
      "step": 1100
    },
    {
      "epoch": 0.762423417290674,
      "grad_norm": 8.679821968078613,
      "learning_rate": 8.843151171043294e-06,
      "loss": 0.1049,
      "step": 1120
    },
    {
      "epoch": 0.7760381211708646,
      "grad_norm": 4.2732038497924805,
      "learning_rate": 8.814762242725338e-06,
      "loss": 0.1013,
      "step": 1140
    },
    {
      "epoch": 0.7896528250510552,
      "grad_norm": 9.020421028137207,
      "learning_rate": 8.786373314407382e-06,
      "loss": 0.101,
      "step": 1160
    },
    {
      "epoch": 0.8032675289312458,
      "grad_norm": 6.960070610046387,
      "learning_rate": 8.757984386089426e-06,
      "loss": 0.1067,
      "step": 1180
    },
    {
      "epoch": 0.8168822328114363,
      "grad_norm": 3.1116604804992676,
      "learning_rate": 8.72959545777147e-06,
      "loss": 0.099,
      "step": 1200
    },
    {
      "epoch": 0.8304969366916269,
      "grad_norm": 3.949913740158081,
      "learning_rate": 8.701206529453514e-06,
      "loss": 0.1068,
      "step": 1220
    },
    {
      "epoch": 0.8441116405718175,
      "grad_norm": 4.617140293121338,
      "learning_rate": 8.672817601135558e-06,
      "loss": 0.1237,
      "step": 1240
    },
    {
      "epoch": 0.8577263444520081,
      "grad_norm": 6.639930248260498,
      "learning_rate": 8.644428672817602e-06,
      "loss": 0.1295,
      "step": 1260
    },
    {
      "epoch": 0.8713410483321987,
      "grad_norm": 5.353048324584961,
      "learning_rate": 8.616039744499646e-06,
      "loss": 0.1088,
      "step": 1280
    },
    {
      "epoch": 0.8849557522123894,
      "grad_norm": 7.117270469665527,
      "learning_rate": 8.587650816181689e-06,
      "loss": 0.112,
      "step": 1300
    },
    {
      "epoch": 0.89857045609258,
      "grad_norm": 5.29574728012085,
      "learning_rate": 8.559261887863733e-06,
      "loss": 0.1188,
      "step": 1320
    },
    {
      "epoch": 0.9121851599727706,
      "grad_norm": 7.008860111236572,
      "learning_rate": 8.530872959545777e-06,
      "loss": 0.1076,
      "step": 1340
    },
    {
      "epoch": 0.9257998638529612,
      "grad_norm": 5.290408134460449,
      "learning_rate": 8.502484031227823e-06,
      "loss": 0.1339,
      "step": 1360
    },
    {
      "epoch": 0.9394145677331518,
      "grad_norm": 4.095961093902588,
      "learning_rate": 8.474095102909867e-06,
      "loss": 0.111,
      "step": 1380
    },
    {
      "epoch": 0.9530292716133424,
      "grad_norm": 4.4582085609436035,
      "learning_rate": 8.44570617459191e-06,
      "loss": 0.1079,
      "step": 1400
    },
    {
      "epoch": 0.966643975493533,
      "grad_norm": 3.397148370742798,
      "learning_rate": 8.417317246273953e-06,
      "loss": 0.0989,
      "step": 1420
    },
    {
      "epoch": 0.9802586793737236,
      "grad_norm": 4.60284948348999,
      "learning_rate": 8.388928317955997e-06,
      "loss": 0.1205,
      "step": 1440
    },
    {
      "epoch": 0.9938733832539143,
      "grad_norm": 4.263270378112793,
      "learning_rate": 8.360539389638041e-06,
      "loss": 0.0981,
      "step": 1460
    },
    {
      "epoch": 1.0074880871341048,
      "grad_norm": 2.734585762023926,
      "learning_rate": 8.332150461320085e-06,
      "loss": 0.0687,
      "step": 1480
    },
    {
      "epoch": 1.0211027910142954,
      "grad_norm": 4.107769966125488,
      "learning_rate": 8.30376153300213e-06,
      "loss": 0.0483,
      "step": 1500
    },
    {
      "epoch": 1.034717494894486,
      "grad_norm": 3.7158114910125732,
      "learning_rate": 8.275372604684174e-06,
      "loss": 0.0523,
      "step": 1520
    },
    {
      "epoch": 1.0483321987746765,
      "grad_norm": 3.725184917449951,
      "learning_rate": 8.246983676366218e-06,
      "loss": 0.0438,
      "step": 1540
    },
    {
      "epoch": 1.0619469026548674,
      "grad_norm": 2.794508218765259,
      "learning_rate": 8.218594748048262e-06,
      "loss": 0.0534,
      "step": 1560
    },
    {
      "epoch": 1.075561606535058,
      "grad_norm": 3.6485531330108643,
      "learning_rate": 8.190205819730306e-06,
      "loss": 0.0443,
      "step": 1580
    },
    {
      "epoch": 1.0891763104152485,
      "grad_norm": 2.9312281608581543,
      "learning_rate": 8.16181689141235e-06,
      "loss": 0.0426,
      "step": 1600
    },
    {
      "epoch": 1.1027910142954391,
      "grad_norm": 3.66829514503479,
      "learning_rate": 8.133427963094394e-06,
      "loss": 0.0558,
      "step": 1620
    },
    {
      "epoch": 1.1164057181756297,
      "grad_norm": 2.831806182861328,
      "learning_rate": 8.105039034776438e-06,
      "loss": 0.0504,
      "step": 1640
    },
    {
      "epoch": 1.1300204220558203,
      "grad_norm": 4.0393171310424805,
      "learning_rate": 8.076650106458482e-06,
      "loss": 0.0447,
      "step": 1660
    },
    {
      "epoch": 1.143635125936011,
      "grad_norm": 3.919926643371582,
      "learning_rate": 8.048261178140526e-06,
      "loss": 0.0432,
      "step": 1680
    },
    {
      "epoch": 1.1572498298162015,
      "grad_norm": 2.3747353553771973,
      "learning_rate": 8.01987224982257e-06,
      "loss": 0.0495,
      "step": 1700
    },
    {
      "epoch": 1.170864533696392,
      "grad_norm": 2.7838926315307617,
      "learning_rate": 7.991483321504614e-06,
      "loss": 0.0477,
      "step": 1720
    },
    {
      "epoch": 1.1844792375765827,
      "grad_norm": 4.423709869384766,
      "learning_rate": 7.963094393186658e-06,
      "loss": 0.0511,
      "step": 1740
    },
    {
      "epoch": 1.1980939414567733,
      "grad_norm": 3.6399760246276855,
      "learning_rate": 7.9347054648687e-06,
      "loss": 0.0512,
      "step": 1760
    },
    {
      "epoch": 1.2117086453369639,
      "grad_norm": 4.124867916107178,
      "learning_rate": 7.906316536550745e-06,
      "loss": 0.0503,
      "step": 1780
    },
    {
      "epoch": 1.2253233492171545,
      "grad_norm": 2.90708589553833,
      "learning_rate": 7.877927608232789e-06,
      "loss": 0.0428,
      "step": 1800
    },
    {
      "epoch": 1.238938053097345,
      "grad_norm": 3.0257723331451416,
      "learning_rate": 7.849538679914835e-06,
      "loss": 0.0505,
      "step": 1820
    },
    {
      "epoch": 1.2525527569775359,
      "grad_norm": 3.9913330078125,
      "learning_rate": 7.821149751596879e-06,
      "loss": 0.051,
      "step": 1840
    },
    {
      "epoch": 1.2661674608577265,
      "grad_norm": 3.413067102432251,
      "learning_rate": 7.792760823278921e-06,
      "loss": 0.0566,
      "step": 1860
    },
    {
      "epoch": 1.279782164737917,
      "grad_norm": 2.3691389560699463,
      "learning_rate": 7.764371894960965e-06,
      "loss": 0.0387,
      "step": 1880
    },
    {
      "epoch": 1.2933968686181077,
      "grad_norm": 2.782249689102173,
      "learning_rate": 7.73598296664301e-06,
      "loss": 0.0548,
      "step": 1900
    },
    {
      "epoch": 1.3070115724982982,
      "grad_norm": 2.8736960887908936,
      "learning_rate": 7.707594038325053e-06,
      "loss": 0.0512,
      "step": 1920
    },
    {
      "epoch": 1.3206262763784888,
      "grad_norm": 2.5277926921844482,
      "learning_rate": 7.679205110007098e-06,
      "loss": 0.0449,
      "step": 1940
    },
    {
      "epoch": 1.3342409802586794,
      "grad_norm": 3.7290806770324707,
      "learning_rate": 7.650816181689142e-06,
      "loss": 0.052,
      "step": 1960
    },
    {
      "epoch": 1.34785568413887,
      "grad_norm": 3.4079151153564453,
      "learning_rate": 7.6224272533711865e-06,
      "loss": 0.0459,
      "step": 1980
    },
    {
      "epoch": 1.3614703880190606,
      "grad_norm": 2.7560651302337646,
      "learning_rate": 7.59403832505323e-06,
      "loss": 0.0536,
      "step": 2000
    },
    {
      "epoch": 1.3614703880190606,
      "eval_loss": 0.05250042304396629,
      "eval_runtime": 4762.58,
      "eval_samples_per_second": 2.467,
      "eval_steps_per_second": 0.308,
      "eval_wer": 3.935134725289306,
      "step": 2000
    },
    {
      "epoch": 1.3750850918992512,
      "grad_norm": 2.759082794189453,
      "learning_rate": 7.565649396735274e-06,
      "loss": 0.0465,
      "step": 2020
    },
    {
      "epoch": 1.3886997957794418,
      "grad_norm": 3.005037307739258,
      "learning_rate": 7.537260468417318e-06,
      "loss": 0.044,
      "step": 2040
    },
    {
      "epoch": 1.4023144996596324,
      "grad_norm": 1.479257345199585,
      "learning_rate": 7.508871540099362e-06,
      "loss": 0.0439,
      "step": 2060
    },
    {
      "epoch": 1.415929203539823,
      "grad_norm": 3.5328667163848877,
      "learning_rate": 7.480482611781405e-06,
      "loss": 0.0489,
      "step": 2080
    },
    {
      "epoch": 1.4295439074200136,
      "grad_norm": 4.166772842407227,
      "learning_rate": 7.452093683463449e-06,
      "loss": 0.047,
      "step": 2100
    },
    {
      "epoch": 1.4431586113002042,
      "grad_norm": 6.129347801208496,
      "learning_rate": 7.423704755145494e-06,
      "loss": 0.0417,
      "step": 2120
    },
    {
      "epoch": 1.4567733151803948,
      "grad_norm": 3.5741307735443115,
      "learning_rate": 7.395315826827538e-06,
      "loss": 0.0531,
      "step": 2140
    },
    {
      "epoch": 1.4703880190605854,
      "grad_norm": 2.9267029762268066,
      "learning_rate": 7.366926898509582e-06,
      "loss": 0.0413,
      "step": 2160
    },
    {
      "epoch": 1.484002722940776,
      "grad_norm": 5.829305648803711,
      "learning_rate": 7.3385379701916255e-06,
      "loss": 0.056,
      "step": 2180
    },
    {
      "epoch": 1.4976174268209665,
      "grad_norm": 5.405384540557861,
      "learning_rate": 7.31014904187367e-06,
      "loss": 0.0466,
      "step": 2200
    },
    {
      "epoch": 1.5112321307011571,
      "grad_norm": 2.835297107696533,
      "learning_rate": 7.281760113555714e-06,
      "loss": 0.0488,
      "step": 2220
    },
    {
      "epoch": 1.5248468345813477,
      "grad_norm": 3.75079607963562,
      "learning_rate": 7.253371185237758e-06,
      "loss": 0.0422,
      "step": 2240
    },
    {
      "epoch": 1.5384615384615383,
      "grad_norm": 3.934020519256592,
      "learning_rate": 7.224982256919801e-06,
      "loss": 0.0554,
      "step": 2260
    },
    {
      "epoch": 1.552076242341729,
      "grad_norm": 2.9037790298461914,
      "learning_rate": 7.196593328601847e-06,
      "loss": 0.0393,
      "step": 2280
    },
    {
      "epoch": 1.5656909462219195,
      "grad_norm": 3.191974639892578,
      "learning_rate": 7.16820440028389e-06,
      "loss": 0.0477,
      "step": 2300
    },
    {
      "epoch": 1.5793056501021103,
      "grad_norm": 3.6773743629455566,
      "learning_rate": 7.139815471965934e-06,
      "loss": 0.0443,
      "step": 2320
    },
    {
      "epoch": 1.592920353982301,
      "grad_norm": 3.3054301738739014,
      "learning_rate": 7.111426543647978e-06,
      "loss": 0.0516,
      "step": 2340
    },
    {
      "epoch": 1.6065350578624915,
      "grad_norm": 3.2623848915100098,
      "learning_rate": 7.083037615330021e-06,
      "loss": 0.0449,
      "step": 2360
    },
    {
      "epoch": 1.620149761742682,
      "grad_norm": 3.678145408630371,
      "learning_rate": 7.0546486870120654e-06,
      "loss": 0.0421,
      "step": 2380
    },
    {
      "epoch": 1.6337644656228727,
      "grad_norm": 1.8990530967712402,
      "learning_rate": 7.0262597586941095e-06,
      "loss": 0.0451,
      "step": 2400
    },
    {
      "epoch": 1.6473791695030633,
      "grad_norm": 3.1517415046691895,
      "learning_rate": 6.997870830376154e-06,
      "loss": 0.0444,
      "step": 2420
    },
    {
      "epoch": 1.6609938733832539,
      "grad_norm": 3.7715275287628174,
      "learning_rate": 6.9694819020581985e-06,
      "loss": 0.0478,
      "step": 2440
    },
    {
      "epoch": 1.6746085772634445,
      "grad_norm": 3.123321533203125,
      "learning_rate": 6.9410929737402426e-06,
      "loss": 0.0492,
      "step": 2460
    },
    {
      "epoch": 1.6882232811436353,
      "grad_norm": 3.15405535697937,
      "learning_rate": 6.912704045422286e-06,
      "loss": 0.0478,
      "step": 2480
    },
    {
      "epoch": 1.7018379850238259,
      "grad_norm": 3.2529942989349365,
      "learning_rate": 6.88431511710433e-06,
      "loss": 0.0478,
      "step": 2500
    },
    {
      "epoch": 1.7154526889040165,
      "grad_norm": 5.611981391906738,
      "learning_rate": 6.855926188786374e-06,
      "loss": 0.0516,
      "step": 2520
    },
    {
      "epoch": 1.729067392784207,
      "grad_norm": 2.6857259273529053,
      "learning_rate": 6.827537260468417e-06,
      "loss": 0.0416,
      "step": 2540
    },
    {
      "epoch": 1.7426820966643977,
      "grad_norm": 3.303100824356079,
      "learning_rate": 6.799148332150461e-06,
      "loss": 0.0431,
      "step": 2560
    },
    {
      "epoch": 1.7562968005445883,
      "grad_norm": 3.7506537437438965,
      "learning_rate": 6.770759403832506e-06,
      "loss": 0.0555,
      "step": 2580
    },
    {
      "epoch": 1.7699115044247788,
      "grad_norm": 2.169769048690796,
      "learning_rate": 6.74237047551455e-06,
      "loss": 0.0449,
      "step": 2600
    },
    {
      "epoch": 1.7835262083049694,
      "grad_norm": 2.864407539367676,
      "learning_rate": 6.713981547196594e-06,
      "loss": 0.045,
      "step": 2620
    },
    {
      "epoch": 1.79714091218516,
      "grad_norm": 2.529905319213867,
      "learning_rate": 6.685592618878638e-06,
      "loss": 0.0598,
      "step": 2640
    },
    {
      "epoch": 1.8107556160653506,
      "grad_norm": 3.385667324066162,
      "learning_rate": 6.657203690560682e-06,
      "loss": 0.05,
      "step": 2660
    },
    {
      "epoch": 1.8243703199455412,
      "grad_norm": 4.172725677490234,
      "learning_rate": 6.628814762242726e-06,
      "loss": 0.0473,
      "step": 2680
    },
    {
      "epoch": 1.8379850238257318,
      "grad_norm": 2.7874386310577393,
      "learning_rate": 6.60042583392477e-06,
      "loss": 0.0541,
      "step": 2700
    },
    {
      "epoch": 1.8515997277059224,
      "grad_norm": 2.623189926147461,
      "learning_rate": 6.572036905606813e-06,
      "loss": 0.0486,
      "step": 2720
    },
    {
      "epoch": 1.865214431586113,
      "grad_norm": 3.5073838233947754,
      "learning_rate": 6.543647977288859e-06,
      "loss": 0.0496,
      "step": 2740
    },
    {
      "epoch": 1.8788291354663036,
      "grad_norm": 2.545915126800537,
      "learning_rate": 6.515259048970902e-06,
      "loss": 0.05,
      "step": 2760
    },
    {
      "epoch": 1.8924438393464942,
      "grad_norm": 3.92095947265625,
      "learning_rate": 6.486870120652946e-06,
      "loss": 0.0576,
      "step": 2780
    },
    {
      "epoch": 1.9060585432266848,
      "grad_norm": 3.5295567512512207,
      "learning_rate": 6.45848119233499e-06,
      "loss": 0.0497,
      "step": 2800
    },
    {
      "epoch": 1.9196732471068754,
      "grad_norm": 2.9580278396606445,
      "learning_rate": 6.430092264017034e-06,
      "loss": 0.0487,
      "step": 2820
    },
    {
      "epoch": 1.933287950987066,
      "grad_norm": 2.3317949771881104,
      "learning_rate": 6.4017033356990774e-06,
      "loss": 0.0457,
      "step": 2840
    },
    {
      "epoch": 1.9469026548672566,
      "grad_norm": 2.8551571369171143,
      "learning_rate": 6.3733144073811215e-06,
      "loss": 0.0507,
      "step": 2860
    },
    {
      "epoch": 1.9605173587474471,
      "grad_norm": 3.363938093185425,
      "learning_rate": 6.344925479063166e-06,
      "loss": 0.0403,
      "step": 2880
    },
    {
      "epoch": 1.9741320626276377,
      "grad_norm": 3.0591518878936768,
      "learning_rate": 6.3165365507452105e-06,
      "loss": 0.0532,
      "step": 2900
    },
    {
      "epoch": 1.9877467665078283,
      "grad_norm": 3.6997368335723877,
      "learning_rate": 6.288147622427255e-06,
      "loss": 0.0572,
      "step": 2920
    },
    {
      "epoch": 2.001361470388019,
      "grad_norm": 1.2867134809494019,
      "learning_rate": 6.259758694109298e-06,
      "loss": 0.0393,
      "step": 2940
    },
    {
      "epoch": 2.0149761742682095,
      "grad_norm": 2.620507001876831,
      "learning_rate": 6.231369765791342e-06,
      "loss": 0.0201,
      "step": 2960
    },
    {
      "epoch": 2.0285908781484,
      "grad_norm": 1.7586274147033691,
      "learning_rate": 6.202980837473386e-06,
      "loss": 0.0164,
      "step": 2980
    },
    {
      "epoch": 2.0422055820285907,
      "grad_norm": 2.2848572731018066,
      "learning_rate": 6.17459190915543e-06,
      "loss": 0.0194,
      "step": 3000
    },
    {
      "epoch": 2.0558202859087813,
      "grad_norm": 1.6622294187545776,
      "learning_rate": 6.146202980837473e-06,
      "loss": 0.0164,
      "step": 3020
    },
    {
      "epoch": 2.069434989788972,
      "grad_norm": 3.2560629844665527,
      "learning_rate": 6.117814052519518e-06,
      "loss": 0.0184,
      "step": 3040
    },
    {
      "epoch": 2.0830496936691625,
      "grad_norm": 1.6647343635559082,
      "learning_rate": 6.089425124201562e-06,
      "loss": 0.014,
      "step": 3060
    },
    {
      "epoch": 2.096664397549353,
      "grad_norm": 1.7806779146194458,
      "learning_rate": 6.061036195883606e-06,
      "loss": 0.0172,
      "step": 3080
    },
    {
      "epoch": 2.110279101429544,
      "grad_norm": 1.9715111255645752,
      "learning_rate": 6.03264726756565e-06,
      "loss": 0.017,
      "step": 3100
    },
    {
      "epoch": 2.1238938053097347,
      "grad_norm": 2.1102495193481445,
      "learning_rate": 6.004258339247694e-06,
      "loss": 0.0154,
      "step": 3120
    },
    {
      "epoch": 2.1375085091899253,
      "grad_norm": 2.978144407272339,
      "learning_rate": 5.975869410929738e-06,
      "loss": 0.0227,
      "step": 3140
    },
    {
      "epoch": 2.151123213070116,
      "grad_norm": 1.914223074913025,
      "learning_rate": 5.947480482611782e-06,
      "loss": 0.0144,
      "step": 3160
    },
    {
      "epoch": 2.1647379169503065,
      "grad_norm": 2.0948495864868164,
      "learning_rate": 5.919091554293826e-06,
      "loss": 0.0161,
      "step": 3180
    },
    {
      "epoch": 2.178352620830497,
      "grad_norm": 2.937959909439087,
      "learning_rate": 5.890702625975871e-06,
      "loss": 0.0177,
      "step": 3200
    },
    {
      "epoch": 2.1919673247106877,
      "grad_norm": 0.9957959651947021,
      "learning_rate": 5.862313697657914e-06,
      "loss": 0.0165,
      "step": 3220
    },
    {
      "epoch": 2.2055820285908783,
      "grad_norm": 1.2504860162734985,
      "learning_rate": 5.833924769339958e-06,
      "loss": 0.0169,
      "step": 3240
    },
    {
      "epoch": 2.219196732471069,
      "grad_norm": 2.532622814178467,
      "learning_rate": 5.805535841022002e-06,
      "loss": 0.0193,
      "step": 3260
    },
    {
      "epoch": 2.2328114363512594,
      "grad_norm": 2.0159101486206055,
      "learning_rate": 5.777146912704046e-06,
      "loss": 0.0189,
      "step": 3280
    },
    {
      "epoch": 2.24642614023145,
      "grad_norm": 2.19897198677063,
      "learning_rate": 5.7487579843860894e-06,
      "loss": 0.0195,
      "step": 3300
    },
    {
      "epoch": 2.2600408441116406,
      "grad_norm": 1.4231209754943848,
      "learning_rate": 5.7203690560681335e-06,
      "loss": 0.0185,
      "step": 3320
    },
    {
      "epoch": 2.2736555479918312,
      "grad_norm": 1.622592568397522,
      "learning_rate": 5.691980127750178e-06,
      "loss": 0.0176,
      "step": 3340
    },
    {
      "epoch": 2.287270251872022,
      "grad_norm": 1.630968451499939,
      "learning_rate": 5.6635911994322225e-06,
      "loss": 0.0179,
      "step": 3360
    },
    {
      "epoch": 2.3008849557522124,
      "grad_norm": 1.7456097602844238,
      "learning_rate": 5.635202271114267e-06,
      "loss": 0.0182,
      "step": 3380
    },
    {
      "epoch": 2.314499659632403,
      "grad_norm": 3.1690380573272705,
      "learning_rate": 5.60681334279631e-06,
      "loss": 0.0149,
      "step": 3400
    },
    {
      "epoch": 2.3281143635125936,
      "grad_norm": 2.187356472015381,
      "learning_rate": 5.578424414478354e-06,
      "loss": 0.0182,
      "step": 3420
    },
    {
      "epoch": 2.341729067392784,
      "grad_norm": 1.772560954093933,
      "learning_rate": 5.550035486160398e-06,
      "loss": 0.022,
      "step": 3440
    },
    {
      "epoch": 2.355343771272975,
      "grad_norm": 1.4121778011322021,
      "learning_rate": 5.521646557842442e-06,
      "loss": 0.0184,
      "step": 3460
    },
    {
      "epoch": 2.3689584751531654,
      "grad_norm": 2.99052357673645,
      "learning_rate": 5.493257629524485e-06,
      "loss": 0.0169,
      "step": 3480
    },
    {
      "epoch": 2.382573179033356,
      "grad_norm": 1.1008409261703491,
      "learning_rate": 5.464868701206529e-06,
      "loss": 0.0139,
      "step": 3500
    },
    {
      "epoch": 2.3961878829135466,
      "grad_norm": 1.6258739233016968,
      "learning_rate": 5.436479772888574e-06,
      "loss": 0.0132,
      "step": 3520
    },
    {
      "epoch": 2.409802586793737,
      "grad_norm": 2.573024034500122,
      "learning_rate": 5.408090844570618e-06,
      "loss": 0.0145,
      "step": 3540
    },
    {
      "epoch": 2.4234172906739277,
      "grad_norm": 1.7651166915893555,
      "learning_rate": 5.379701916252662e-06,
      "loss": 0.018,
      "step": 3560
    },
    {
      "epoch": 2.4370319945541183,
      "grad_norm": 1.000501036643982,
      "learning_rate": 5.351312987934706e-06,
      "loss": 0.0174,
      "step": 3580
    },
    {
      "epoch": 2.450646698434309,
      "grad_norm": 1.9365277290344238,
      "learning_rate": 5.32292405961675e-06,
      "loss": 0.0239,
      "step": 3600
    },
    {
      "epoch": 2.4642614023144995,
      "grad_norm": 1.6012334823608398,
      "learning_rate": 5.294535131298794e-06,
      "loss": 0.0158,
      "step": 3620
    },
    {
      "epoch": 2.47787610619469,
      "grad_norm": 1.9519362449645996,
      "learning_rate": 5.266146202980838e-06,
      "loss": 0.0172,
      "step": 3640
    },
    {
      "epoch": 2.4914908100748807,
      "grad_norm": 1.6032203435897827,
      "learning_rate": 5.237757274662883e-06,
      "loss": 0.0168,
      "step": 3660
    },
    {
      "epoch": 2.5051055139550717,
      "grad_norm": 1.5285481214523315,
      "learning_rate": 5.209368346344926e-06,
      "loss": 0.0197,
      "step": 3680
    },
    {
      "epoch": 2.518720217835262,
      "grad_norm": 0.9293742179870605,
      "learning_rate": 5.18097941802697e-06,
      "loss": 0.0169,
      "step": 3700
    },
    {
      "epoch": 2.532334921715453,
      "grad_norm": 1.169976830482483,
      "learning_rate": 5.152590489709014e-06,
      "loss": 0.0173,
      "step": 3720
    },
    {
      "epoch": 2.545949625595643,
      "grad_norm": 1.4823977947235107,
      "learning_rate": 5.124201561391058e-06,
      "loss": 0.0166,
      "step": 3740
    },
    {
      "epoch": 2.559564329475834,
      "grad_norm": 0.9930800199508667,
      "learning_rate": 5.0958126330731014e-06,
      "loss": 0.0179,
      "step": 3760
    },
    {
      "epoch": 2.5731790333560243,
      "grad_norm": 1.3004252910614014,
      "learning_rate": 5.0674237047551455e-06,
      "loss": 0.0189,
      "step": 3780
    },
    {
      "epoch": 2.5867937372362153,
      "grad_norm": 2.19750714302063,
      "learning_rate": 5.03903477643719e-06,
      "loss": 0.0181,
      "step": 3800
    },
    {
      "epoch": 2.6004084411164055,
      "grad_norm": 2.087999105453491,
      "learning_rate": 5.0106458481192345e-06,
      "loss": 0.0151,
      "step": 3820
    },
    {
      "epoch": 2.6140231449965965,
      "grad_norm": 1.2687196731567383,
      "learning_rate": 4.982256919801278e-06,
      "loss": 0.018,
      "step": 3840
    },
    {
      "epoch": 2.627637848876787,
      "grad_norm": 1.2243746519088745,
      "learning_rate": 4.953867991483322e-06,
      "loss": 0.0173,
      "step": 3860
    },
    {
      "epoch": 2.6412525527569777,
      "grad_norm": 0.9616296291351318,
      "learning_rate": 4.925479063165366e-06,
      "loss": 0.02,
      "step": 3880
    },
    {
      "epoch": 2.6548672566371683,
      "grad_norm": 0.7424539923667908,
      "learning_rate": 4.89709013484741e-06,
      "loss": 0.0143,
      "step": 3900
    },
    {
      "epoch": 2.668481960517359,
      "grad_norm": 2.386852979660034,
      "learning_rate": 4.868701206529454e-06,
      "loss": 0.0165,
      "step": 3920
    },
    {
      "epoch": 2.6820966643975495,
      "grad_norm": 1.4014451503753662,
      "learning_rate": 4.840312278211498e-06,
      "loss": 0.0231,
      "step": 3940
    },
    {
      "epoch": 2.69571136827774,
      "grad_norm": 1.2173271179199219,
      "learning_rate": 4.811923349893542e-06,
      "loss": 0.0181,
      "step": 3960
    },
    {
      "epoch": 2.7093260721579306,
      "grad_norm": 3.268631935119629,
      "learning_rate": 4.783534421575585e-06,
      "loss": 0.0179,
      "step": 3980
    },
    {
      "epoch": 2.7229407760381212,
      "grad_norm": 1.9503110647201538,
      "learning_rate": 4.7551454932576295e-06,
      "loss": 0.0154,
      "step": 4000
    },
    {
      "epoch": 2.7229407760381212,
      "eval_loss": 0.03271039202809334,
      "eval_runtime": 4672.049,
      "eval_samples_per_second": 2.515,
      "eval_steps_per_second": 0.314,
      "eval_wer": 2.1779719650083877,
      "step": 4000
    }
  ],
  "logging_steps": 20,
  "max_steps": 7345,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 2000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.693258236215296e+19,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}