{
  "best_metric": 0.5749832987785339,
  "best_model_checkpoint": "./Melanoma-Classification/checkpoint-3800",
  "epoch": 4.0,
  "eval_steps": 100,
  "global_step": 5068,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "grad_norm": 1.4924156665802002,
      "learning_rate": 0.00019960536700868193,
      "loss": 1.5595,
      "step": 10
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.687912106513977,
      "learning_rate": 0.00019921073401736386,
      "loss": 1.3491,
      "step": 20
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1836044788360596,
      "learning_rate": 0.0001988161010260458,
      "loss": 1.1583,
      "step": 30
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.7617909908294678,
      "learning_rate": 0.00019842146803472773,
      "loss": 1.4836,
      "step": 40
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7186870574951172,
      "learning_rate": 0.00019802683504340965,
      "loss": 1.1296,
      "step": 50
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4756615161895752,
      "learning_rate": 0.00019763220205209155,
      "loss": 1.0864,
      "step": 60
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.723090648651123,
      "learning_rate": 0.00019723756906077347,
      "loss": 1.1339,
      "step": 70
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.304826021194458,
      "learning_rate": 0.0001968429360694554,
      "loss": 1.0937,
      "step": 80
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5404345989227295,
      "learning_rate": 0.00019644830307813735,
      "loss": 1.0765,
      "step": 90
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.1473004817962646,
      "learning_rate": 0.00019605367008681927,
      "loss": 0.9779,
      "step": 100
    },
    {
      "epoch": 0.08,
      "eval_accuracy": 0.6041049930925597,
      "eval_loss": 1.1158318519592285,
      "eval_runtime": 91.4624,
      "eval_samples_per_second": 55.4,
      "eval_steps_per_second": 6.932,
      "step": 100
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.886522054672241,
      "learning_rate": 0.0001956590370955012,
      "loss": 1.0835,
      "step": 110
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.163769006729126,
      "learning_rate": 0.00019526440410418312,
      "loss": 1.1449,
      "step": 120
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7149088382720947,
      "learning_rate": 0.00019486977111286504,
      "loss": 1.115,
      "step": 130
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.6274876594543457,
      "learning_rate": 0.00019447513812154697,
      "loss": 1.1516,
      "step": 140
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5108281373977661,
      "learning_rate": 0.00019408050513022892,
      "loss": 1.0328,
      "step": 150
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.575366497039795,
      "learning_rate": 0.00019368587213891084,
      "loss": 1.0288,
      "step": 160
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9967373609542847,
      "learning_rate": 0.00019329123914759276,
      "loss": 0.89,
      "step": 170
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.47580885887146,
      "learning_rate": 0.00019289660615627466,
      "loss": 1.0238,
      "step": 180
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.0836246013641357,
      "learning_rate": 0.00019250197316495658,
      "loss": 1.034,
      "step": 190
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9168826341629028,
      "learning_rate": 0.0001921073401736385,
      "loss": 0.9934,
      "step": 200
    },
    {
      "epoch": 0.16,
      "eval_accuracy": 0.650088809946714,
      "eval_loss": 1.0227293968200684,
      "eval_runtime": 91.1657,
      "eval_samples_per_second": 55.58,
      "eval_steps_per_second": 6.954,
      "step": 200
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.0433690547943115,
      "learning_rate": 0.00019171270718232046,
      "loss": 0.9561,
      "step": 210
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9006619453430176,
      "learning_rate": 0.00019131807419100238,
      "loss": 1.1087,
      "step": 220
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.693882942199707,
      "learning_rate": 0.0001909234411996843,
      "loss": 1.0212,
      "step": 230
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.44376802444458,
      "learning_rate": 0.00019052880820836623,
      "loss": 0.9192,
      "step": 240
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.168645977973938,
      "learning_rate": 0.00019013417521704815,
      "loss": 0.9394,
      "step": 250
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.2017979621887207,
      "learning_rate": 0.00018973954222573008,
      "loss": 1.0321,
      "step": 260
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.930981397628784,
      "learning_rate": 0.00018934490923441203,
      "loss": 1.019,
      "step": 270
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.8777360916137695,
      "learning_rate": 0.00018895027624309395,
      "loss": 1.1093,
      "step": 280
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.2270009517669678,
      "learning_rate": 0.00018855564325177585,
      "loss": 0.9979,
      "step": 290
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8015016317367554,
      "learning_rate": 0.00018816101026045777,
      "loss": 0.9562,
      "step": 300
    },
    {
      "epoch": 0.24,
      "eval_accuracy": 0.6747582395895007,
      "eval_loss": 0.9276195168495178,
      "eval_runtime": 91.6369,
      "eval_samples_per_second": 55.294,
      "eval_steps_per_second": 6.919,
      "step": 300
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.6960203647613525,
      "learning_rate": 0.0001877663772691397,
      "loss": 0.95,
      "step": 310
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5449939966201782,
      "learning_rate": 0.00018737174427782162,
      "loss": 0.9869,
      "step": 320
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.7589311599731445,
      "learning_rate": 0.00018697711128650357,
      "loss": 0.8735,
      "step": 330
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1408846378326416,
      "learning_rate": 0.0001865824782951855,
      "loss": 0.8768,
      "step": 340
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8114129304885864,
      "learning_rate": 0.00018618784530386741,
      "loss": 0.8912,
      "step": 350
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.1425697803497314,
      "learning_rate": 0.00018579321231254934,
      "loss": 0.9495,
      "step": 360
    },
    {
      "epoch": 0.29,
      "grad_norm": 2.0084307193756104,
      "learning_rate": 0.00018539857932123126,
      "loss": 1.0388,
      "step": 370
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8898825645446777,
      "learning_rate": 0.00018500394632991319,
      "loss": 0.9202,
      "step": 380
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.13079833984375,
      "learning_rate": 0.00018460931333859514,
      "loss": 0.9567,
      "step": 390
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.111664056777954,
      "learning_rate": 0.00018421468034727706,
      "loss": 1.0995,
      "step": 400
    },
    {
      "epoch": 0.32,
      "eval_accuracy": 0.6836392342609039,
      "eval_loss": 0.9088240265846252,
      "eval_runtime": 91.7358,
      "eval_samples_per_second": 55.235,
      "eval_steps_per_second": 6.911,
      "step": 400
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.5040955543518066,
      "learning_rate": 0.00018382004735595896,
      "loss": 0.7407,
      "step": 410
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1850649118423462,
      "learning_rate": 0.00018342541436464088,
      "loss": 0.8685,
      "step": 420
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.618028163909912,
      "learning_rate": 0.0001830307813733228,
      "loss": 0.9087,
      "step": 430
    },
    {
      "epoch": 0.35,
      "grad_norm": 3.015228509902954,
      "learning_rate": 0.00018263614838200473,
      "loss": 1.0173,
      "step": 440
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2800592184066772,
      "learning_rate": 0.00018224151539068668,
      "loss": 1.1207,
      "step": 450
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.40296471118927,
      "learning_rate": 0.0001818468823993686,
      "loss": 0.7813,
      "step": 460
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.0328774452209473,
      "learning_rate": 0.00018145224940805052,
      "loss": 0.8776,
      "step": 470
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.16869854927063,
      "learning_rate": 0.00018105761641673245,
      "loss": 1.1089,
      "step": 480
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.867370367050171,
      "learning_rate": 0.00018066298342541437,
      "loss": 0.9493,
      "step": 490
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.4438893795013428,
      "learning_rate": 0.0001802683504340963,
      "loss": 0.8198,
      "step": 500
    },
    {
      "epoch": 0.39,
      "eval_accuracy": 0.6948884941780146,
      "eval_loss": 0.8580543398857117,
      "eval_runtime": 92.2035,
      "eval_samples_per_second": 54.955,
      "eval_steps_per_second": 6.876,
      "step": 500
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3336217403411865,
      "learning_rate": 0.00017987371744277825,
      "loss": 0.7372,
      "step": 510
    },
    {
      "epoch": 0.41,
      "grad_norm": 2.482779026031494,
      "learning_rate": 0.00017947908445146017,
      "loss": 0.8516,
      "step": 520
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5005723237991333,
      "learning_rate": 0.00017908445146014207,
      "loss": 0.7964,
      "step": 530
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8133363723754883,
      "learning_rate": 0.000178689818468824,
      "loss": 0.8267,
      "step": 540
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.8624382019042969,
      "learning_rate": 0.0001782951854775059,
      "loss": 0.8205,
      "step": 550
    },
    {
      "epoch": 0.44,
      "grad_norm": 2.319140911102295,
      "learning_rate": 0.00017790055248618784,
      "loss": 0.8772,
      "step": 560
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.922405958175659,
      "learning_rate": 0.0001775059194948698,
      "loss": 0.9475,
      "step": 570
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8725391030311584,
      "learning_rate": 0.0001771112865035517,
      "loss": 0.8155,
      "step": 580
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.170085906982422,
      "learning_rate": 0.00017671665351223363,
      "loss": 1.0075,
      "step": 590
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.645709276199341,
      "learning_rate": 0.00017632202052091556,
      "loss": 0.8034,
      "step": 600
    },
    {
      "epoch": 0.47,
      "eval_accuracy": 0.6966646931122953,
      "eval_loss": 0.8444026112556458,
      "eval_runtime": 91.3762,
      "eval_samples_per_second": 55.452,
      "eval_steps_per_second": 6.938,
      "step": 600
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4042195081710815,
      "learning_rate": 0.00017592738752959748,
      "loss": 0.8112,
      "step": 610
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9543468952178955,
      "learning_rate": 0.0001755327545382794,
      "loss": 0.8895,
      "step": 620
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4769212007522583,
      "learning_rate": 0.00017513812154696135,
      "loss": 0.8816,
      "step": 630
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.843929648399353,
      "learning_rate": 0.00017474348855564325,
      "loss": 0.834,
      "step": 640
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.1005053520202637,
      "learning_rate": 0.00017434885556432517,
      "loss": 0.8459,
      "step": 650
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.8301466703414917,
      "learning_rate": 0.0001739542225730071,
      "loss": 0.745,
      "step": 660
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.7254234552383423,
      "learning_rate": 0.00017355958958168902,
      "loss": 0.7985,
      "step": 670
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.2055907249450684,
      "learning_rate": 0.00017316495659037095,
      "loss": 0.768,
      "step": 680
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.17266845703125,
      "learning_rate": 0.0001727703235990529,
      "loss": 0.8608,
      "step": 690
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5750426054000854,
      "learning_rate": 0.00017237569060773482,
      "loss": 0.8319,
      "step": 700
    },
    {
      "epoch": 0.55,
      "eval_accuracy": 0.7148213933293862,
      "eval_loss": 0.819585919380188,
      "eval_runtime": 91.97,
      "eval_samples_per_second": 55.094,
      "eval_steps_per_second": 6.894,
      "step": 700
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.9275529384613037,
      "learning_rate": 0.00017198105761641674,
      "loss": 0.8694,
      "step": 710
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.029745101928711,
      "learning_rate": 0.00017158642462509867,
      "loss": 0.814,
      "step": 720
    },
    {
      "epoch": 0.58,
      "grad_norm": 2.5681614875793457,
      "learning_rate": 0.0001711917916337806,
      "loss": 0.8117,
      "step": 730
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8536593914031982,
      "learning_rate": 0.00017079715864246251,
      "loss": 0.8844,
      "step": 740
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9550455808639526,
      "learning_rate": 0.00017040252565114446,
      "loss": 0.8574,
      "step": 750
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.0112359523773193,
      "learning_rate": 0.00017000789265982636,
      "loss": 0.7197,
      "step": 760
    },
    {
      "epoch": 0.61,
      "grad_norm": 2.2003233432769775,
      "learning_rate": 0.00016961325966850828,
      "loss": 0.8009,
      "step": 770
    },
    {
      "epoch": 0.62,
      "grad_norm": 2.5492188930511475,
      "learning_rate": 0.0001692186266771902,
      "loss": 0.7504,
      "step": 780
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8917009830474854,
      "learning_rate": 0.00016882399368587213,
      "loss": 0.7638,
      "step": 790
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2614909410476685,
      "learning_rate": 0.00016842936069455408,
      "loss": 0.787,
      "step": 800
    },
    {
      "epoch": 0.63,
      "eval_accuracy": 0.6974541148608644,
      "eval_loss": 0.8360146880149841,
      "eval_runtime": 91.0831,
      "eval_samples_per_second": 55.63,
      "eval_steps_per_second": 6.961,
      "step": 800
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.555303692817688,
      "learning_rate": 0.000168034727703236,
      "loss": 0.9073,
      "step": 810
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3937816619873047,
      "learning_rate": 0.00016764009471191793,
      "loss": 0.7977,
      "step": 820
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7314428091049194,
      "learning_rate": 0.00016724546172059985,
      "loss": 0.8166,
      "step": 830
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.7047677040100098,
      "learning_rate": 0.00016685082872928178,
      "loss": 0.9118,
      "step": 840
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.1908257007598877,
      "learning_rate": 0.0001664561957379637,
      "loss": 0.7528,
      "step": 850
    },
    {
      "epoch": 0.68,
      "grad_norm": 2.966177463531494,
      "learning_rate": 0.00016606156274664565,
      "loss": 0.8306,
      "step": 860
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.368105411529541,
      "learning_rate": 0.00016566692975532755,
      "loss": 0.8289,
      "step": 870
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.197537660598755,
      "learning_rate": 0.00016527229676400947,
      "loss": 0.9994,
      "step": 880
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.863797187805176,
      "learning_rate": 0.0001648776637726914,
      "loss": 0.852,
      "step": 890
    },
    {
      "epoch": 0.71,
      "grad_norm": 2.704500198364258,
      "learning_rate": 0.00016448303078137332,
      "loss": 0.8642,
      "step": 900
    },
    {
      "epoch": 0.71,
      "eval_accuracy": 0.7008091572922834,
      "eval_loss": 0.8250275254249573,
      "eval_runtime": 90.6591,
      "eval_samples_per_second": 55.891,
      "eval_steps_per_second": 6.993,
      "step": 900
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.438966751098633,
      "learning_rate": 0.00016408839779005524,
      "loss": 0.7703,
      "step": 910
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.4163198471069336,
      "learning_rate": 0.0001636937647987372,
      "loss": 0.7739,
      "step": 920
    },
    {
      "epoch": 0.73,
      "grad_norm": 3.480591297149658,
      "learning_rate": 0.00016329913180741912,
      "loss": 0.8631,
      "step": 930
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.24957537651062,
      "learning_rate": 0.00016290449881610104,
      "loss": 0.8008,
      "step": 940
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.1198952198028564,
      "learning_rate": 0.00016250986582478296,
      "loss": 0.8948,
      "step": 950
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7699896097183228,
      "learning_rate": 0.00016211523283346489,
      "loss": 0.8871,
      "step": 960
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3637826442718506,
      "learning_rate": 0.0001617205998421468,
      "loss": 0.7904,
      "step": 970
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.0125231742858887,
      "learning_rate": 0.00016132596685082876,
      "loss": 0.6222,
      "step": 980
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.767691731452942,
      "learning_rate": 0.00016093133385951066,
      "loss": 0.8987,
      "step": 990
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2865104675292969,
      "learning_rate": 0.00016053670086819258,
      "loss": 0.8329,
      "step": 1000
    },
    {
      "epoch": 0.79,
      "eval_accuracy": 0.7171896585750938,
      "eval_loss": 0.79388028383255,
      "eval_runtime": 90.2373,
      "eval_samples_per_second": 56.152,
      "eval_steps_per_second": 7.026,
      "step": 1000
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.2386913299560547,
      "learning_rate": 0.0001601420678768745,
      "loss": 0.7724,
      "step": 1010
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5864025354385376,
      "learning_rate": 0.00015974743488555643,
      "loss": 0.7434,
      "step": 1020
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.7106010913848877,
      "learning_rate": 0.00015935280189423835,
      "loss": 0.6527,
      "step": 1030
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.2001113891601562,
      "learning_rate": 0.0001589581689029203,
      "loss": 0.7301,
      "step": 1040
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8353298902511597,
      "learning_rate": 0.00015856353591160222,
      "loss": 0.7831,
      "step": 1050
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.228792190551758,
      "learning_rate": 0.00015816890292028415,
      "loss": 0.7633,
      "step": 1060
    },
    {
      "epoch": 0.84,
      "grad_norm": 3.594202756881714,
      "learning_rate": 0.00015777426992896607,
      "loss": 0.8804,
      "step": 1070
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.274918794631958,
      "learning_rate": 0.000157379636937648,
      "loss": 0.9008,
      "step": 1080
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.0492002964019775,
      "learning_rate": 0.00015698500394632992,
      "loss": 0.6768,
      "step": 1090
    },
    {
      "epoch": 0.87,
      "grad_norm": 2.415807008743286,
      "learning_rate": 0.00015659037095501187,
      "loss": 0.9678,
      "step": 1100
    },
    {
      "epoch": 0.87,
      "eval_accuracy": 0.7331754489836195,
      "eval_loss": 0.7660978436470032,
      "eval_runtime": 92.2043,
      "eval_samples_per_second": 54.954,
      "eval_steps_per_second": 6.876,
      "step": 1100
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.9822345972061157,
      "learning_rate": 0.00015619573796369377,
      "loss": 0.7861,
      "step": 1110
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.869279384613037,
      "learning_rate": 0.0001558011049723757,
      "loss": 0.7472,
      "step": 1120
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.2551605701446533,
      "learning_rate": 0.0001554064719810576,
      "loss": 0.6913,
      "step": 1130
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.4671690464019775,
      "learning_rate": 0.00015501183898973954,
      "loss": 0.6751,
      "step": 1140
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8815134763717651,
      "learning_rate": 0.00015461720599842146,
      "loss": 0.6414,
      "step": 1150
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.746532917022705,
      "learning_rate": 0.0001542225730071034,
      "loss": 0.849,
      "step": 1160
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.844710111618042,
      "learning_rate": 0.00015382794001578533,
      "loss": 0.8739,
      "step": 1170
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.393015146255493,
      "learning_rate": 0.00015343330702446726,
      "loss": 0.7106,
      "step": 1180
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9429343938827515,
      "learning_rate": 0.00015303867403314918,
      "loss": 0.7832,
      "step": 1190
    },
    {
      "epoch": 0.95,
      "grad_norm": 2.263422727584839,
      "learning_rate": 0.0001526440410418311,
      "loss": 0.8226,
      "step": 1200
    },
    {
      "epoch": 0.95,
      "eval_accuracy": 0.7373199131636077,
      "eval_loss": 0.7284455895423889,
      "eval_runtime": 91.1355,
      "eval_samples_per_second": 55.599,
      "eval_steps_per_second": 6.957,
      "step": 1200
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.1170480251312256,
      "learning_rate": 0.00015224940805051303,
      "loss": 0.8428,
      "step": 1210
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.143359422683716,
      "learning_rate": 0.00015185477505919495,
      "loss": 0.8429,
      "step": 1220
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9562945365905762,
      "learning_rate": 0.00015146014206787688,
      "loss": 0.7666,
      "step": 1230
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7012232542037964,
      "learning_rate": 0.0001510655090765588,
      "loss": 0.7828,
      "step": 1240
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.563348412513733,
      "learning_rate": 0.00015067087608524072,
      "loss": 0.8311,
      "step": 1250
    },
    {
      "epoch": 0.99,
      "grad_norm": 2.0692856311798096,
      "learning_rate": 0.00015027624309392265,
      "loss": 0.7133,
      "step": 1260
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.52514910697937,
      "learning_rate": 0.00014988161010260457,
      "loss": 0.8335,
      "step": 1270
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.8508520126342773,
      "learning_rate": 0.00014948697711128652,
      "loss": 0.5996,
      "step": 1280
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.858475685119629,
      "learning_rate": 0.00014909234411996844,
      "loss": 0.6434,
      "step": 1290
    },
    {
      "epoch": 1.03,
      "grad_norm": 4.348726272583008,
      "learning_rate": 0.00014869771112865037,
      "loss": 0.7967,
      "step": 1300
    },
    {
      "epoch": 1.03,
      "eval_accuracy": 0.7410696664693113,
      "eval_loss": 0.7354866862297058,
      "eval_runtime": 91.4806,
      "eval_samples_per_second": 55.389,
      "eval_steps_per_second": 6.93,
      "step": 1300
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.194589614868164,
      "learning_rate": 0.0001483030781373323,
      "loss": 0.5958,
      "step": 1310
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.5000673532485962,
      "learning_rate": 0.00014790844514601421,
      "loss": 0.7041,
      "step": 1320
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.6893486976623535,
      "learning_rate": 0.00014751381215469614,
      "loss": 0.6521,
      "step": 1330
    },
    {
      "epoch": 1.06,
      "grad_norm": 2.065706968307495,
      "learning_rate": 0.00014711917916337806,
      "loss": 0.6359,
      "step": 1340
    },
    {
      "epoch": 1.07,
      "grad_norm": 2.6649951934814453,
      "learning_rate": 0.00014672454617205998,
      "loss": 0.7392,
      "step": 1350
    },
    {
      "epoch": 1.07,
      "grad_norm": 2.844940185546875,
      "learning_rate": 0.0001463299131807419,
      "loss": 0.6207,
      "step": 1360
    },
    {
      "epoch": 1.08,
      "grad_norm": 2.581749677658081,
      "learning_rate": 0.00014593528018942383,
      "loss": 0.551,
      "step": 1370
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.7631639242172241,
      "learning_rate": 0.00014554064719810576,
      "loss": 0.6514,
      "step": 1380
    },
    {
      "epoch": 1.1,
      "grad_norm": 4.243679523468018,
      "learning_rate": 0.00014514601420678768,
      "loss": 0.5341,
      "step": 1390
    },
    {
      "epoch": 1.1,
      "grad_norm": 3.421487808227539,
      "learning_rate": 0.00014475138121546963,
      "loss": 0.6531,
      "step": 1400
    },
    {
      "epoch": 1.1,
      "eval_accuracy": 0.7246891651865008,
      "eval_loss": 0.7560569047927856,
      "eval_runtime": 91.2074,
      "eval_samples_per_second": 55.555,
      "eval_steps_per_second": 6.951,
      "step": 1400
    },
    {
      "epoch": 1.11,
      "grad_norm": 4.670535564422607,
      "learning_rate": 0.00014435674822415155,
      "loss": 0.6982,
      "step": 1410
    },
    {
      "epoch": 1.12,
      "grad_norm": 3.2248194217681885,
      "learning_rate": 0.00014396211523283348,
      "loss": 0.5345,
      "step": 1420
    },
    {
      "epoch": 1.13,
      "grad_norm": 3.1267943382263184,
      "learning_rate": 0.0001435674822415154,
      "loss": 0.7239,
      "step": 1430
    },
    {
      "epoch": 1.14,
      "grad_norm": 3.158299684524536,
      "learning_rate": 0.00014317284925019732,
      "loss": 0.6854,
      "step": 1440
    },
    {
      "epoch": 1.14,
      "grad_norm": 2.1028189659118652,
      "learning_rate": 0.00014277821625887925,
      "loss": 0.6367,
      "step": 1450
    },
    {
      "epoch": 1.15,
      "grad_norm": 2.811735153198242,
      "learning_rate": 0.00014238358326756117,
      "loss": 0.6858,
      "step": 1460
    },
    {
      "epoch": 1.16,
      "grad_norm": 2.150059461593628,
      "learning_rate": 0.0001419889502762431,
      "loss": 0.7487,
      "step": 1470
    },
    {
      "epoch": 1.17,
      "grad_norm": 2.536576509475708,
      "learning_rate": 0.00014159431728492502,
      "loss": 0.5976,
      "step": 1480
    },
    {
      "epoch": 1.18,
      "grad_norm": 2.9393413066864014,
      "learning_rate": 0.00014119968429360694,
      "loss": 0.5959,
      "step": 1490
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.6796705722808838,
      "learning_rate": 0.00014080505130228886,
      "loss": 0.5719,
      "step": 1500
    },
    {
      "epoch": 1.18,
      "eval_accuracy": 0.7637655417406749,
      "eval_loss": 0.6839306354522705,
      "eval_runtime": 91.2672,
      "eval_samples_per_second": 55.518,
      "eval_steps_per_second": 6.947,
      "step": 1500
    },
    {
      "epoch": 1.19,
      "grad_norm": 2.542813301086426,
      "learning_rate": 0.00014041041831097082,
      "loss": 0.5247,
      "step": 1510
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9640020132064819,
      "learning_rate": 0.00014001578531965274,
      "loss": 0.4843,
      "step": 1520
    },
    {
      "epoch": 1.21,
      "grad_norm": 2.0161819458007812,
      "learning_rate": 0.00013962115232833466,
      "loss": 0.676,
      "step": 1530
    },
    {
      "epoch": 1.22,
      "grad_norm": 2.028648853302002,
      "learning_rate": 0.00013922651933701659,
      "loss": 0.5935,
      "step": 1540
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.2919349670410156,
      "learning_rate": 0.0001388318863456985,
      "loss": 0.5941,
      "step": 1550
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.8597910404205322,
      "learning_rate": 0.00013843725335438043,
      "loss": 0.6447,
      "step": 1560
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.6460645198822021,
      "learning_rate": 0.00013804262036306236,
      "loss": 0.5576,
      "step": 1570
    },
    {
      "epoch": 1.25,
      "grad_norm": 3.1643693447113037,
      "learning_rate": 0.00013764798737174428,
      "loss": 0.6632,
      "step": 1580
    },
    {
      "epoch": 1.25,
      "grad_norm": 3.991893768310547,
      "learning_rate": 0.0001372533543804262,
      "loss": 0.6406,
      "step": 1590
    },
    {
      "epoch": 1.26,
      "grad_norm": 3.314627170562744,
      "learning_rate": 0.00013685872138910813,
      "loss": 0.6123,
      "step": 1600
    },
    {
      "epoch": 1.26,
      "eval_accuracy": 0.7584369449378331,
      "eval_loss": 0.6857106685638428,
      "eval_runtime": 91.5334,
      "eval_samples_per_second": 55.357,
      "eval_steps_per_second": 6.926,
      "step": 1600
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.986220121383667,
      "learning_rate": 0.00013646408839779005,
      "loss": 0.7161,
      "step": 1610
    },
    {
      "epoch": 1.28,
      "grad_norm": 4.052389144897461,
      "learning_rate": 0.00013606945540647197,
      "loss": 0.7073,
      "step": 1620
    },
    {
      "epoch": 1.29,
      "grad_norm": 2.59963321685791,
      "learning_rate": 0.00013567482241515392,
      "loss": 0.6871,
      "step": 1630
    },
    {
      "epoch": 1.29,
      "grad_norm": 3.5070436000823975,
      "learning_rate": 0.00013528018942383585,
      "loss": 0.6279,
      "step": 1640
    },
    {
      "epoch": 1.3,
      "grad_norm": 2.9049060344696045,
      "learning_rate": 0.00013488555643251777,
      "loss": 0.5972,
      "step": 1650
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.6428894996643066,
      "learning_rate": 0.0001344909234411997,
      "loss": 0.5128,
      "step": 1660
    },
    {
      "epoch": 1.32,
      "grad_norm": 2.659148693084717,
      "learning_rate": 0.00013409629044988162,
      "loss": 0.7127,
      "step": 1670
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.5462294816970825,
      "learning_rate": 0.00013370165745856354,
      "loss": 0.6444,
      "step": 1680
    },
    {
      "epoch": 1.33,
      "grad_norm": 2.8382019996643066,
      "learning_rate": 0.00013330702446724547,
      "loss": 0.5236,
      "step": 1690
    },
    {
      "epoch": 1.34,
      "grad_norm": 2.1057918071746826,
      "learning_rate": 0.0001329123914759274,
      "loss": 0.6504,
      "step": 1700
    },
    {
      "epoch": 1.34,
      "eval_accuracy": 0.7531083481349912,
      "eval_loss": 0.6970384120941162,
      "eval_runtime": 92.1039,
      "eval_samples_per_second": 55.014,
      "eval_steps_per_second": 6.884,
      "step": 1700
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.670453429222107,
      "learning_rate": 0.0001325177584846093,
      "loss": 0.5204,
      "step": 1710
    },
    {
      "epoch": 1.36,
      "grad_norm": 3.400521755218506,
      "learning_rate": 0.00013212312549329124,
      "loss": 0.5707,
      "step": 1720
    },
    {
      "epoch": 1.37,
      "grad_norm": 2.378593683242798,
      "learning_rate": 0.00013172849250197316,
      "loss": 0.6149,
      "step": 1730
    },
    {
      "epoch": 1.37,
      "grad_norm": 4.090423583984375,
      "learning_rate": 0.00013133385951065508,
      "loss": 0.7014,
      "step": 1740
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.9423013925552368,
      "learning_rate": 0.00013093922651933703,
      "loss": 0.4657,
      "step": 1750
    },
    {
      "epoch": 1.39,
      "grad_norm": 2.23858380317688,
      "learning_rate": 0.00013054459352801896,
      "loss": 0.5976,
      "step": 1760
    },
    {
      "epoch": 1.4,
      "grad_norm": 3.8164286613464355,
      "learning_rate": 0.00013014996053670088,
      "loss": 0.5735,
      "step": 1770
    },
    {
      "epoch": 1.4,
      "grad_norm": 3.1099419593811035,
      "learning_rate": 0.0001297553275453828,
      "loss": 0.5255,
      "step": 1780
    },
    {
      "epoch": 1.41,
      "grad_norm": 3.818204879760742,
      "learning_rate": 0.00012936069455406473,
      "loss": 0.6007,
      "step": 1790
    },
    {
      "epoch": 1.42,
      "grad_norm": 3.2938928604125977,
      "learning_rate": 0.00012896606156274665,
      "loss": 0.6214,
      "step": 1800
    },
    {
      "epoch": 1.42,
      "eval_accuracy": 0.7576475231892639,
      "eval_loss": 0.6841119527816772,
      "eval_runtime": 91.5596,
      "eval_samples_per_second": 55.341,
      "eval_steps_per_second": 6.924,
      "step": 1800
    },
    {
      "epoch": 1.43,
      "grad_norm": 3.2717528343200684,
      "learning_rate": 0.00012857142857142858,
      "loss": 0.6118,
      "step": 1810
    },
    {
      "epoch": 1.44,
      "grad_norm": 4.299696445465088,
      "learning_rate": 0.0001281767955801105,
      "loss": 0.5825,
      "step": 1820
    },
    {
      "epoch": 1.44,
      "grad_norm": 3.3908324241638184,
      "learning_rate": 0.00012778216258879242,
      "loss": 0.6013,
      "step": 1830
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.867518424987793,
      "learning_rate": 0.00012738752959747435,
      "loss": 0.4792,
      "step": 1840
    },
    {
      "epoch": 1.46,
      "grad_norm": 4.130500793457031,
      "learning_rate": 0.00012699289660615627,
      "loss": 0.624,
      "step": 1850
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.728039026260376,
      "learning_rate": 0.0001265982636148382,
      "loss": 0.534,
      "step": 1860
    },
    {
      "epoch": 1.48,
      "grad_norm": 3.357947826385498,
      "learning_rate": 0.00012620363062352014,
      "loss": 0.6773,
      "step": 1870
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.5584005117416382,
      "learning_rate": 0.00012580899763220207,
      "loss": 0.634,
      "step": 1880
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.4084281921386719,
      "learning_rate": 0.000125414364640884,
      "loss": 0.5695,
      "step": 1890
    },
    {
      "epoch": 1.5,
      "grad_norm": 3.2529728412628174,
      "learning_rate": 0.00012501973164956591,
      "loss": 0.4925,
      "step": 1900
    },
    {
      "epoch": 1.5,
      "eval_accuracy": 0.7641602526149596,
      "eval_loss": 0.6623676419258118,
      "eval_runtime": 91.2699,
      "eval_samples_per_second": 55.517,
      "eval_steps_per_second": 6.946,
      "step": 1900
    },
    {
      "epoch": 1.51,
      "grad_norm": 2.2992420196533203,
      "learning_rate": 0.00012462509865824784,
      "loss": 0.554,
      "step": 1910
    },
    {
      "epoch": 1.52,
      "grad_norm": 2.778937339782715,
      "learning_rate": 0.00012423046566692976,
      "loss": 0.6068,
      "step": 1920
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.7722057104110718,
      "learning_rate": 0.00012383583267561169,
      "loss": 0.6889,
      "step": 1930
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0860401391983032,
      "learning_rate": 0.0001234411996842936,
      "loss": 0.5685,
      "step": 1940
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.8181560039520264,
      "learning_rate": 0.00012304656669297553,
      "loss": 0.4731,
      "step": 1950
    },
    {
      "epoch": 1.55,
      "grad_norm": 2.5842840671539307,
      "learning_rate": 0.00012265193370165746,
      "loss": 0.6794,
      "step": 1960
    },
    {
      "epoch": 1.55,
      "grad_norm": 2.0809130668640137,
      "learning_rate": 0.00012225730071033938,
      "loss": 0.602,
      "step": 1970
    },
    {
      "epoch": 1.56,
      "grad_norm": 4.095306396484375,
      "learning_rate": 0.0001218626677190213,
      "loss": 0.5275,
      "step": 1980
    },
    {
      "epoch": 1.57,
      "grad_norm": 2.1503665447235107,
      "learning_rate": 0.00012146803472770325,
      "loss": 0.5502,
      "step": 1990
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.8762742280960083,
      "learning_rate": 0.00012107340173638518,
      "loss": 0.5797,
      "step": 2000
    },
    {
      "epoch": 1.58,
      "eval_accuracy": 0.7708703374777975,
      "eval_loss": 0.6286503076553345,
      "eval_runtime": 91.5869,
      "eval_samples_per_second": 55.325,
      "eval_steps_per_second": 6.922,
      "step": 2000
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.8651785850524902,
      "learning_rate": 0.0001206787687450671,
      "loss": 0.4879,
      "step": 2010
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.7065809965133667,
      "learning_rate": 0.00012028413575374901,
      "loss": 0.4858,
      "step": 2020
    },
    {
      "epoch": 1.6,
      "grad_norm": 2.897512435913086,
      "learning_rate": 0.00011988950276243093,
      "loss": 0.5983,
      "step": 2030
    },
    {
      "epoch": 1.61,
      "grad_norm": 2.6783411502838135,
      "learning_rate": 0.00011949486977111286,
      "loss": 0.475,
      "step": 2040
    },
    {
      "epoch": 1.62,
      "grad_norm": 3.3606057167053223,
      "learning_rate": 0.00011910023677979481,
      "loss": 0.6417,
      "step": 2050
    },
    {
      "epoch": 1.63,
      "grad_norm": 2.4237632751464844,
      "learning_rate": 0.00011870560378847673,
      "loss": 0.543,
      "step": 2060
    },
    {
      "epoch": 1.63,
      "grad_norm": 2.7784054279327393,
      "learning_rate": 0.00011831097079715866,
      "loss": 0.6355,
      "step": 2070
    },
    {
      "epoch": 1.64,
      "grad_norm": 2.962653398513794,
      "learning_rate": 0.00011791633780584057,
      "loss": 0.6289,
      "step": 2080
    },
    {
      "epoch": 1.65,
      "grad_norm": 2.9475271701812744,
      "learning_rate": 0.00011752170481452249,
      "loss": 0.4379,
      "step": 2090
    },
    {
      "epoch": 1.66,
      "grad_norm": 2.1250455379486084,
      "learning_rate": 0.00011712707182320441,
      "loss": 0.6018,
      "step": 2100
    },
    {
      "epoch": 1.66,
      "eval_accuracy": 0.7621866982435366,
      "eval_loss": 0.6536552309989929,
      "eval_runtime": 91.6935,
      "eval_samples_per_second": 55.26,
      "eval_steps_per_second": 6.914,
      "step": 2100
    },
    {
      "epoch": 1.67,
      "grad_norm": 3.269709348678589,
      "learning_rate": 0.00011673243883188636,
      "loss": 0.6077,
      "step": 2110
    },
    {
      "epoch": 1.67,
      "grad_norm": 2.2141342163085938,
      "learning_rate": 0.00011633780584056829,
      "loss": 0.543,
      "step": 2120
    },
    {
      "epoch": 1.68,
      "grad_norm": 3.65494966506958,
      "learning_rate": 0.00011594317284925021,
      "loss": 0.5,
      "step": 2130
    },
    {
      "epoch": 1.69,
      "grad_norm": 2.4164280891418457,
      "learning_rate": 0.00011554853985793212,
      "loss": 0.5186,
      "step": 2140
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.6607398986816406,
      "learning_rate": 0.00011515390686661404,
      "loss": 0.5087,
      "step": 2150
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.4734584093093872,
      "learning_rate": 0.00011475927387529597,
      "loss": 0.5642,
      "step": 2160
    },
    {
      "epoch": 1.71,
      "grad_norm": 2.652514934539795,
      "learning_rate": 0.00011436464088397792,
      "loss": 0.5276,
      "step": 2170
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.7451727390289307,
      "learning_rate": 0.00011397000789265984,
      "loss": 0.4694,
      "step": 2180
    },
    {
      "epoch": 1.73,
      "grad_norm": 2.4469006061553955,
      "learning_rate": 0.00011357537490134175,
      "loss": 0.611,
      "step": 2190
    },
    {
      "epoch": 1.74,
      "grad_norm": 2.6196885108947754,
      "learning_rate": 0.00011318074191002367,
      "loss": 0.6334,
      "step": 2200
    },
    {
      "epoch": 1.74,
      "eval_accuracy": 0.7712650483520821,
      "eval_loss": 0.6413267254829407,
      "eval_runtime": 91.8747,
      "eval_samples_per_second": 55.151,
      "eval_steps_per_second": 6.901,
      "step": 2200
    },
    {
      "epoch": 1.74,
      "grad_norm": 3.294318437576294,
      "learning_rate": 0.0001127861089187056,
      "loss": 0.5822,
      "step": 2210
    },
    {
      "epoch": 1.75,
      "grad_norm": 3.1869616508483887,
      "learning_rate": 0.00011239147592738752,
      "loss": 0.5093,
      "step": 2220
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.102219820022583,
      "learning_rate": 0.00011199684293606947,
      "loss": 0.6136,
      "step": 2230
    },
    {
      "epoch": 1.77,
      "grad_norm": 2.6941049098968506,
      "learning_rate": 0.0001116022099447514,
      "loss": 0.5639,
      "step": 2240
    },
    {
      "epoch": 1.78,
      "grad_norm": 2.5528244972229004,
      "learning_rate": 0.0001112075769534333,
      "loss": 0.5713,
      "step": 2250
    },
    {
      "epoch": 1.78,
      "grad_norm": 2.0887537002563477,
      "learning_rate": 0.00011081294396211523,
      "loss": 0.3819,
      "step": 2260
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9082605242729187,
      "learning_rate": 0.00011041831097079715,
      "loss": 0.6017,
      "step": 2270
    },
    {
      "epoch": 1.8,
      "grad_norm": 3.7689263820648193,
      "learning_rate": 0.0001100236779794791,
      "loss": 0.6816,
      "step": 2280
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2326387166976929,
      "learning_rate": 0.00010962904498816103,
      "loss": 0.4845,
      "step": 2290
    },
    {
      "epoch": 1.82,
      "grad_norm": 2.5216434001922607,
      "learning_rate": 0.00010923441199684295,
      "loss": 0.4111,
      "step": 2300
    },
    {
      "epoch": 1.82,
      "eval_accuracy": 0.7785671995263469,
      "eval_loss": 0.6241939663887024,
      "eval_runtime": 92.1094,
      "eval_samples_per_second": 55.011,
      "eval_steps_per_second": 6.883,
      "step": 2300
    },
    {
      "epoch": 1.82,
      "grad_norm": 3.051318407058716,
      "learning_rate": 0.00010883977900552486,
      "loss": 0.4867,
      "step": 2310
    },
    {
      "epoch": 1.83,
      "grad_norm": 2.2874298095703125,
      "learning_rate": 0.00010844514601420678,
      "loss": 0.594,
      "step": 2320
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.954172134399414,
      "learning_rate": 0.00010805051302288871,
      "loss": 0.4688,
      "step": 2330
    },
    {
      "epoch": 1.85,
      "grad_norm": 2.589738607406616,
      "learning_rate": 0.00010765588003157066,
      "loss": 0.5692,
      "step": 2340
    },
    {
      "epoch": 1.85,
      "grad_norm": 2.662523031234741,
      "learning_rate": 0.00010726124704025258,
      "loss": 0.6033,
      "step": 2350
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6577619314193726,
      "learning_rate": 0.0001068666140489345,
      "loss": 0.4772,
      "step": 2360
    },
    {
      "epoch": 1.87,
      "grad_norm": 2.9956750869750977,
      "learning_rate": 0.00010647198105761642,
      "loss": 0.4724,
      "step": 2370
    },
    {
      "epoch": 1.88,
      "grad_norm": 2.4675168991088867,
      "learning_rate": 0.00010607734806629834,
      "loss": 0.5699,
      "step": 2380
    },
    {
      "epoch": 1.89,
      "grad_norm": 3.3173062801361084,
      "learning_rate": 0.00010568271507498026,
      "loss": 0.7105,
      "step": 2390
    },
    {
      "epoch": 1.89,
      "grad_norm": 2.4665205478668213,
      "learning_rate": 0.00010528808208366221,
      "loss": 0.4779,
      "step": 2400
    },
    {
      "epoch": 1.89,
      "eval_accuracy": 0.7789619104006316,
      "eval_loss": 0.6259533166885376,
      "eval_runtime": 92.0715,
      "eval_samples_per_second": 55.033,
      "eval_steps_per_second": 6.886,
      "step": 2400
    },
    {
      "epoch": 1.9,
      "grad_norm": 2.5439658164978027,
      "learning_rate": 0.00010489344909234414,
      "loss": 0.5352,
      "step": 2410
    },
    {
      "epoch": 1.91,
      "grad_norm": 2.340139627456665,
      "learning_rate": 0.00010449881610102606,
      "loss": 0.5646,
      "step": 2420
    },
    {
      "epoch": 1.92,
      "grad_norm": 2.9097132682800293,
      "learning_rate": 0.00010410418310970797,
      "loss": 0.6391,
      "step": 2430
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.203121662139893,
      "learning_rate": 0.0001037095501183899,
      "loss": 0.4706,
      "step": 2440
    },
    {
      "epoch": 1.93,
      "grad_norm": 4.017784595489502,
      "learning_rate": 0.00010331491712707182,
      "loss": 0.514,
      "step": 2450
    },
    {
      "epoch": 1.94,
      "grad_norm": 2.4876112937927246,
      "learning_rate": 0.00010292028413575377,
      "loss": 0.4246,
      "step": 2460
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.8711539506912231,
      "learning_rate": 0.00010252565114443569,
      "loss": 0.4967,
      "step": 2470
    },
    {
      "epoch": 1.96,
      "grad_norm": 2.180922508239746,
      "learning_rate": 0.0001021310181531176,
      "loss": 0.5203,
      "step": 2480
    },
    {
      "epoch": 1.97,
      "grad_norm": 2.5203304290771484,
      "learning_rate": 0.00010173638516179952,
      "loss": 0.5806,
      "step": 2490
    },
    {
      "epoch": 1.97,
      "grad_norm": 2.462113857269287,
      "learning_rate": 0.00010134175217048145,
      "loss": 0.5488,
      "step": 2500
    },
    {
      "epoch": 1.97,
      "eval_accuracy": 0.7807381093349122,
      "eval_loss": 0.6145808696746826,
      "eval_runtime": 92.3214,
      "eval_samples_per_second": 54.884,
      "eval_steps_per_second": 6.867,
      "step": 2500
    },
    {
      "epoch": 1.98,
      "grad_norm": 2.9260380268096924,
      "learning_rate": 0.00010094711917916337,
      "loss": 0.5797,
      "step": 2510
    },
    {
      "epoch": 1.99,
      "grad_norm": 2.5453708171844482,
      "learning_rate": 0.00010055248618784532,
      "loss": 0.5223,
      "step": 2520
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0558793544769287,
      "learning_rate": 0.00010015785319652725,
      "loss": 0.3824,
      "step": 2530
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.0377509593963623,
      "learning_rate": 9.976322020520916e-05,
      "loss": 0.4497,
      "step": 2540
    },
    {
      "epoch": 2.01,
      "grad_norm": 2.023796319961548,
      "learning_rate": 9.936858721389108e-05,
      "loss": 0.3679,
      "step": 2550
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.737573504447937,
      "learning_rate": 9.8973954222573e-05,
      "loss": 0.2755,
      "step": 2560
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.711060881614685,
      "learning_rate": 9.857932123125494e-05,
      "loss": 0.2844,
      "step": 2570
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0967369079589844,
      "learning_rate": 9.818468823993686e-05,
      "loss": 0.3734,
      "step": 2580
    },
    {
      "epoch": 2.04,
      "grad_norm": 4.402666091918945,
      "learning_rate": 9.779005524861879e-05,
      "loss": 0.3492,
      "step": 2590
    },
    {
      "epoch": 2.05,
      "grad_norm": 0.23064565658569336,
      "learning_rate": 9.739542225730071e-05,
      "loss": 0.3212,
      "step": 2600
    },
    {
      "epoch": 2.05,
      "eval_accuracy": 0.7706729820406553,
      "eval_loss": 0.697548508644104,
      "eval_runtime": 93.0724,
      "eval_samples_per_second": 54.441,
      "eval_steps_per_second": 6.812,
      "step": 2600
    },
    {
      "epoch": 2.06,
      "grad_norm": 2.1419265270233154,
      "learning_rate": 9.700078926598263e-05,
      "loss": 0.3762,
      "step": 2610
    },
    {
      "epoch": 2.07,
      "grad_norm": 2.5868382453918457,
      "learning_rate": 9.660615627466457e-05,
      "loss": 0.3973,
      "step": 2620
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.9295076131820679,
      "learning_rate": 9.62115232833465e-05,
      "loss": 0.3264,
      "step": 2630
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.8431601524353027,
      "learning_rate": 9.581689029202842e-05,
      "loss": 0.2892,
      "step": 2640
    },
    {
      "epoch": 2.09,
      "grad_norm": 2.848076105117798,
      "learning_rate": 9.542225730071036e-05,
      "loss": 0.3393,
      "step": 2650
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3948113918304443,
      "learning_rate": 9.502762430939227e-05,
      "loss": 0.3074,
      "step": 2660
    },
    {
      "epoch": 2.11,
      "grad_norm": 7.114987850189209,
      "learning_rate": 9.463299131807419e-05,
      "loss": 0.3326,
      "step": 2670
    },
    {
      "epoch": 2.12,
      "grad_norm": 3.4275946617126465,
      "learning_rate": 9.423835832675613e-05,
      "loss": 0.3915,
      "step": 2680
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.183617353439331,
      "learning_rate": 9.384372533543805e-05,
      "loss": 0.3841,
      "step": 2690
    },
    {
      "epoch": 2.13,
      "grad_norm": 2.8311221599578857,
      "learning_rate": 9.344909234411997e-05,
      "loss": 0.4282,
      "step": 2700
    },
    {
      "epoch": 2.13,
      "eval_accuracy": 0.7789619104006316,
      "eval_loss": 0.634354293346405,
      "eval_runtime": 93.2533,
      "eval_samples_per_second": 54.336,
      "eval_steps_per_second": 6.799,
      "step": 2700
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.839000701904297,
      "learning_rate": 9.305445935280191e-05,
      "loss": 0.2869,
      "step": 2710
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4117017984390259,
      "learning_rate": 9.265982636148382e-05,
      "loss": 0.35,
      "step": 2720
    },
    {
      "epoch": 2.15,
      "grad_norm": 6.646859645843506,
      "learning_rate": 9.226519337016574e-05,
      "loss": 0.1956,
      "step": 2730
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.4397947788238525,
      "learning_rate": 9.187056037884768e-05,
      "loss": 0.3768,
      "step": 2740
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.0888396501541138,
      "learning_rate": 9.14759273875296e-05,
      "loss": 0.2825,
      "step": 2750
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1045578718185425,
      "learning_rate": 9.108129439621153e-05,
      "loss": 0.1746,
      "step": 2760
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.6195358037948608,
      "learning_rate": 9.068666140489345e-05,
      "loss": 0.2362,
      "step": 2770
    },
    {
      "epoch": 2.19,
      "grad_norm": 2.6458139419555664,
      "learning_rate": 9.029202841357538e-05,
      "loss": 0.2583,
      "step": 2780
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.6736512184143066,
      "learning_rate": 8.98973954222573e-05,
      "loss": 0.4217,
      "step": 2790
    },
    {
      "epoch": 2.21,
      "grad_norm": 2.508667230606079,
      "learning_rate": 8.950276243093924e-05,
      "loss": 0.2822,
      "step": 2800
    },
    {
      "epoch": 2.21,
      "eval_accuracy": 0.7844878626406158,
      "eval_loss": 0.6984525322914124,
      "eval_runtime": 93.5792,
      "eval_samples_per_second": 54.147,
      "eval_steps_per_second": 6.775,
      "step": 2800
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4989869594573975,
      "learning_rate": 8.910812943962116e-05,
      "loss": 0.3101,
      "step": 2810
    },
    {
      "epoch": 2.23,
      "grad_norm": 0.30821582674980164,
      "learning_rate": 8.871349644830308e-05,
      "loss": 0.2419,
      "step": 2820
    },
    {
      "epoch": 2.23,
      "grad_norm": 5.179461479187012,
      "learning_rate": 8.8318863456985e-05,
      "loss": 0.4989,
      "step": 2830
    },
    {
      "epoch": 2.24,
      "grad_norm": 3.597341775894165,
      "learning_rate": 8.792423046566693e-05,
      "loss": 0.3628,
      "step": 2840
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.0267224311828613,
      "learning_rate": 8.752959747434885e-05,
      "loss": 0.2395,
      "step": 2850
    },
    {
      "epoch": 2.26,
      "grad_norm": 4.339228630065918,
      "learning_rate": 8.713496448303079e-05,
      "loss": 0.385,
      "step": 2860
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.7694778442382812,
      "learning_rate": 8.674033149171271e-05,
      "loss": 0.3129,
      "step": 2870
    },
    {
      "epoch": 2.27,
      "grad_norm": 3.7710163593292236,
      "learning_rate": 8.634569850039464e-05,
      "loss": 0.3818,
      "step": 2880
    },
    {
      "epoch": 2.28,
      "grad_norm": 3.384129047393799,
      "learning_rate": 8.595106550907656e-05,
      "loss": 0.2772,
      "step": 2890
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.9485032558441162,
      "learning_rate": 8.555643251775848e-05,
      "loss": 0.3003,
      "step": 2900
    },
    {
      "epoch": 2.29,
      "eval_accuracy": 0.7992895204262878,
      "eval_loss": 0.5954017639160156,
      "eval_runtime": 90.8299,
      "eval_samples_per_second": 55.786,
      "eval_steps_per_second": 6.98,
      "step": 2900
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.648350715637207,
      "learning_rate": 8.516179952644041e-05,
      "loss": 0.3255,
      "step": 2910
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.1421117782592773,
      "learning_rate": 8.476716653512235e-05,
      "loss": 0.3721,
      "step": 2920
    },
    {
      "epoch": 2.31,
      "grad_norm": 3.3461079597473145,
      "learning_rate": 8.437253354380427e-05,
      "loss": 0.3533,
      "step": 2930
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.0906074047088623,
      "learning_rate": 8.397790055248619e-05,
      "loss": 0.2805,
      "step": 2940
    },
    {
      "epoch": 2.33,
      "grad_norm": 2.1087684631347656,
      "learning_rate": 8.358326756116812e-05,
      "loss": 0.2588,
      "step": 2950
    },
    {
      "epoch": 2.34,
      "grad_norm": 4.184020042419434,
      "learning_rate": 8.318863456985004e-05,
      "loss": 0.3646,
      "step": 2960
    },
    {
      "epoch": 2.34,
      "grad_norm": 0.7979826927185059,
      "learning_rate": 8.279400157853196e-05,
      "loss": 0.2358,
      "step": 2970
    },
    {
      "epoch": 2.35,
      "grad_norm": 3.317765235900879,
      "learning_rate": 8.23993685872139e-05,
      "loss": 0.3134,
      "step": 2980
    },
    {
      "epoch": 2.36,
      "grad_norm": 4.421618938446045,
      "learning_rate": 8.200473559589582e-05,
      "loss": 0.3644,
      "step": 2990
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.3369879722595215,
      "learning_rate": 8.161010260457775e-05,
      "loss": 0.2982,
      "step": 3000
    },
    {
      "epoch": 2.37,
      "eval_accuracy": 0.7939609236234458,
      "eval_loss": 0.6156049966812134,
      "eval_runtime": 90.2416,
      "eval_samples_per_second": 56.149,
      "eval_steps_per_second": 7.026,
      "step": 3000
    },
    {
      "epoch": 2.38,
      "grad_norm": 3.2008187770843506,
      "learning_rate": 8.121546961325967e-05,
      "loss": 0.3314,
      "step": 3010
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.550990343093872,
      "learning_rate": 8.08208366219416e-05,
      "loss": 0.2522,
      "step": 3020
    },
    {
      "epoch": 2.39,
      "grad_norm": 0.7245291471481323,
      "learning_rate": 8.042620363062352e-05,
      "loss": 0.325,
      "step": 3030
    },
    {
      "epoch": 2.4,
      "grad_norm": 3.1475908756256104,
      "learning_rate": 8.003157063930545e-05,
      "loss": 0.3212,
      "step": 3040
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6802873611450195,
      "learning_rate": 7.963693764798738e-05,
      "loss": 0.2484,
      "step": 3050
    },
    {
      "epoch": 2.42,
      "grad_norm": 3.3150057792663574,
      "learning_rate": 7.92423046566693e-05,
      "loss": 0.3014,
      "step": 3060
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.040041208267212,
      "learning_rate": 7.884767166535123e-05,
      "loss": 0.324,
      "step": 3070
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5080398321151733,
      "learning_rate": 7.845303867403315e-05,
      "loss": 0.2253,
      "step": 3080
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.1280837059020996,
      "learning_rate": 7.805840568271507e-05,
      "loss": 0.2373,
      "step": 3090
    },
    {
      "epoch": 2.45,
      "grad_norm": 4.508062839508057,
      "learning_rate": 7.766377269139701e-05,
      "loss": 0.2628,
      "step": 3100
    },
    {
      "epoch": 2.45,
      "eval_accuracy": 0.7963291888691534,
      "eval_loss": 0.6317601203918457,
      "eval_runtime": 91.5531,
      "eval_samples_per_second": 55.345,
      "eval_steps_per_second": 6.925,
      "step": 3100
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.481145977973938,
      "learning_rate": 7.726913970007893e-05,
      "loss": 0.3059,
      "step": 3110
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.0478160381317139,
      "learning_rate": 7.687450670876086e-05,
      "loss": 0.2453,
      "step": 3120
    },
    {
      "epoch": 2.47,
      "grad_norm": 3.6211369037628174,
      "learning_rate": 7.647987371744278e-05,
      "loss": 0.293,
      "step": 3130
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.617844343185425,
      "learning_rate": 7.60852407261247e-05,
      "loss": 0.2081,
      "step": 3140
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.3399300575256348,
      "learning_rate": 7.569060773480663e-05,
      "loss": 0.239,
      "step": 3150
    },
    {
      "epoch": 2.49,
      "grad_norm": 5.697124481201172,
      "learning_rate": 7.529597474348856e-05,
      "loss": 0.2893,
      "step": 3160
    },
    {
      "epoch": 2.5,
      "grad_norm": 3.405291795730591,
      "learning_rate": 7.490134175217049e-05,
      "loss": 0.3245,
      "step": 3170
    },
    {
      "epoch": 2.51,
      "grad_norm": 3.2485592365264893,
      "learning_rate": 7.450670876085241e-05,
      "loss": 0.4137,
      "step": 3180
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.0946143865585327,
      "learning_rate": 7.411207576953433e-05,
      "loss": 0.1969,
      "step": 3190
    },
    {
      "epoch": 2.53,
      "grad_norm": 0.3789944052696228,
      "learning_rate": 7.371744277821626e-05,
      "loss": 0.2987,
      "step": 3200
    },
    {
      "epoch": 2.53,
      "eval_accuracy": 0.8030392737319914,
      "eval_loss": 0.6494620442390442,
      "eval_runtime": 92.4486,
      "eval_samples_per_second": 54.809,
      "eval_steps_per_second": 6.858,
      "step": 3200
    },
    {
      "epoch": 2.53,
      "grad_norm": 3.4358811378479004,
      "learning_rate": 7.332280978689818e-05,
      "loss": 0.2757,
      "step": 3210
    },
    {
      "epoch": 2.54,
      "grad_norm": 3.544767141342163,
      "learning_rate": 7.292817679558012e-05,
      "loss": 0.2126,
      "step": 3220
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.8036537170410156,
      "learning_rate": 7.253354380426204e-05,
      "loss": 0.2455,
      "step": 3230
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.1251418590545654,
      "learning_rate": 7.213891081294397e-05,
      "loss": 0.2136,
      "step": 3240
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.44752311706543,
      "learning_rate": 7.174427782162589e-05,
      "loss": 0.3819,
      "step": 3250
    },
    {
      "epoch": 2.57,
      "grad_norm": 0.8341973423957825,
      "learning_rate": 7.134964483030781e-05,
      "loss": 0.2532,
      "step": 3260
    },
    {
      "epoch": 2.58,
      "grad_norm": 3.4972732067108154,
      "learning_rate": 7.095501183898974e-05,
      "loss": 0.2948,
      "step": 3270
    },
    {
      "epoch": 2.59,
      "grad_norm": 3.1307532787323,
      "learning_rate": 7.056037884767167e-05,
      "loss": 0.2987,
      "step": 3280
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.040160894393921,
      "learning_rate": 7.01657458563536e-05,
      "loss": 0.2939,
      "step": 3290
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.5896205902099609,
      "learning_rate": 6.977111286503552e-05,
      "loss": 0.2714,
      "step": 3300
    },
    {
      "epoch": 2.6,
      "eval_accuracy": 0.8052101835405565,
      "eval_loss": 0.6017640829086304,
      "eval_runtime": 92.2458,
      "eval_samples_per_second": 54.929,
      "eval_steps_per_second": 6.873,
      "step": 3300
    },
    {
      "epoch": 2.61,
      "grad_norm": 4.136446475982666,
      "learning_rate": 6.937647987371744e-05,
      "loss": 0.2848,
      "step": 3310
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.1288126707077026,
      "learning_rate": 6.898184688239937e-05,
      "loss": 0.3115,
      "step": 3320
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.7702901363372803,
      "learning_rate": 6.858721389108129e-05,
      "loss": 0.2547,
      "step": 3330
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.859607696533203,
      "learning_rate": 6.819258089976323e-05,
      "loss": 0.2793,
      "step": 3340
    },
    {
      "epoch": 2.64,
      "grad_norm": 5.98909854888916,
      "learning_rate": 6.779794790844515e-05,
      "loss": 0.2718,
      "step": 3350
    },
    {
      "epoch": 2.65,
      "grad_norm": 4.067326545715332,
      "learning_rate": 6.740331491712708e-05,
      "loss": 0.3174,
      "step": 3360
    },
    {
      "epoch": 2.66,
      "grad_norm": 5.376368522644043,
      "learning_rate": 6.7008681925809e-05,
      "loss": 0.3494,
      "step": 3370
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.6737453937530518,
      "learning_rate": 6.661404893449092e-05,
      "loss": 0.293,
      "step": 3380
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.012742042541504,
      "learning_rate": 6.621941594317286e-05,
      "loss": 0.2291,
      "step": 3390
    },
    {
      "epoch": 2.68,
      "grad_norm": 4.339968204498291,
      "learning_rate": 6.582478295185478e-05,
      "loss": 0.3059,
      "step": 3400
    },
    {
      "epoch": 2.68,
      "eval_accuracy": 0.8077758042234063,
      "eval_loss": 0.5943508148193359,
      "eval_runtime": 92.7623,
      "eval_samples_per_second": 54.624,
      "eval_steps_per_second": 6.835,
      "step": 3400
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6194040775299072,
      "learning_rate": 6.543014996053671e-05,
      "loss": 0.3704,
      "step": 3410
    },
    {
      "epoch": 2.7,
      "grad_norm": 4.174539089202881,
      "learning_rate": 6.503551696921863e-05,
      "loss": 0.3716,
      "step": 3420
    },
    {
      "epoch": 2.71,
      "grad_norm": 4.487131118774414,
      "learning_rate": 6.464088397790055e-05,
      "loss": 0.3054,
      "step": 3430
    },
    {
      "epoch": 2.72,
      "grad_norm": 3.870863437652588,
      "learning_rate": 6.424625098658248e-05,
      "loss": 0.2811,
      "step": 3440
    },
    {
      "epoch": 2.72,
      "grad_norm": 4.740366458892822,
      "learning_rate": 6.385161799526441e-05,
      "loss": 0.2143,
      "step": 3450
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.337819814682007,
      "learning_rate": 6.345698500394634e-05,
      "loss": 0.4391,
      "step": 3460
    },
    {
      "epoch": 2.74,
      "grad_norm": 6.155264854431152,
      "learning_rate": 6.306235201262826e-05,
      "loss": 0.3491,
      "step": 3470
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.4382271766662598,
      "learning_rate": 6.266771902131019e-05,
      "loss": 0.2899,
      "step": 3480
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.5207793712615967,
      "learning_rate": 6.227308602999211e-05,
      "loss": 0.2522,
      "step": 3490
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.243215322494507,
      "learning_rate": 6.187845303867403e-05,
      "loss": 0.2762,
      "step": 3500
    },
    {
      "epoch": 2.76,
      "eval_accuracy": 0.7935662127491613,
      "eval_loss": 0.6296067237854004,
      "eval_runtime": 92.1328,
      "eval_samples_per_second": 54.997,
      "eval_steps_per_second": 6.881,
      "step": 3500
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.194077968597412,
      "learning_rate": 6.148382004735597e-05,
      "loss": 0.2279,
      "step": 3510
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.3025989532470703,
      "learning_rate": 6.108918705603789e-05,
      "loss": 0.3366,
      "step": 3520
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.9970479011535645,
      "learning_rate": 6.069455406471981e-05,
      "loss": 0.2698,
      "step": 3530
    },
    {
      "epoch": 2.79,
      "grad_norm": 3.0827438831329346,
      "learning_rate": 6.029992107340175e-05,
      "loss": 0.2195,
      "step": 3540
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.7207252979278564,
      "learning_rate": 5.9905288082083663e-05,
      "loss": 0.2549,
      "step": 3550
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.6517016887664795,
      "learning_rate": 5.951065509076559e-05,
      "loss": 0.2294,
      "step": 3560
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5923064947128296,
      "learning_rate": 5.9116022099447524e-05,
      "loss": 0.2465,
      "step": 3570
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.049023151397705,
      "learning_rate": 5.872138910812944e-05,
      "loss": 0.2712,
      "step": 3580
    },
    {
      "epoch": 2.83,
      "grad_norm": 0.7295699119567871,
      "learning_rate": 5.8326756116811364e-05,
      "loss": 0.3309,
      "step": 3590
    },
    {
      "epoch": 2.84,
      "grad_norm": 3.531484365463257,
      "learning_rate": 5.79321231254933e-05,
      "loss": 0.3685,
      "step": 3600
    },
    {
      "epoch": 2.84,
      "eval_accuracy": 0.8016577856719953,
      "eval_loss": 0.6276537179946899,
      "eval_runtime": 92.0744,
      "eval_samples_per_second": 55.032,
      "eval_steps_per_second": 6.886,
      "step": 3600
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.7010326385498047,
      "learning_rate": 5.753749013417522e-05,
      "loss": 0.1898,
      "step": 3610
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.8486061096191406,
      "learning_rate": 5.714285714285714e-05,
      "loss": 0.2965,
      "step": 3620
    },
    {
      "epoch": 2.87,
      "grad_norm": 5.980003833770752,
      "learning_rate": 5.674822415153908e-05,
      "loss": 0.297,
      "step": 3630
    },
    {
      "epoch": 2.87,
      "grad_norm": 0.3644309639930725,
      "learning_rate": 5.6353591160220996e-05,
      "loss": 0.2627,
      "step": 3640
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.6136064529418945,
      "learning_rate": 5.595895816890292e-05,
      "loss": 0.3327,
      "step": 3650
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.8576825857162476,
      "learning_rate": 5.556432517758485e-05,
      "loss": 0.2064,
      "step": 3660
    },
    {
      "epoch": 2.9,
      "grad_norm": 3.2825279235839844,
      "learning_rate": 5.516969218626677e-05,
      "loss": 0.2582,
      "step": 3670
    },
    {
      "epoch": 2.9,
      "grad_norm": 4.109126567840576,
      "learning_rate": 5.4775059194948696e-05,
      "loss": 0.2412,
      "step": 3680
    },
    {
      "epoch": 2.91,
      "grad_norm": 0.9957514405250549,
      "learning_rate": 5.438042620363063e-05,
      "loss": 0.1818,
      "step": 3690
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.8618041276931763,
      "learning_rate": 5.398579321231255e-05,
      "loss": 0.2299,
      "step": 3700
    },
    {
      "epoch": 2.92,
      "eval_accuracy": 0.8125123347148214,
      "eval_loss": 0.5834479928016663,
      "eval_runtime": 91.926,
      "eval_samples_per_second": 55.12,
      "eval_steps_per_second": 6.897,
      "step": 3700
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.655374526977539,
      "learning_rate": 5.3591160220994474e-05,
      "loss": 0.3183,
      "step": 3710
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.711798906326294,
      "learning_rate": 5.3196527229676404e-05,
      "loss": 0.3336,
      "step": 3720
    },
    {
      "epoch": 2.94,
      "grad_norm": 4.209230422973633,
      "learning_rate": 5.280189423835833e-05,
      "loss": 0.2944,
      "step": 3730
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.0983171463012695,
      "learning_rate": 5.240726124704025e-05,
      "loss": 0.2846,
      "step": 3740
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.4771692752838135,
      "learning_rate": 5.201262825572218e-05,
      "loss": 0.3192,
      "step": 3750
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.7510647773742676,
      "learning_rate": 5.1617995264404105e-05,
      "loss": 0.1813,
      "step": 3760
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.2778220176696777,
      "learning_rate": 5.122336227308603e-05,
      "loss": 0.1935,
      "step": 3770
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.6012797355651855,
      "learning_rate": 5.082872928176796e-05,
      "loss": 0.2111,
      "step": 3780
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.29512357711792,
      "learning_rate": 5.043409629044988e-05,
      "loss": 0.2987,
      "step": 3790
    },
    {
      "epoch": 3.0,
      "grad_norm": 5.790306568145752,
      "learning_rate": 5.0039463299131806e-05,
      "loss": 0.3414,
      "step": 3800
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8166567988948096,
      "eval_loss": 0.5749832987785339,
      "eval_runtime": 92.3841,
      "eval_samples_per_second": 54.847,
      "eval_steps_per_second": 6.863,
      "step": 3800
    },
    {
      "epoch": 3.01,
      "grad_norm": 4.339576721191406,
      "learning_rate": 4.9644830307813736e-05,
      "loss": 0.1144,
      "step": 3810
    },
    {
      "epoch": 3.01,
      "grad_norm": 0.7903636693954468,
      "learning_rate": 4.925019731649566e-05,
      "loss": 0.0767,
      "step": 3820
    },
    {
      "epoch": 3.02,
      "grad_norm": 2.8627429008483887,
      "learning_rate": 4.885556432517759e-05,
      "loss": 0.0859,
      "step": 3830
    },
    {
      "epoch": 3.03,
      "grad_norm": 4.783523082733154,
      "learning_rate": 4.8460931333859514e-05,
      "loss": 0.1071,
      "step": 3840
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.7647588849067688,
      "learning_rate": 4.806629834254144e-05,
      "loss": 0.1531,
      "step": 3850
    },
    {
      "epoch": 3.05,
      "grad_norm": 2.228163719177246,
      "learning_rate": 4.767166535122337e-05,
      "loss": 0.1582,
      "step": 3860
    },
    {
      "epoch": 3.05,
      "grad_norm": 0.918712854385376,
      "learning_rate": 4.727703235990529e-05,
      "loss": 0.1197,
      "step": 3870
    },
    {
      "epoch": 3.06,
      "grad_norm": 4.140162467956543,
      "learning_rate": 4.6882399368587215e-05,
      "loss": 0.0733,
      "step": 3880
    },
    {
      "epoch": 3.07,
      "grad_norm": 2.8325231075286865,
      "learning_rate": 4.6487766377269145e-05,
      "loss": 0.1651,
      "step": 3890
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.49401429295539856,
      "learning_rate": 4.609313338595106e-05,
      "loss": 0.1082,
      "step": 3900
    },
    {
      "epoch": 3.08,
      "eval_accuracy": 0.819617130451944,
      "eval_loss": 0.6201203465461731,
      "eval_runtime": 92.3164,
      "eval_samples_per_second": 54.887,
      "eval_steps_per_second": 6.868,
      "step": 3900
    },
    {
      "epoch": 3.09,
      "grad_norm": 0.34822145104408264,
      "learning_rate": 4.569850039463299e-05,
      "loss": 0.0457,
      "step": 3910
    },
    {
      "epoch": 3.09,
      "grad_norm": 1.0006756782531738,
      "learning_rate": 4.530386740331492e-05,
      "loss": 0.1194,
      "step": 3920
    },
    {
      "epoch": 3.1,
      "grad_norm": 0.7993775010108948,
      "learning_rate": 4.4909234411996846e-05,
      "loss": 0.1325,
      "step": 3930
    },
    {
      "epoch": 3.11,
      "grad_norm": 6.907358646392822,
      "learning_rate": 4.451460142067877e-05,
      "loss": 0.0605,
      "step": 3940
    },
    {
      "epoch": 3.12,
      "grad_norm": 1.2168715000152588,
      "learning_rate": 4.41199684293607e-05,
      "loss": 0.0968,
      "step": 3950
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.48231208324432373,
      "learning_rate": 4.372533543804262e-05,
      "loss": 0.0868,
      "step": 3960
    },
    {
      "epoch": 3.13,
      "grad_norm": 0.09531491994857788,
      "learning_rate": 4.333070244672455e-05,
      "loss": 0.1161,
      "step": 3970
    },
    {
      "epoch": 3.14,
      "grad_norm": 3.662717342376709,
      "learning_rate": 4.293606945540648e-05,
      "loss": 0.1188,
      "step": 3980
    },
    {
      "epoch": 3.15,
      "grad_norm": 8.440380096435547,
      "learning_rate": 4.25414364640884e-05,
      "loss": 0.1319,
      "step": 3990
    },
    {
      "epoch": 3.16,
      "grad_norm": 2.019827127456665,
      "learning_rate": 4.2146803472770324e-05,
      "loss": 0.049,
      "step": 4000
    },
    {
      "epoch": 3.16,
      "eval_accuracy": 0.8160647325833826,
      "eval_loss": 0.6474657654762268,
      "eval_runtime": 92.4783,
      "eval_samples_per_second": 54.791,
      "eval_steps_per_second": 6.856,
      "step": 4000
    },
    {
      "epoch": 3.16,
      "grad_norm": 4.42202615737915,
      "learning_rate": 4.1752170481452254e-05,
      "loss": 0.0794,
      "step": 4010
    },
    {
      "epoch": 3.17,
      "grad_norm": 4.54102087020874,
      "learning_rate": 4.135753749013418e-05,
      "loss": 0.1152,
      "step": 4020
    },
    {
      "epoch": 3.18,
      "grad_norm": 0.31936466693878174,
      "learning_rate": 4.09629044988161e-05,
      "loss": 0.0945,
      "step": 4030
    },
    {
      "epoch": 3.19,
      "grad_norm": 0.12120810896158218,
      "learning_rate": 4.056827150749803e-05,
      "loss": 0.1196,
      "step": 4040
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.3144413232803345,
      "learning_rate": 4.0173638516179955e-05,
      "loss": 0.1226,
      "step": 4050
    },
    {
      "epoch": 3.2,
      "grad_norm": 5.312731742858887,
      "learning_rate": 3.977900552486188e-05,
      "loss": 0.1807,
      "step": 4060
    },
    {
      "epoch": 3.21,
      "grad_norm": 0.05327579751610756,
      "learning_rate": 3.938437253354381e-05,
      "loss": 0.1521,
      "step": 4070
    },
    {
      "epoch": 3.22,
      "grad_norm": 1.8903406858444214,
      "learning_rate": 3.898973954222573e-05,
      "loss": 0.1712,
      "step": 4080
    },
    {
      "epoch": 3.23,
      "grad_norm": 0.07272280752658844,
      "learning_rate": 3.8595106550907656e-05,
      "loss": 0.0731,
      "step": 4090
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.41295164823532104,
      "learning_rate": 3.8200473559589587e-05,
      "loss": 0.102,
      "step": 4100
    },
    {
      "epoch": 3.24,
      "eval_accuracy": 0.8097493585948293,
      "eval_loss": 0.6791070103645325,
      "eval_runtime": 92.1285,
      "eval_samples_per_second": 54.999,
      "eval_steps_per_second": 6.882,
      "step": 4100
    },
    {
      "epoch": 3.24,
      "grad_norm": 5.161396026611328,
      "learning_rate": 3.780584056827151e-05,
      "loss": 0.0824,
      "step": 4110
    },
    {
      "epoch": 3.25,
      "grad_norm": 0.2593359053134918,
      "learning_rate": 3.7411207576953434e-05,
      "loss": 0.1567,
      "step": 4120
    },
    {
      "epoch": 3.26,
      "grad_norm": 1.5720149278640747,
      "learning_rate": 3.7016574585635364e-05,
      "loss": 0.1422,
      "step": 4130
    },
    {
      "epoch": 3.27,
      "grad_norm": 6.186444282531738,
      "learning_rate": 3.662194159431729e-05,
      "loss": 0.0811,
      "step": 4140
    },
    {
      "epoch": 3.28,
      "grad_norm": 1.9182838201522827,
      "learning_rate": 3.622730860299921e-05,
      "loss": 0.1454,
      "step": 4150
    },
    {
      "epoch": 3.28,
      "grad_norm": 3.1142055988311768,
      "learning_rate": 3.583267561168114e-05,
      "loss": 0.1304,
      "step": 4160
    },
    {
      "epoch": 3.29,
      "grad_norm": 0.17880991101264954,
      "learning_rate": 3.5438042620363065e-05,
      "loss": 0.0791,
      "step": 4170
    },
    {
      "epoch": 3.3,
      "grad_norm": 3.2896628379821777,
      "learning_rate": 3.504340962904499e-05,
      "loss": 0.0903,
      "step": 4180
    },
    {
      "epoch": 3.31,
      "grad_norm": 3.1752073764801025,
      "learning_rate": 3.464877663772691e-05,
      "loss": 0.0988,
      "step": 4190
    },
    {
      "epoch": 3.31,
      "grad_norm": 1.282556414604187,
      "learning_rate": 3.425414364640884e-05,
      "loss": 0.0483,
      "step": 4200
    },
    {
      "epoch": 3.31,
      "eval_accuracy": 0.8215906848233668,
      "eval_loss": 0.6581845283508301,
      "eval_runtime": 92.2168,
      "eval_samples_per_second": 54.947,
      "eval_steps_per_second": 6.875,
      "step": 4200
    },
    {
      "epoch": 3.32,
      "grad_norm": 5.438603401184082,
      "learning_rate": 3.3859510655090766e-05,
      "loss": 0.084,
      "step": 4210
    },
    {
      "epoch": 3.33,
      "grad_norm": 0.6808891296386719,
      "learning_rate": 3.346487766377269e-05,
      "loss": 0.1193,
      "step": 4220
    },
    {
      "epoch": 3.34,
      "grad_norm": 0.4623468220233917,
      "learning_rate": 3.307024467245462e-05,
      "loss": 0.0575,
      "step": 4230
    },
    {
      "epoch": 3.35,
      "grad_norm": 0.5891864895820618,
      "learning_rate": 3.267561168113654e-05,
      "loss": 0.0353,
      "step": 4240
    },
    {
      "epoch": 3.35,
      "grad_norm": 3.8541829586029053,
      "learning_rate": 3.2280978689818467e-05,
      "loss": 0.1207,
      "step": 4250
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.4876788854598999,
      "learning_rate": 3.18863456985004e-05,
      "loss": 0.0386,
      "step": 4260
    },
    {
      "epoch": 3.37,
      "grad_norm": 1.7822741270065308,
      "learning_rate": 3.149171270718232e-05,
      "loss": 0.0753,
      "step": 4270
    },
    {
      "epoch": 3.38,
      "grad_norm": 0.6429558396339417,
      "learning_rate": 3.1097079715864244e-05,
      "loss": 0.0799,
      "step": 4280
    },
    {
      "epoch": 3.39,
      "grad_norm": 6.067298889160156,
      "learning_rate": 3.0702446724546174e-05,
      "loss": 0.1465,
      "step": 4290
    },
    {
      "epoch": 3.39,
      "grad_norm": 3.5951452255249023,
      "learning_rate": 3.03078137332281e-05,
      "loss": 0.1204,
      "step": 4300
    },
    {
      "epoch": 3.39,
      "eval_accuracy": 0.8221827511347938,
      "eval_loss": 0.6602674722671509,
      "eval_runtime": 91.9479,
      "eval_samples_per_second": 55.107,
      "eval_steps_per_second": 6.895,
      "step": 4300
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.12355990707874298,
      "learning_rate": 2.9913180741910025e-05,
      "loss": 0.0873,
      "step": 4310
    },
    {
      "epoch": 3.41,
      "grad_norm": 1.412264347076416,
      "learning_rate": 2.951854775059195e-05,
      "loss": 0.0222,
      "step": 4320
    },
    {
      "epoch": 3.42,
      "grad_norm": 0.25506913661956787,
      "learning_rate": 2.912391475927388e-05,
      "loss": 0.049,
      "step": 4330
    },
    {
      "epoch": 3.43,
      "grad_norm": 1.5517828464508057,
      "learning_rate": 2.8729281767955802e-05,
      "loss": 0.0323,
      "step": 4340
    },
    {
      "epoch": 3.43,
      "grad_norm": 9.105844497680664,
      "learning_rate": 2.833464877663773e-05,
      "loss": 0.1014,
      "step": 4350
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.06447356194257736,
      "learning_rate": 2.7940015785319656e-05,
      "loss": 0.049,
      "step": 4360
    },
    {
      "epoch": 3.45,
      "grad_norm": 6.344263076782227,
      "learning_rate": 2.754538279400158e-05,
      "loss": 0.0841,
      "step": 4370
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.45002487301826477,
      "learning_rate": 2.7150749802683506e-05,
      "loss": 0.1012,
      "step": 4380
    },
    {
      "epoch": 3.46,
      "grad_norm": 0.10588052868843079,
      "learning_rate": 2.6756116811365433e-05,
      "loss": 0.0076,
      "step": 4390
    },
    {
      "epoch": 3.47,
      "grad_norm": 3.3592453002929688,
      "learning_rate": 2.6361483820047357e-05,
      "loss": 0.0611,
      "step": 4400
    },
    {
      "epoch": 3.47,
      "eval_accuracy": 0.8190250641405171,
      "eval_loss": 0.717377781867981,
      "eval_runtime": 91.8985,
      "eval_samples_per_second": 55.137,
      "eval_steps_per_second": 6.899,
      "step": 4400
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.4130847156047821,
      "learning_rate": 2.5966850828729284e-05,
      "loss": 0.1135,
      "step": 4410
    },
    {
      "epoch": 3.49,
      "grad_norm": 4.881414890289307,
      "learning_rate": 2.557221783741121e-05,
      "loss": 0.1324,
      "step": 4420
    },
    {
      "epoch": 3.5,
      "grad_norm": 2.830336332321167,
      "learning_rate": 2.5177584846093134e-05,
      "loss": 0.0686,
      "step": 4430
    },
    {
      "epoch": 3.5,
      "grad_norm": 0.41728538274765015,
      "learning_rate": 2.478295185477506e-05,
      "loss": 0.0649,
      "step": 4440
    },
    {
      "epoch": 3.51,
      "grad_norm": 2.339663505554199,
      "learning_rate": 2.4388318863456985e-05,
      "loss": 0.1187,
      "step": 4450
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.06545936316251755,
      "learning_rate": 2.399368587213891e-05,
      "loss": 0.0995,
      "step": 4460
    },
    {
      "epoch": 3.53,
      "grad_norm": 0.9462890625,
      "learning_rate": 2.359905288082084e-05,
      "loss": 0.0796,
      "step": 4470
    },
    {
      "epoch": 3.54,
      "grad_norm": 3.4971237182617188,
      "learning_rate": 2.3204419889502762e-05,
      "loss": 0.062,
      "step": 4480
    },
    {
      "epoch": 3.54,
      "grad_norm": 3.370915412902832,
      "learning_rate": 2.280978689818469e-05,
      "loss": 0.1375,
      "step": 4490
    },
    {
      "epoch": 3.55,
      "grad_norm": 0.047411490231752396,
      "learning_rate": 2.2415153906866616e-05,
      "loss": 0.0555,
      "step": 4500
    },
    {
      "epoch": 3.55,
      "eval_accuracy": 0.8235642391947898,
      "eval_loss": 0.6841058135032654,
      "eval_runtime": 92.265,
      "eval_samples_per_second": 54.918,
      "eval_steps_per_second": 6.872,
      "step": 4500
    },
    {
      "epoch": 3.56,
      "grad_norm": 9.102856636047363,
      "learning_rate": 2.202052091554854e-05,
      "loss": 0.0936,
      "step": 4510
    },
    {
      "epoch": 3.57,
      "grad_norm": 1.5080801248550415,
      "learning_rate": 2.1625887924230466e-05,
      "loss": 0.0577,
      "step": 4520
    },
    {
      "epoch": 3.58,
      "grad_norm": 1.3888694047927856,
      "learning_rate": 2.1231254932912393e-05,
      "loss": 0.1027,
      "step": 4530
    },
    {
      "epoch": 3.58,
      "grad_norm": 3.6194889545440674,
      "learning_rate": 2.0836621941594317e-05,
      "loss": 0.0755,
      "step": 4540
    },
    {
      "epoch": 3.59,
      "grad_norm": 0.7071271538734436,
      "learning_rate": 2.0441988950276244e-05,
      "loss": 0.0276,
      "step": 4550
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.07247108966112137,
      "learning_rate": 2.004735595895817e-05,
      "loss": 0.1272,
      "step": 4560
    },
    {
      "epoch": 3.61,
      "grad_norm": 1.932975172996521,
      "learning_rate": 1.9652722967640098e-05,
      "loss": 0.1341,
      "step": 4570
    },
    {
      "epoch": 3.61,
      "grad_norm": 0.6965919137001038,
      "learning_rate": 1.925808997632202e-05,
      "loss": 0.0614,
      "step": 4580
    },
    {
      "epoch": 3.62,
      "grad_norm": 3.4213573932647705,
      "learning_rate": 1.8863456985003948e-05,
      "loss": 0.0764,
      "step": 4590
    },
    {
      "epoch": 3.63,
      "grad_norm": 0.1402810513973236,
      "learning_rate": 1.8468823993685875e-05,
      "loss": 0.0188,
      "step": 4600
    },
    {
      "epoch": 3.63,
      "eval_accuracy": 0.8239589500690744,
      "eval_loss": 0.7009280920028687,
      "eval_runtime": 92.5146,
      "eval_samples_per_second": 54.77,
      "eval_steps_per_second": 6.853,
      "step": 4600
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.07443057000637054,
      "learning_rate": 1.80741910023678e-05,
      "loss": 0.0846,
      "step": 4610
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.20932039618492126,
      "learning_rate": 1.7679558011049725e-05,
      "loss": 0.0747,
      "step": 4620
    },
    {
      "epoch": 3.65,
      "grad_norm": 0.04756965488195419,
      "learning_rate": 1.7284925019731652e-05,
      "loss": 0.0628,
      "step": 4630
    },
    {
      "epoch": 3.66,
      "grad_norm": 13.026732444763184,
      "learning_rate": 1.6890292028413576e-05,
      "loss": 0.1348,
      "step": 4640
    },
    {
      "epoch": 3.67,
      "grad_norm": 0.08388219773769379,
      "learning_rate": 1.64956590370955e-05,
      "loss": 0.0186,
      "step": 4650
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.08429060131311417,
      "learning_rate": 1.610102604577743e-05,
      "loss": 0.0342,
      "step": 4660
    },
    {
      "epoch": 3.69,
      "grad_norm": 0.22585107386112213,
      "learning_rate": 1.5706393054459353e-05,
      "loss": 0.0371,
      "step": 4670
    },
    {
      "epoch": 3.69,
      "grad_norm": 8.003403663635254,
      "learning_rate": 1.5311760063141277e-05,
      "loss": 0.1692,
      "step": 4680
    },
    {
      "epoch": 3.7,
      "grad_norm": 1.522824764251709,
      "learning_rate": 1.4917127071823205e-05,
      "loss": 0.0612,
      "step": 4690
    },
    {
      "epoch": 3.71,
      "grad_norm": 2.2803661823272705,
      "learning_rate": 1.452249408050513e-05,
      "loss": 0.1292,
      "step": 4700
    },
    {
      "epoch": 3.71,
      "eval_accuracy": 0.8204065522005132,
      "eval_loss": 0.703969419002533,
      "eval_runtime": 92.794,
      "eval_samples_per_second": 54.605,
      "eval_steps_per_second": 6.832,
      "step": 4700
    },
    {
      "epoch": 3.72,
      "grad_norm": 6.225048065185547,
      "learning_rate": 1.4127861089187056e-05,
      "loss": 0.0557,
      "step": 4710
    },
    {
      "epoch": 3.73,
      "grad_norm": 0.4492289423942566,
      "learning_rate": 1.3733228097868983e-05,
      "loss": 0.1578,
      "step": 4720
    },
    {
      "epoch": 3.73,
      "grad_norm": 1.797982096672058,
      "learning_rate": 1.3338595106550908e-05,
      "loss": 0.019,
      "step": 4730
    },
    {
      "epoch": 3.74,
      "grad_norm": 0.4045000970363617,
      "learning_rate": 1.2943962115232833e-05,
      "loss": 0.0614,
      "step": 4740
    },
    {
      "epoch": 3.75,
      "grad_norm": 0.08240614086389542,
      "learning_rate": 1.254932912391476e-05,
      "loss": 0.1421,
      "step": 4750
    },
    {
      "epoch": 3.76,
      "grad_norm": 4.642998695373535,
      "learning_rate": 1.2154696132596685e-05,
      "loss": 0.0562,
      "step": 4760
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.16091428697109222,
      "learning_rate": 1.176006314127861e-05,
      "loss": 0.0638,
      "step": 4770
    },
    {
      "epoch": 3.77,
      "grad_norm": 1.0345463752746582,
      "learning_rate": 1.1365430149960538e-05,
      "loss": 0.0634,
      "step": 4780
    },
    {
      "epoch": 3.78,
      "grad_norm": 0.042663924396038055,
      "learning_rate": 1.0970797158642463e-05,
      "loss": 0.1053,
      "step": 4790
    },
    {
      "epoch": 3.79,
      "grad_norm": 1.7890504598617554,
      "learning_rate": 1.0576164167324388e-05,
      "loss": 0.0661,
      "step": 4800
    },
    {
      "epoch": 3.79,
      "eval_accuracy": 0.8237615946319321,
      "eval_loss": 0.7073862552642822,
      "eval_runtime": 91.8421,
      "eval_samples_per_second": 55.171,
      "eval_steps_per_second": 6.903,
      "step": 4800
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.08139993995428085,
      "learning_rate": 1.0181531176006315e-05,
      "loss": 0.0623,
      "step": 4810
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.02715960517525673,
      "learning_rate": 9.786898184688242e-06,
      "loss": 0.0532,
      "step": 4820
    },
    {
      "epoch": 3.81,
      "grad_norm": 5.507246494293213,
      "learning_rate": 9.392265193370165e-06,
      "loss": 0.0829,
      "step": 4830
    },
    {
      "epoch": 3.82,
      "grad_norm": 0.11962206661701202,
      "learning_rate": 8.997632202052092e-06,
      "loss": 0.0887,
      "step": 4840
    },
    {
      "epoch": 3.83,
      "grad_norm": 0.14435319602489471,
      "learning_rate": 8.602999210734018e-06,
      "loss": 0.0478,
      "step": 4850
    },
    {
      "epoch": 3.84,
      "grad_norm": 4.151905536651611,
      "learning_rate": 8.208366219415943e-06,
      "loss": 0.1795,
      "step": 4860
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.10827145725488663,
      "learning_rate": 7.81373322809787e-06,
      "loss": 0.0566,
      "step": 4870
    },
    {
      "epoch": 3.85,
      "grad_norm": 0.5536698698997498,
      "learning_rate": 7.419100236779796e-06,
      "loss": 0.0602,
      "step": 4880
    },
    {
      "epoch": 3.86,
      "grad_norm": 0.1482585370540619,
      "learning_rate": 7.02446724546172e-06,
      "loss": 0.0995,
      "step": 4890
    },
    {
      "epoch": 3.87,
      "grad_norm": 0.6495879888534546,
      "learning_rate": 6.629834254143646e-06,
      "loss": 0.1061,
      "step": 4900
    },
    {
      "epoch": 3.87,
      "eval_accuracy": 0.82099861851194,
      "eval_loss": 0.6983892321586609,
      "eval_runtime": 92.4464,
      "eval_samples_per_second": 54.81,
      "eval_steps_per_second": 6.858,
      "step": 4900
    },
    {
      "epoch": 3.88,
      "grad_norm": 1.4518879652023315,
      "learning_rate": 6.235201262825572e-06,
      "loss": 0.0968,
      "step": 4910
    },
    {
      "epoch": 3.88,
      "grad_norm": 2.563098669052124,
      "learning_rate": 5.840568271507498e-06,
      "loss": 0.0877,
      "step": 4920
    },
    {
      "epoch": 3.89,
      "grad_norm": 5.312118053436279,
      "learning_rate": 5.445935280189424e-06,
      "loss": 0.1662,
      "step": 4930
    },
    {
      "epoch": 3.9,
      "grad_norm": 2.835618257522583,
      "learning_rate": 5.05130228887135e-06,
      "loss": 0.0757,
      "step": 4940
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.9459186792373657,
      "learning_rate": 4.656669297553276e-06,
      "loss": 0.0844,
      "step": 4950
    },
    {
      "epoch": 3.91,
      "grad_norm": 1.6886509656906128,
      "learning_rate": 4.262036306235202e-06,
      "loss": 0.1373,
      "step": 4960
    },
    {
      "epoch": 3.92,
      "grad_norm": 9.272232055664062,
      "learning_rate": 3.867403314917127e-06,
      "loss": 0.0566,
      "step": 4970
    },
    {
      "epoch": 3.93,
      "grad_norm": 0.17531900107860565,
      "learning_rate": 3.472770323599053e-06,
      "loss": 0.0691,
      "step": 4980
    },
    {
      "epoch": 3.94,
      "grad_norm": 0.08186766505241394,
      "learning_rate": 3.0781373322809787e-06,
      "loss": 0.0586,
      "step": 4990
    },
    {
      "epoch": 3.95,
      "grad_norm": 3.012854814529419,
      "learning_rate": 2.683504340962905e-06,
      "loss": 0.0861,
      "step": 5000
    },
    {
      "epoch": 3.95,
      "eval_accuracy": 0.8229721728833629,
      "eval_loss": 0.6912640929222107,
      "eval_runtime": 91.3989,
      "eval_samples_per_second": 55.438,
      "eval_steps_per_second": 6.937,
      "step": 5000
    },
    {
      "epoch": 3.95,
      "grad_norm": 0.05941695719957352,
      "learning_rate": 2.2888713496448305e-06,
      "loss": 0.0622,
      "step": 5010
    },
    {
      "epoch": 3.96,
      "grad_norm": 7.365468502044678,
      "learning_rate": 1.8942383583267563e-06,
      "loss": 0.1702,
      "step": 5020
    },
    {
      "epoch": 3.97,
      "grad_norm": 2.009979724884033,
      "learning_rate": 1.499605367008682e-06,
      "loss": 0.0164,
      "step": 5030
    },
    {
      "epoch": 3.98,
      "grad_norm": 4.683029651641846,
      "learning_rate": 1.1049723756906078e-06,
      "loss": 0.0963,
      "step": 5040
    },
    {
      "epoch": 3.99,
      "grad_norm": 8.475516319274902,
      "learning_rate": 7.103393843725336e-07,
      "loss": 0.071,
      "step": 5050
    },
    {
      "epoch": 3.99,
      "grad_norm": 0.3904523253440857,
      "learning_rate": 3.1570639305445935e-07,
      "loss": 0.0701,
      "step": 5060
    },
    {
      "epoch": 4.0,
      "step": 5068,
      "total_flos": 6.281528488153842e+18,
      "train_loss": 0.4664627312864129,
      "train_runtime": 8576.6293,
      "train_samples_per_second": 9.451,
      "train_steps_per_second": 0.591
    }
  ],
  "logging_steps": 10,
  "max_steps": 5068,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 100,
  "total_flos": 6.281528488153842e+18,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}