{
  "best_metric": 0.3352562189102173,
  "best_model_checkpoint": "./exper_batch_8_e4/checkpoint-5100",
  "epoch": 4.0,
  "global_step": 5108,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 0.0001996084573218481,
      "loss": 4.9435,
      "step": 10
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019921691464369617,
      "loss": 4.8897,
      "step": 20
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00019882537196554424,
      "loss": 4.8087,
      "step": 30
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00019843382928739234,
      "loss": 4.7267,
      "step": 40
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019804228660924044,
      "loss": 4.6033,
      "step": 50
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019765074393108848,
      "loss": 4.5838,
      "step": 60
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00019725920125293657,
      "loss": 4.4214,
      "step": 70
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00019686765857478467,
      "loss": 4.2695,
      "step": 80
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00019647611589663274,
      "loss": 4.2364,
      "step": 90
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0001960845732184808,
      "loss": 4.2251,
      "step": 100
    },
    {
      "epoch": 0.08,
      "eval_accuracy": 0.1202775636083269,
      "eval_loss": 4.150820732116699,
      "eval_runtime": 42.3584,
      "eval_samples_per_second": 61.239,
      "eval_steps_per_second": 7.673,
      "step": 100
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0001956930305403289,
      "loss": 4.0458,
      "step": 110
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.000195301487862177,
      "loss": 4.1893,
      "step": 120
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.00019490994518402507,
      "loss": 4.0755,
      "step": 130
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019451840250587314,
      "loss": 3.8684,
      "step": 140
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.00019412685982772123,
      "loss": 3.8903,
      "step": 150
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00019373531714956933,
      "loss": 3.9097,
      "step": 160
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0001933437744714174,
      "loss": 3.8331,
      "step": 170
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.00019295223179326547,
      "loss": 3.924,
      "step": 180
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00019256068911511356,
      "loss": 3.6402,
      "step": 190
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00019216914643696163,
      "loss": 3.4942,
      "step": 200
    },
    {
      "epoch": 0.16,
      "eval_accuracy": 0.2081727062451812,
      "eval_loss": 3.5565712451934814,
      "eval_runtime": 41.4085,
      "eval_samples_per_second": 62.644,
      "eval_steps_per_second": 7.849,
      "step": 200
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0001917776037588097,
      "loss": 3.5131,
      "step": 210
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0001913860610806578,
      "loss": 3.6997,
      "step": 220
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0001909945184025059,
      "loss": 3.4877,
      "step": 230
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00019060297572435396,
      "loss": 3.4052,
      "step": 240
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00019021143304620203,
      "loss": 3.3865,
      "step": 250
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00018981989036805013,
      "loss": 3.3681,
      "step": 260
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00018942834768989822,
      "loss": 3.2912,
      "step": 270
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0001890368050117463,
      "loss": 3.1632,
      "step": 280
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00018864526233359436,
      "loss": 3.1283,
      "step": 290
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00018825371965544246,
      "loss": 3.2871,
      "step": 300
    },
    {
      "epoch": 0.23,
      "eval_accuracy": 0.3091750192752506,
      "eval_loss": 3.0942108631134033,
      "eval_runtime": 41.1598,
      "eval_samples_per_second": 63.023,
      "eval_steps_per_second": 7.896,
      "step": 300
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00018786217697729053,
      "loss": 3.0961,
      "step": 310
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00018747063429913862,
      "loss": 3.0243,
      "step": 320
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0001870790916209867,
      "loss": 2.9358,
      "step": 330
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.0001866875489428348,
      "loss": 2.9325,
      "step": 340
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00018629600626468286,
      "loss": 2.9971,
      "step": 350
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00018590446358653093,
      "loss": 2.7911,
      "step": 360
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00018551292090837902,
      "loss": 3.0211,
      "step": 370
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00018512137823022712,
      "loss": 2.6767,
      "step": 380
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0001847298355520752,
      "loss": 3.089,
      "step": 390
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00018433829287392326,
      "loss": 2.7273,
      "step": 400
    },
    {
      "epoch": 0.31,
      "eval_accuracy": 0.330763299922899,
      "eval_loss": 2.8338043689727783,
      "eval_runtime": 40.8134,
      "eval_samples_per_second": 63.557,
      "eval_steps_per_second": 7.963,
      "step": 400
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00018394675019577135,
      "loss": 2.757,
      "step": 410
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00018355520751761942,
      "loss": 2.5827,
      "step": 420
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00018316366483946752,
      "loss": 2.7527,
      "step": 430
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0001827721221613156,
      "loss": 2.793,
      "step": 440
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00018238057948316368,
      "loss": 2.8126,
      "step": 450
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00018198903680501175,
      "loss": 2.7034,
      "step": 460
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00018159749412685985,
      "loss": 2.4759,
      "step": 470
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00018120595144870792,
      "loss": 2.5824,
      "step": 480
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00018081440877055599,
      "loss": 2.3941,
      "step": 490
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00018042286609240408,
      "loss": 2.4984,
      "step": 500
    },
    {
      "epoch": 0.39,
      "eval_accuracy": 0.43407864302235927,
      "eval_loss": 2.48604154586792,
      "eval_runtime": 40.6933,
      "eval_samples_per_second": 63.745,
      "eval_steps_per_second": 7.987,
      "step": 500
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00018003132341425218,
      "loss": 2.6472,
      "step": 510
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00017963978073610025,
      "loss": 2.4425,
      "step": 520
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00017924823805794832,
      "loss": 2.4469,
      "step": 530
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0001788566953797964,
      "loss": 2.2719,
      "step": 540
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00017846515270164448,
      "loss": 2.6521,
      "step": 550
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00017807361002349258,
      "loss": 2.7091,
      "step": 560
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00017768206734534065,
      "loss": 2.1837,
      "step": 570
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00017729052466718874,
      "loss": 2.5009,
      "step": 580
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.0001768989819890368,
      "loss": 2.1391,
      "step": 590
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00017650743931088488,
      "loss": 2.3423,
      "step": 600
    },
    {
      "epoch": 0.47,
      "eval_accuracy": 0.47956823438704704,
      "eval_loss": 2.2201321125030518,
      "eval_runtime": 40.719,
      "eval_samples_per_second": 63.705,
      "eval_steps_per_second": 7.982,
      "step": 600
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00017611589663273298,
      "loss": 2.3181,
      "step": 610
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00017572435395458107,
      "loss": 2.4361,
      "step": 620
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00017533281127642914,
      "loss": 2.1049,
      "step": 630
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0001749412685982772,
      "loss": 2.3202,
      "step": 640
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0001745497259201253,
      "loss": 2.2411,
      "step": 650
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0001741581832419734,
      "loss": 2.2504,
      "step": 660
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00017376664056382147,
      "loss": 2.4506,
      "step": 670
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00017337509788566954,
      "loss": 2.1783,
      "step": 680
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00017298355520751764,
      "loss": 2.1607,
      "step": 690
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0001725920125293657,
      "loss": 1.8785,
      "step": 700
    },
    {
      "epoch": 0.55,
      "eval_accuracy": 0.4653045489591365,
      "eval_loss": 2.188953161239624,
      "eval_runtime": 40.7627,
      "eval_samples_per_second": 63.637,
      "eval_steps_per_second": 7.973,
      "step": 700
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00017220046985121377,
      "loss": 1.9736,
      "step": 710
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00017180892717306187,
      "loss": 1.8289,
      "step": 720
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.00017141738449490997,
      "loss": 2.0143,
      "step": 730
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00017102584181675804,
      "loss": 1.9026,
      "step": 740
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0001706342991386061,
      "loss": 1.954,
      "step": 750
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0001702427564604542,
      "loss": 2.1157,
      "step": 760
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0001698512137823023,
      "loss": 1.9459,
      "step": 770
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00016945967110415037,
      "loss": 2.0861,
      "step": 780
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00016906812842599843,
      "loss": 1.9725,
      "step": 790
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00016867658574784653,
      "loss": 1.8012,
      "step": 800
    },
    {
      "epoch": 0.63,
      "eval_accuracy": 0.4865073245952197,
      "eval_loss": 1.9900634288787842,
      "eval_runtime": 40.6805,
      "eval_samples_per_second": 63.765,
      "eval_steps_per_second": 7.989,
      "step": 800
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00016828504306969463,
      "loss": 1.8629,
      "step": 810
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00016789350039154267,
      "loss": 1.9081,
      "step": 820
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00016750195771339076,
      "loss": 1.9606,
      "step": 830
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.00016711041503523886,
      "loss": 1.9441,
      "step": 840
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00016671887235708693,
      "loss": 1.8987,
      "step": 850
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.000166327329678935,
      "loss": 1.8268,
      "step": 860
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0001659357870007831,
      "loss": 1.7508,
      "step": 870
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.0001655442443226312,
      "loss": 1.7049,
      "step": 880
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00016515270164447926,
      "loss": 1.632,
      "step": 890
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00016476115896632733,
      "loss": 1.7236,
      "step": 900
    },
    {
      "epoch": 0.7,
      "eval_accuracy": 0.5736314572089437,
      "eval_loss": 1.6821279525756836,
      "eval_runtime": 40.8079,
      "eval_samples_per_second": 63.566,
      "eval_steps_per_second": 7.964,
      "step": 900
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00016436961628817542,
      "loss": 1.7729,
      "step": 910
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00016397807361002352,
      "loss": 1.8805,
      "step": 920
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.00016358653093187156,
      "loss": 1.4968,
      "step": 930
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00016319498825371966,
      "loss": 1.9607,
      "step": 940
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00016280344557556775,
      "loss": 1.5626,
      "step": 950
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.00016241190289741582,
      "loss": 1.7008,
      "step": 960
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0001620203602192639,
      "loss": 1.7529,
      "step": 970
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.000161628817541112,
      "loss": 1.59,
      "step": 980
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.00016123727486296008,
      "loss": 1.4606,
      "step": 990
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.00016084573218480815,
      "loss": 1.4949,
      "step": 1000
    },
    {
      "epoch": 0.78,
      "eval_accuracy": 0.6083269082498073,
      "eval_loss": 1.5422368049621582,
      "eval_runtime": 41.3155,
      "eval_samples_per_second": 62.785,
      "eval_steps_per_second": 7.866,
      "step": 1000
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.00016045418950665622,
      "loss": 1.3951,
      "step": 1010
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.00016006264682850432,
      "loss": 1.3543,
      "step": 1020
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.00015967110415035241,
      "loss": 1.3775,
      "step": 1030
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.00015927956147220046,
      "loss": 1.6018,
      "step": 1040
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00015888801879404855,
      "loss": 1.6941,
      "step": 1050
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.00015849647611589665,
      "loss": 1.5474,
      "step": 1060
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00015810493343774472,
      "loss": 1.2869,
      "step": 1070
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0001577133907595928,
      "loss": 1.4885,
      "step": 1080
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.00015732184808144088,
      "loss": 1.4994,
      "step": 1090
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.00015693030540328898,
      "loss": 1.5573,
      "step": 1100
    },
    {
      "epoch": 0.86,
      "eval_accuracy": 0.6110254433307633,
      "eval_loss": 1.5435744524002075,
      "eval_runtime": 40.569,
      "eval_samples_per_second": 63.94,
      "eval_steps_per_second": 8.011,
      "step": 1100
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.00015653876272513705,
      "loss": 1.5131,
      "step": 1110
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.00015614722004698512,
      "loss": 1.262,
      "step": 1120
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.0001557556773688332,
      "loss": 1.2866,
      "step": 1130
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0001553641346906813,
      "loss": 1.3828,
      "step": 1140
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00015497259201252938,
      "loss": 1.3434,
      "step": 1150
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.00015458104933437745,
      "loss": 1.324,
      "step": 1160
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.00015418950665622554,
      "loss": 1.2011,
      "step": 1170
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0001537979639780736,
      "loss": 1.315,
      "step": 1180
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0001534064212999217,
      "loss": 1.1749,
      "step": 1190
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.00015301487862176978,
      "loss": 1.3241,
      "step": 1200
    },
    {
      "epoch": 0.94,
      "eval_accuracy": 0.6206630686198921,
      "eval_loss": 1.4076749086380005,
      "eval_runtime": 40.5566,
      "eval_samples_per_second": 63.96,
      "eval_steps_per_second": 8.013,
      "step": 1200
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.00015262333594361787,
      "loss": 1.393,
      "step": 1210
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.00015223179326546594,
      "loss": 1.3495,
      "step": 1220
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.000151840250587314,
      "loss": 1.2607,
      "step": 1230
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0001514487079091621,
      "loss": 1.3498,
      "step": 1240
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.00015105716523101018,
      "loss": 1.4737,
      "step": 1250
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00015066562255285827,
      "loss": 1.2173,
      "step": 1260
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00015027407987470634,
      "loss": 1.1518,
      "step": 1270
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.00014988253719655444,
      "loss": 1.049,
      "step": 1280
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0001494909945184025,
      "loss": 1.1152,
      "step": 1290
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0001490994518402506,
      "loss": 1.0773,
      "step": 1300
    },
    {
      "epoch": 1.02,
      "eval_accuracy": 0.6915959907478797,
      "eval_loss": 1.1416738033294678,
      "eval_runtime": 41.0868,
      "eval_samples_per_second": 63.135,
      "eval_steps_per_second": 7.91,
      "step": 1300
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.00014870790916209867,
      "loss": 0.927,
      "step": 1310
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.00014831636648394677,
      "loss": 0.9421,
      "step": 1320
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.00014792482380579484,
      "loss": 0.9346,
      "step": 1330
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.00014753328112764293,
      "loss": 0.9591,
      "step": 1340
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.000147141738449491,
      "loss": 0.7073,
      "step": 1350
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.00014675019577133907,
      "loss": 0.7688,
      "step": 1360
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00014635865309318717,
      "loss": 1.0054,
      "step": 1370
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.00014596711041503524,
      "loss": 0.9498,
      "step": 1380
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.00014557556773688333,
      "loss": 0.8585,
      "step": 1390
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0001451840250587314,
      "loss": 0.7935,
      "step": 1400
    },
    {
      "epoch": 1.1,
      "eval_accuracy": 0.6931380107941403,
      "eval_loss": 1.119438648223877,
      "eval_runtime": 40.9452,
      "eval_samples_per_second": 63.353,
      "eval_steps_per_second": 7.937,
      "step": 1400
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0001447924823805795,
      "loss": 0.9609,
      "step": 1410
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.00014440093970242757,
      "loss": 0.7703,
      "step": 1420
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.00014400939702427566,
      "loss": 0.7874,
      "step": 1430
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.00014361785434612373,
      "loss": 1.1383,
      "step": 1440
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.00014322631166797183,
      "loss": 0.8565,
      "step": 1450
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0001428347689898199,
      "loss": 0.7929,
      "step": 1460
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00014244322631166796,
      "loss": 0.8146,
      "step": 1470
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00014205168363351606,
      "loss": 0.9915,
      "step": 1480
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.00014166014095536416,
      "loss": 0.8703,
      "step": 1490
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.00014126859827721223,
      "loss": 0.7677,
      "step": 1500
    },
    {
      "epoch": 1.17,
      "eval_accuracy": 0.7166538164996145,
      "eval_loss": 1.0726888179779053,
      "eval_runtime": 38.2695,
      "eval_samples_per_second": 67.782,
      "eval_steps_per_second": 8.492,
      "step": 1500
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0001408770555990603,
      "loss": 0.6706,
      "step": 1510
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0001404855129209084,
      "loss": 0.9404,
      "step": 1520
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.0001400939702427565,
      "loss": 0.6985,
      "step": 1530
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.00013970242756460456,
      "loss": 1.0434,
      "step": 1540
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.00013931088488645262,
      "loss": 0.981,
      "step": 1550
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.00013891934220830072,
      "loss": 0.6079,
      "step": 1560
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0001385277995301488,
      "loss": 0.6289,
      "step": 1570
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.00013813625685199686,
      "loss": 0.9693,
      "step": 1580
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00013774471417384496,
      "loss": 0.6964,
      "step": 1590
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00013735317149569305,
      "loss": 0.9468,
      "step": 1600
    },
    {
      "epoch": 1.25,
      "eval_accuracy": 0.7135697764070933,
      "eval_loss": 1.0706665515899658,
      "eval_runtime": 40.6503,
      "eval_samples_per_second": 63.813,
      "eval_steps_per_second": 7.995,
      "step": 1600
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00013696162881754112,
      "loss": 0.7422,
      "step": 1610
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0001365700861393892,
      "loss": 0.8951,
      "step": 1620
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00013617854346123729,
      "loss": 0.8085,
      "step": 1630
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00013578700078308538,
      "loss": 1.1696,
      "step": 1640
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00013539545810493345,
      "loss": 0.7189,
      "step": 1650
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00013500391542678152,
      "loss": 0.7836,
      "step": 1660
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.00013461237274862962,
      "loss": 0.9709,
      "step": 1670
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0001342208300704777,
      "loss": 0.7392,
      "step": 1680
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00013382928739232575,
      "loss": 0.8334,
      "step": 1690
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00013343774471417385,
      "loss": 0.7563,
      "step": 1700
    },
    {
      "epoch": 1.33,
      "eval_accuracy": 0.7390131071703933,
      "eval_loss": 0.9426913261413574,
      "eval_runtime": 41.0106,
      "eval_samples_per_second": 63.252,
      "eval_steps_per_second": 7.925,
      "step": 1700
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00013304620203602195,
      "loss": 0.8852,
      "step": 1710
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.00013265465935787001,
      "loss": 0.679,
      "step": 1720
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.00013226311667971808,
      "loss": 0.711,
      "step": 1730
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00013187157400156618,
      "loss": 0.6945,
      "step": 1740
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.00013148003132341428,
      "loss": 0.6833,
      "step": 1750
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.00013108848864526232,
      "loss": 0.8873,
      "step": 1760
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0001306969459671104,
      "loss": 0.6331,
      "step": 1770
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0001303054032889585,
      "loss": 0.8011,
      "step": 1780
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0001299138606108066,
      "loss": 0.7751,
      "step": 1790
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00012952231793265465,
      "loss": 0.8471,
      "step": 1800
    },
    {
      "epoch": 1.41,
      "eval_accuracy": 0.7571318427139553,
      "eval_loss": 0.8905778527259827,
      "eval_runtime": 40.6713,
      "eval_samples_per_second": 63.78,
      "eval_steps_per_second": 7.991,
      "step": 1800
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00012913077525450274,
      "loss": 0.7046,
      "step": 1810
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.00012873923257635084,
      "loss": 0.532,
      "step": 1820
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0001283476898981989,
      "loss": 0.7583,
      "step": 1830
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.00012795614722004698,
      "loss": 0.6438,
      "step": 1840
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.00012756460454189507,
      "loss": 0.6597,
      "step": 1850
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.00012717306186374317,
      "loss": 1.0176,
      "step": 1860
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.00012678151918559124,
      "loss": 0.5534,
      "step": 1870
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0001263899765074393,
      "loss": 0.7411,
      "step": 1880
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0001259984338292874,
      "loss": 0.5911,
      "step": 1890
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0001256068911511355,
      "loss": 0.9998,
      "step": 1900
    },
    {
      "epoch": 1.49,
      "eval_accuracy": 0.7845026985350809,
      "eval_loss": 0.8098413944244385,
      "eval_runtime": 40.972,
      "eval_samples_per_second": 63.312,
      "eval_steps_per_second": 7.932,
      "step": 1900
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.00012521534847298354,
      "loss": 0.5899,
      "step": 1910
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.00012482380579483164,
      "loss": 0.4877,
      "step": 1920
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.00012443226311667973,
      "loss": 0.7054,
      "step": 1930
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0001240407204385278,
      "loss": 0.4887,
      "step": 1940
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.00012364917776037587,
      "loss": 0.4256,
      "step": 1950
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.00012325763508222397,
      "loss": 0.4679,
      "step": 1960
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.00012286609240407206,
      "loss": 0.4745,
      "step": 1970
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.00012247454972592013,
      "loss": 0.583,
      "step": 1980
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0001220830070477682,
      "loss": 0.6025,
      "step": 1990
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.0001216914643696163,
      "loss": 0.6039,
      "step": 2000
    },
    {
      "epoch": 1.57,
      "eval_accuracy": 0.8033924441017734,
      "eval_loss": 0.724382221698761,
      "eval_runtime": 41.1355,
      "eval_samples_per_second": 63.06,
      "eval_steps_per_second": 7.901,
      "step": 2000
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.00012129992169146438,
      "loss": 0.5208,
      "step": 2010
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.00012090837901331246,
      "loss": 0.6528,
      "step": 2020
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.00012051683633516053,
      "loss": 0.4718,
      "step": 2030
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.00012012529365700861,
      "loss": 0.5781,
      "step": 2040
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.00011973375097885671,
      "loss": 0.535,
      "step": 2050
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.00011934220830070477,
      "loss": 0.6573,
      "step": 2060
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.00011895066562255286,
      "loss": 0.5547,
      "step": 2070
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.00011855912294440094,
      "loss": 0.4013,
      "step": 2080
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.00011816758026624904,
      "loss": 0.4209,
      "step": 2090
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.0001177760375880971,
      "loss": 0.7052,
      "step": 2100
    },
    {
      "epoch": 1.64,
      "eval_accuracy": 0.7952968388589051,
      "eval_loss": 0.7880675792694092,
      "eval_runtime": 40.536,
      "eval_samples_per_second": 63.993,
      "eval_steps_per_second": 8.018,
      "step": 2100
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.00011738449490994519,
      "loss": 0.7183,
      "step": 2110
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.00011699295223179327,
      "loss": 0.499,
      "step": 2120
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.00011660140955364136,
      "loss": 0.8091,
      "step": 2130
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.00011620986687548943,
      "loss": 0.5392,
      "step": 2140
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.00011581832419733751,
      "loss": 0.6083,
      "step": 2150
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.0001154267815191856,
      "loss": 0.6992,
      "step": 2160
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.00011503523884103369,
      "loss": 0.3624,
      "step": 2170
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.00011464369616288176,
      "loss": 0.5923,
      "step": 2180
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.00011425215348472984,
      "loss": 0.5746,
      "step": 2190
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.00011386061080657792,
      "loss": 0.6753,
      "step": 2200
    },
    {
      "epoch": 1.72,
      "eval_accuracy": 0.7925983037779492,
      "eval_loss": 0.745768129825592,
      "eval_runtime": 41.0623,
      "eval_samples_per_second": 63.172,
      "eval_steps_per_second": 7.915,
      "step": 2200
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.00011346906812842602,
      "loss": 0.4564,
      "step": 2210
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.00011307752545027409,
      "loss": 0.7629,
      "step": 2220
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.00011268598277212217,
      "loss": 0.6863,
      "step": 2230
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.00011229444009397025,
      "loss": 0.4225,
      "step": 2240
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.00011190289741581832,
      "loss": 0.5484,
      "step": 2250
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.0001115113547376664,
      "loss": 0.3271,
      "step": 2260
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.0001111198120595145,
      "loss": 0.6182,
      "step": 2270
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.00011072826938136258,
      "loss": 0.6526,
      "step": 2280
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.00011033672670321065,
      "loss": 0.6346,
      "step": 2290
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.00010994518402505873,
      "loss": 0.3758,
      "step": 2300
    },
    {
      "epoch": 1.8,
      "eval_accuracy": 0.8022359290670779,
      "eval_loss": 0.6987051963806152,
      "eval_runtime": 41.0604,
      "eval_samples_per_second": 63.175,
      "eval_steps_per_second": 7.915,
      "step": 2300
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.00010955364134690682,
      "loss": 0.6213,
      "step": 2310
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.00010916209866875491,
      "loss": 0.6481,
      "step": 2320
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.00010877055599060297,
      "loss": 0.4049,
      "step": 2330
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.00010837901331245106,
      "loss": 0.5083,
      "step": 2340
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.00010798747063429915,
      "loss": 0.4128,
      "step": 2350
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.00010759592795614724,
      "loss": 0.3613,
      "step": 2360
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.0001072043852779953,
      "loss": 0.5291,
      "step": 2370
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.0001068128425998434,
      "loss": 0.3833,
      "step": 2380
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.00010642129992169148,
      "loss": 0.4959,
      "step": 2390
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.00010602975724353954,
      "loss": 0.4985,
      "step": 2400
    },
    {
      "epoch": 1.88,
      "eval_accuracy": 0.8265227447956823,
      "eval_loss": 0.6285678744316101,
      "eval_runtime": 40.7249,
      "eval_samples_per_second": 63.696,
      "eval_steps_per_second": 7.98,
      "step": 2400
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.00010563821456538763,
      "loss": 0.5699,
      "step": 2410
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.00010524667188723571,
      "loss": 0.6911,
      "step": 2420
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.0001048551292090838,
      "loss": 0.3526,
      "step": 2430
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.00010446358653093186,
      "loss": 0.4432,
      "step": 2440
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.00010407204385277996,
      "loss": 0.3668,
      "step": 2450
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.00010368050117462804,
      "loss": 0.45,
      "step": 2460
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.00010328895849647614,
      "loss": 0.4812,
      "step": 2470
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.00010289741581832419,
      "loss": 0.6194,
      "step": 2480
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.00010250587314017229,
      "loss": 0.6149,
      "step": 2490
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.00010211433046202037,
      "loss": 0.4122,
      "step": 2500
    },
    {
      "epoch": 1.96,
      "eval_accuracy": 0.8357748650732459,
      "eval_loss": 0.5948623418807983,
      "eval_runtime": 41.2758,
      "eval_samples_per_second": 62.846,
      "eval_steps_per_second": 7.874,
      "step": 2500
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.00010172278778386845,
      "loss": 0.5993,
      "step": 2510
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.00010133124510571652,
      "loss": 0.3777,
      "step": 2520
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.0001009397024275646,
      "loss": 0.3447,
      "step": 2530
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.0001005481597494127,
      "loss": 0.4143,
      "step": 2540
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.00010015661707126078,
      "loss": 0.5844,
      "step": 2550
    },
    {
      "epoch": 2.0,
      "learning_rate": 9.976507439310885e-05,
      "loss": 0.3812,
      "step": 2560
    },
    {
      "epoch": 2.01,
      "learning_rate": 9.937353171495693e-05,
      "loss": 0.2064,
      "step": 2570
    },
    {
      "epoch": 2.02,
      "learning_rate": 9.898198903680502e-05,
      "loss": 0.239,
      "step": 2580
    },
    {
      "epoch": 2.03,
      "learning_rate": 9.85904463586531e-05,
      "loss": 0.2053,
      "step": 2590
    },
    {
      "epoch": 2.04,
      "learning_rate": 9.819890368050118e-05,
      "loss": 0.1286,
      "step": 2600
    },
    {
      "epoch": 2.04,
      "eval_accuracy": 0.838473400154202,
      "eval_loss": 0.5690719485282898,
      "eval_runtime": 41.0005,
      "eval_samples_per_second": 63.268,
      "eval_steps_per_second": 7.927,
      "step": 2600
    },
    {
      "epoch": 2.04,
      "learning_rate": 9.780736100234926e-05,
      "loss": 0.1961,
      "step": 2610
    },
    {
      "epoch": 2.05,
      "learning_rate": 9.741581832419733e-05,
      "loss": 0.1303,
      "step": 2620
    },
    {
      "epoch": 2.06,
      "learning_rate": 9.702427564604543e-05,
      "loss": 0.1351,
      "step": 2630
    },
    {
      "epoch": 2.07,
      "learning_rate": 9.66327329678935e-05,
      "loss": 0.3014,
      "step": 2640
    },
    {
      "epoch": 2.08,
      "learning_rate": 9.62411902897416e-05,
      "loss": 0.2057,
      "step": 2650
    },
    {
      "epoch": 2.08,
      "learning_rate": 9.584964761158966e-05,
      "loss": 0.2423,
      "step": 2660
    },
    {
      "epoch": 2.09,
      "learning_rate": 9.545810493343776e-05,
      "loss": 0.2505,
      "step": 2670
    },
    {
      "epoch": 2.1,
      "learning_rate": 9.506656225528583e-05,
      "loss": 0.1811,
      "step": 2680
    },
    {
      "epoch": 2.11,
      "learning_rate": 9.467501957713391e-05,
      "loss": 0.1692,
      "step": 2690
    },
    {
      "epoch": 2.11,
      "learning_rate": 9.4283476898982e-05,
      "loss": 0.1989,
      "step": 2700
    },
    {
      "epoch": 2.11,
      "eval_accuracy": 0.8388589051657671,
      "eval_loss": 0.5534732937812805,
      "eval_runtime": 40.7155,
      "eval_samples_per_second": 63.71,
      "eval_steps_per_second": 7.982,
      "step": 2700
    },
    {
      "epoch": 2.12,
      "learning_rate": 9.389193422083008e-05,
      "loss": 0.3107,
      "step": 2710
    },
    {
      "epoch": 2.13,
      "learning_rate": 9.350039154267816e-05,
      "loss": 0.5274,
      "step": 2720
    },
    {
      "epoch": 2.14,
      "learning_rate": 9.310884886452623e-05,
      "loss": 0.2806,
      "step": 2730
    },
    {
      "epoch": 2.15,
      "learning_rate": 9.271730618637432e-05,
      "loss": 0.261,
      "step": 2740
    },
    {
      "epoch": 2.15,
      "learning_rate": 9.232576350822239e-05,
      "loss": 0.3325,
      "step": 2750
    },
    {
      "epoch": 2.16,
      "learning_rate": 9.193422083007049e-05,
      "loss": 0.2559,
      "step": 2760
    },
    {
      "epoch": 2.17,
      "learning_rate": 9.154267815191856e-05,
      "loss": 0.2627,
      "step": 2770
    },
    {
      "epoch": 2.18,
      "learning_rate": 9.115113547376664e-05,
      "loss": 0.3315,
      "step": 2780
    },
    {
      "epoch": 2.18,
      "learning_rate": 9.075959279561472e-05,
      "loss": 0.1876,
      "step": 2790
    },
    {
      "epoch": 2.19,
      "learning_rate": 9.03680501174628e-05,
      "loss": 0.3304,
      "step": 2800
    },
    {
      "epoch": 2.19,
      "eval_accuracy": 0.8519660755589823,
      "eval_loss": 0.5260942578315735,
      "eval_runtime": 41.0357,
      "eval_samples_per_second": 63.213,
      "eval_steps_per_second": 7.92,
      "step": 2800
    },
    {
      "epoch": 2.2,
      "learning_rate": 8.997650743931089e-05,
      "loss": 0.2837,
      "step": 2810
    },
    {
      "epoch": 2.21,
      "learning_rate": 8.958496476115897e-05,
      "loss": 0.2487,
      "step": 2820
    },
    {
      "epoch": 2.22,
      "learning_rate": 8.919342208300705e-05,
      "loss": 0.175,
      "step": 2830
    },
    {
      "epoch": 2.22,
      "learning_rate": 8.880187940485514e-05,
      "loss": 0.2872,
      "step": 2840
    },
    {
      "epoch": 2.23,
      "learning_rate": 8.841033672670322e-05,
      "loss": 0.1787,
      "step": 2850
    },
    {
      "epoch": 2.24,
      "learning_rate": 8.80187940485513e-05,
      "loss": 0.2126,
      "step": 2860
    },
    {
      "epoch": 2.25,
      "learning_rate": 8.762725137039938e-05,
      "loss": 0.1382,
      "step": 2870
    },
    {
      "epoch": 2.26,
      "learning_rate": 8.723570869224747e-05,
      "loss": 0.273,
      "step": 2880
    },
    {
      "epoch": 2.26,
      "learning_rate": 8.684416601409553e-05,
      "loss": 0.3135,
      "step": 2890
    },
    {
      "epoch": 2.27,
      "learning_rate": 8.645262333594362e-05,
      "loss": 0.3415,
      "step": 2900
    },
    {
      "epoch": 2.27,
      "eval_accuracy": 0.8477255204317656,
      "eval_loss": 0.5503605008125305,
      "eval_runtime": 40.5607,
      "eval_samples_per_second": 63.954,
      "eval_steps_per_second": 8.013,
      "step": 2900
    },
    {
      "epoch": 2.28,
      "learning_rate": 8.60610806577917e-05,
      "loss": 0.2213,
      "step": 2910
    },
    {
      "epoch": 2.29,
      "learning_rate": 8.566953797963978e-05,
      "loss": 0.1369,
      "step": 2920
    },
    {
      "epoch": 2.29,
      "learning_rate": 8.527799530148786e-05,
      "loss": 0.1375,
      "step": 2930
    },
    {
      "epoch": 2.3,
      "learning_rate": 8.488645262333595e-05,
      "loss": 0.1758,
      "step": 2940
    },
    {
      "epoch": 2.31,
      "learning_rate": 8.449490994518403e-05,
      "loss": 0.2331,
      "step": 2950
    },
    {
      "epoch": 2.32,
      "learning_rate": 8.410336726703211e-05,
      "loss": 0.1038,
      "step": 2960
    },
    {
      "epoch": 2.33,
      "learning_rate": 8.37118245888802e-05,
      "loss": 0.154,
      "step": 2970
    },
    {
      "epoch": 2.33,
      "learning_rate": 8.332028191072828e-05,
      "loss": 0.1511,
      "step": 2980
    },
    {
      "epoch": 2.34,
      "learning_rate": 8.292873923257636e-05,
      "loss": 0.1806,
      "step": 2990
    },
    {
      "epoch": 2.35,
      "learning_rate": 8.253719655442443e-05,
      "loss": 0.4066,
      "step": 3000
    },
    {
      "epoch": 2.35,
      "eval_accuracy": 0.8496530454895914,
      "eval_loss": 0.5417803525924683,
      "eval_runtime": 40.572,
      "eval_samples_per_second": 63.936,
      "eval_steps_per_second": 8.01,
      "step": 3000
    },
    {
      "epoch": 2.36,
      "learning_rate": 8.214565387627252e-05,
      "loss": 0.1892,
      "step": 3010
    },
    {
      "epoch": 2.36,
      "learning_rate": 8.17541111981206e-05,
      "loss": 0.0577,
      "step": 3020
    },
    {
      "epoch": 2.37,
      "learning_rate": 8.136256851996869e-05,
      "loss": 0.0838,
      "step": 3030
    },
    {
      "epoch": 2.38,
      "learning_rate": 8.097102584181676e-05,
      "loss": 0.2254,
      "step": 3040
    },
    {
      "epoch": 2.39,
      "learning_rate": 8.057948316366485e-05,
      "loss": 0.1103,
      "step": 3050
    },
    {
      "epoch": 2.4,
      "learning_rate": 8.018794048551292e-05,
      "loss": 0.3653,
      "step": 3060
    },
    {
      "epoch": 2.4,
      "learning_rate": 7.9796397807361e-05,
      "loss": 0.3203,
      "step": 3070
    },
    {
      "epoch": 2.41,
      "learning_rate": 7.940485512920909e-05,
      "loss": 0.1933,
      "step": 3080
    },
    {
      "epoch": 2.42,
      "learning_rate": 7.901331245105716e-05,
      "loss": 0.2115,
      "step": 3090
    },
    {
      "epoch": 2.43,
      "learning_rate": 7.862176977290525e-05,
      "loss": 0.1208,
      "step": 3100
    },
    {
      "epoch": 2.43,
      "eval_accuracy": 0.8612181958365459,
      "eval_loss": 0.515609860420227,
      "eval_runtime": 40.5458,
      "eval_samples_per_second": 63.977,
      "eval_steps_per_second": 8.016,
      "step": 3100
    },
    {
      "epoch": 2.44,
      "learning_rate": 7.823022709475332e-05,
      "loss": 0.1831,
      "step": 3110
    },
    {
      "epoch": 2.44,
      "learning_rate": 7.783868441660142e-05,
      "loss": 0.1672,
      "step": 3120
    },
    {
      "epoch": 2.45,
      "learning_rate": 7.744714173844949e-05,
      "loss": 0.1637,
      "step": 3130
    },
    {
      "epoch": 2.46,
      "learning_rate": 7.705559906029758e-05,
      "loss": 0.108,
      "step": 3140
    },
    {
      "epoch": 2.47,
      "learning_rate": 7.666405638214565e-05,
      "loss": 0.1537,
      "step": 3150
    },
    {
      "epoch": 2.47,
      "learning_rate": 7.627251370399374e-05,
      "loss": 0.1982,
      "step": 3160
    },
    {
      "epoch": 2.48,
      "learning_rate": 7.588097102584182e-05,
      "loss": 0.1982,
      "step": 3170
    },
    {
      "epoch": 2.49,
      "learning_rate": 7.54894283476899e-05,
      "loss": 0.1579,
      "step": 3180
    },
    {
      "epoch": 2.5,
      "learning_rate": 7.509788566953798e-05,
      "loss": 0.2772,
      "step": 3190
    },
    {
      "epoch": 2.51,
      "learning_rate": 7.470634299138607e-05,
      "loss": 0.1668,
      "step": 3200
    },
    {
      "epoch": 2.51,
      "eval_accuracy": 0.853893600616808,
      "eval_loss": 0.565523624420166,
      "eval_runtime": 40.5285,
      "eval_samples_per_second": 64.004,
      "eval_steps_per_second": 8.019,
      "step": 3200
    },
    {
      "epoch": 2.51,
      "learning_rate": 7.431480031323415e-05,
      "loss": 0.1441,
      "step": 3210
    },
    {
      "epoch": 2.52,
      "learning_rate": 7.392325763508223e-05,
      "loss": 0.1452,
      "step": 3220
    },
    {
      "epoch": 2.53,
      "learning_rate": 7.353171495693031e-05,
      "loss": 0.1785,
      "step": 3230
    },
    {
      "epoch": 2.54,
      "learning_rate": 7.314017227877838e-05,
      "loss": 0.3255,
      "step": 3240
    },
    {
      "epoch": 2.55,
      "learning_rate": 7.274862960062648e-05,
      "loss": 0.1776,
      "step": 3250
    },
    {
      "epoch": 2.55,
      "learning_rate": 7.235708692247455e-05,
      "loss": 0.132,
      "step": 3260
    },
    {
      "epoch": 2.56,
      "learning_rate": 7.196554424432263e-05,
      "loss": 0.0892,
      "step": 3270
    },
    {
      "epoch": 2.57,
      "learning_rate": 7.157400156617071e-05,
      "loss": 0.1919,
      "step": 3280
    },
    {
      "epoch": 2.58,
      "learning_rate": 7.11824588880188e-05,
      "loss": 0.1852,
      "step": 3290
    },
    {
      "epoch": 2.58,
      "learning_rate": 7.079091620986688e-05,
      "loss": 0.0727,
      "step": 3300
    },
    {
      "epoch": 2.58,
      "eval_accuracy": 0.8658442559753277,
      "eval_loss": 0.4970535933971405,
      "eval_runtime": 40.5488,
      "eval_samples_per_second": 63.972,
      "eval_steps_per_second": 8.015,
      "step": 3300
    },
    {
      "epoch": 2.59,
      "learning_rate": 7.039937353171496e-05,
      "loss": 0.1923,
      "step": 3310
    },
    {
      "epoch": 2.6,
      "learning_rate": 7.000783085356304e-05,
      "loss": 0.2059,
      "step": 3320
    },
    {
      "epoch": 2.61,
      "learning_rate": 6.961628817541112e-05,
      "loss": 0.3617,
      "step": 3330
    },
    {
      "epoch": 2.62,
      "learning_rate": 6.922474549725921e-05,
      "loss": 0.2229,
      "step": 3340
    },
    {
      "epoch": 2.62,
      "learning_rate": 6.883320281910729e-05,
      "loss": 0.3029,
      "step": 3350
    },
    {
      "epoch": 2.63,
      "learning_rate": 6.844166014095537e-05,
      "loss": 0.117,
      "step": 3360
    },
    {
      "epoch": 2.64,
      "learning_rate": 6.805011746280345e-05,
      "loss": 0.3387,
      "step": 3370
    },
    {
      "epoch": 2.65,
      "learning_rate": 6.765857478465152e-05,
      "loss": 0.116,
      "step": 3380
    },
    {
      "epoch": 2.65,
      "learning_rate": 6.726703210649962e-05,
      "loss": 0.2186,
      "step": 3390
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.687548942834769e-05,
      "loss": 0.0929,
      "step": 3400
    },
    {
      "epoch": 2.66,
      "eval_accuracy": 0.8635312259059368,
      "eval_loss": 0.49616244435310364,
      "eval_runtime": 40.5352,
      "eval_samples_per_second": 63.994,
      "eval_steps_per_second": 8.018,
      "step": 3400
    },
    {
      "epoch": 2.67,
      "learning_rate": 6.648394675019577e-05,
      "loss": 0.1702,
      "step": 3410
    },
    {
      "epoch": 2.68,
      "learning_rate": 6.609240407204385e-05,
      "loss": 0.2464,
      "step": 3420
    },
    {
      "epoch": 2.69,
      "learning_rate": 6.570086139389194e-05,
      "loss": 0.1713,
      "step": 3430
    },
    {
      "epoch": 2.69,
      "learning_rate": 6.530931871574002e-05,
      "loss": 0.0984,
      "step": 3440
    },
    {
      "epoch": 2.7,
      "learning_rate": 6.49177760375881e-05,
      "loss": 0.1126,
      "step": 3450
    },
    {
      "epoch": 2.71,
      "learning_rate": 6.452623335943618e-05,
      "loss": 0.2286,
      "step": 3460
    },
    {
      "epoch": 2.72,
      "learning_rate": 6.413469068128425e-05,
      "loss": 0.1258,
      "step": 3470
    },
    {
      "epoch": 2.73,
      "learning_rate": 6.374314800313235e-05,
      "loss": 0.0973,
      "step": 3480
    },
    {
      "epoch": 2.73,
      "learning_rate": 6.335160532498042e-05,
      "loss": 0.1675,
      "step": 3490
    },
    {
      "epoch": 2.74,
      "learning_rate": 6.296006264682851e-05,
      "loss": 0.0678,
      "step": 3500
    },
    {
      "epoch": 2.74,
      "eval_accuracy": 0.8670007710100232,
      "eval_loss": 0.49030396342277527,
      "eval_runtime": 40.9495,
      "eval_samples_per_second": 63.346,
      "eval_steps_per_second": 7.937,
      "step": 3500
    },
    {
      "epoch": 2.75,
      "learning_rate": 6.256851996867658e-05,
      "loss": 0.093,
      "step": 3510
    },
    {
      "epoch": 2.76,
      "learning_rate": 6.217697729052468e-05,
      "loss": 0.1365,
      "step": 3520
    },
    {
      "epoch": 2.76,
      "learning_rate": 6.178543461237275e-05,
      "loss": 0.0985,
      "step": 3530
    },
    {
      "epoch": 2.77,
      "learning_rate": 6.139389193422083e-05,
      "loss": 0.1239,
      "step": 3540
    },
    {
      "epoch": 2.78,
      "learning_rate": 6.100234925606891e-05,
      "loss": 0.0351,
      "step": 3550
    },
    {
      "epoch": 2.79,
      "learning_rate": 6.0610806577917e-05,
      "loss": 0.1373,
      "step": 3560
    },
    {
      "epoch": 2.8,
      "learning_rate": 6.021926389976508e-05,
      "loss": 0.1985,
      "step": 3570
    },
    {
      "epoch": 2.8,
      "learning_rate": 5.9827721221613154e-05,
      "loss": 0.1776,
      "step": 3580
    },
    {
      "epoch": 2.81,
      "learning_rate": 5.943617854346124e-05,
      "loss": 0.3687,
      "step": 3590
    },
    {
      "epoch": 2.82,
      "learning_rate": 5.904463586530932e-05,
      "loss": 0.1212,
      "step": 3600
    },
    {
      "epoch": 2.82,
      "eval_accuracy": 0.8866615265998458,
      "eval_loss": 0.4357341229915619,
      "eval_runtime": 40.7756,
      "eval_samples_per_second": 63.616,
      "eval_steps_per_second": 7.97,
      "step": 3600
    },
    {
      "epoch": 2.83,
      "learning_rate": 5.86530931871574e-05,
      "loss": 0.1552,
      "step": 3610
    },
    {
      "epoch": 2.83,
      "learning_rate": 5.826155050900548e-05,
      "loss": 0.1493,
      "step": 3620
    },
    {
      "epoch": 2.84,
      "learning_rate": 5.787000783085357e-05,
      "loss": 0.2062,
      "step": 3630
    },
    {
      "epoch": 2.85,
      "learning_rate": 5.747846515270164e-05,
      "loss": 0.144,
      "step": 3640
    },
    {
      "epoch": 2.86,
      "learning_rate": 5.708692247454973e-05,
      "loss": 0.1866,
      "step": 3650
    },
    {
      "epoch": 2.87,
      "learning_rate": 5.669537979639781e-05,
      "loss": 0.0389,
      "step": 3660
    },
    {
      "epoch": 2.87,
      "learning_rate": 5.63038371182459e-05,
      "loss": 0.2919,
      "step": 3670
    },
    {
      "epoch": 2.88,
      "learning_rate": 5.591229444009397e-05,
      "loss": 0.1491,
      "step": 3680
    },
    {
      "epoch": 2.89,
      "learning_rate": 5.5520751761942055e-05,
      "loss": 0.3148,
      "step": 3690
    },
    {
      "epoch": 2.9,
      "learning_rate": 5.512920908379014e-05,
      "loss": 0.1579,
      "step": 3700
    },
    {
      "epoch": 2.9,
      "eval_accuracy": 0.8739398612181959,
      "eval_loss": 0.4641892611980438,
      "eval_runtime": 41.3852,
      "eval_samples_per_second": 62.679,
      "eval_steps_per_second": 7.853,
      "step": 3700
    },
    {
      "epoch": 2.91,
      "learning_rate": 5.473766640563822e-05,
      "loss": 0.0813,
      "step": 3710
    },
    {
      "epoch": 2.91,
      "learning_rate": 5.4346123727486296e-05,
      "loss": 0.1219,
      "step": 3720
    },
    {
      "epoch": 2.92,
      "learning_rate": 5.3954581049334385e-05,
      "loss": 0.2151,
      "step": 3730
    },
    {
      "epoch": 2.93,
      "learning_rate": 5.356303837118246e-05,
      "loss": 0.0864,
      "step": 3740
    },
    {
      "epoch": 2.94,
      "learning_rate": 5.317149569303054e-05,
      "loss": 0.2892,
      "step": 3750
    },
    {
      "epoch": 2.94,
      "learning_rate": 5.2779953014878626e-05,
      "loss": 0.1045,
      "step": 3760
    },
    {
      "epoch": 2.95,
      "learning_rate": 5.23884103367267e-05,
      "loss": 0.1124,
      "step": 3770
    },
    {
      "epoch": 2.96,
      "learning_rate": 5.199686765857479e-05,
      "loss": 0.1724,
      "step": 3780
    },
    {
      "epoch": 2.97,
      "learning_rate": 5.160532498042287e-05,
      "loss": 0.0993,
      "step": 3790
    },
    {
      "epoch": 2.98,
      "learning_rate": 5.121378230227095e-05,
      "loss": 0.2625,
      "step": 3800
    },
    {
      "epoch": 2.98,
      "eval_accuracy": 0.8951426368542791,
      "eval_loss": 0.39944082498550415,
      "eval_runtime": 40.9786,
      "eval_samples_per_second": 63.301,
      "eval_steps_per_second": 7.931,
      "step": 3800
    },
    {
      "epoch": 2.98,
      "learning_rate": 5.0822239624119025e-05,
      "loss": 0.0913,
      "step": 3810
    },
    {
      "epoch": 2.99,
      "learning_rate": 5.0430696945967114e-05,
      "loss": 0.1575,
      "step": 3820
    },
    {
      "epoch": 3.0,
      "learning_rate": 5.003915426781519e-05,
      "loss": 0.2052,
      "step": 3830
    },
    {
      "epoch": 3.01,
      "learning_rate": 4.964761158966327e-05,
      "loss": 0.0363,
      "step": 3840
    },
    {
      "epoch": 3.01,
      "learning_rate": 4.9256068911511355e-05,
      "loss": 0.0477,
      "step": 3850
    },
    {
      "epoch": 3.02,
      "learning_rate": 4.886452623335944e-05,
      "loss": 0.0262,
      "step": 3860
    },
    {
      "epoch": 3.03,
      "learning_rate": 4.847298355520752e-05,
      "loss": 0.0329,
      "step": 3870
    },
    {
      "epoch": 3.04,
      "learning_rate": 4.80814408770556e-05,
      "loss": 0.1238,
      "step": 3880
    },
    {
      "epoch": 3.05,
      "learning_rate": 4.7689898198903685e-05,
      "loss": 0.0234,
      "step": 3890
    },
    {
      "epoch": 3.05,
      "learning_rate": 4.729835552075176e-05,
      "loss": 0.024,
      "step": 3900
    },
    {
      "epoch": 3.05,
      "eval_accuracy": 0.8970701619121049,
      "eval_loss": 0.39525991678237915,
      "eval_runtime": 41.0486,
      "eval_samples_per_second": 63.193,
      "eval_steps_per_second": 7.917,
      "step": 3900
    },
    {
      "epoch": 3.06,
      "learning_rate": 4.6906812842599844e-05,
      "loss": 0.023,
      "step": 3910
    },
    {
      "epoch": 3.07,
      "learning_rate": 4.6515270164447926e-05,
      "loss": 0.0514,
      "step": 3920
    },
    {
      "epoch": 3.08,
      "learning_rate": 4.612372748629601e-05,
      "loss": 0.0687,
      "step": 3930
    },
    {
      "epoch": 3.09,
      "learning_rate": 4.573218480814409e-05,
      "loss": 0.1139,
      "step": 3940
    },
    {
      "epoch": 3.09,
      "learning_rate": 4.5340642129992174e-05,
      "loss": 0.0813,
      "step": 3950
    },
    {
      "epoch": 3.1,
      "learning_rate": 4.4949099451840256e-05,
      "loss": 0.0347,
      "step": 3960
    },
    {
      "epoch": 3.11,
      "learning_rate": 4.455755677368834e-05,
      "loss": 0.031,
      "step": 3970
    },
    {
      "epoch": 3.12,
      "learning_rate": 4.4166014095536414e-05,
      "loss": 0.0696,
      "step": 3980
    },
    {
      "epoch": 3.12,
      "learning_rate": 4.37744714173845e-05,
      "loss": 0.0243,
      "step": 3990
    },
    {
      "epoch": 3.13,
      "learning_rate": 4.338292873923258e-05,
      "loss": 0.0696,
      "step": 4000
    },
    {
      "epoch": 3.13,
      "eval_accuracy": 0.9055512721665382,
      "eval_loss": 0.38827046751976013,
      "eval_runtime": 40.6562,
      "eval_samples_per_second": 63.803,
      "eval_steps_per_second": 7.994,
      "step": 4000
    },
    {
      "epoch": 3.14,
      "learning_rate": 4.2991386061080655e-05,
      "loss": 0.0264,
      "step": 4010
    },
    {
      "epoch": 3.15,
      "learning_rate": 4.259984338292874e-05,
      "loss": 0.0602,
      "step": 4020
    },
    {
      "epoch": 3.16,
      "learning_rate": 4.220830070477682e-05,
      "loss": 0.0119,
      "step": 4030
    },
    {
      "epoch": 3.16,
      "learning_rate": 4.18167580266249e-05,
      "loss": 0.0386,
      "step": 4040
    },
    {
      "epoch": 3.17,
      "learning_rate": 4.1425215348472985e-05,
      "loss": 0.0287,
      "step": 4050
    },
    {
      "epoch": 3.18,
      "learning_rate": 4.103367267032107e-05,
      "loss": 0.0278,
      "step": 4060
    },
    {
      "epoch": 3.19,
      "learning_rate": 4.064212999216915e-05,
      "loss": 0.0169,
      "step": 4070
    },
    {
      "epoch": 3.19,
      "learning_rate": 4.025058731401723e-05,
      "loss": 0.0209,
      "step": 4080
    },
    {
      "epoch": 3.2,
      "learning_rate": 3.985904463586531e-05,
      "loss": 0.0731,
      "step": 4090
    },
    {
      "epoch": 3.21,
      "learning_rate": 3.946750195771339e-05,
      "loss": 0.0169,
      "step": 4100
    },
    {
      "epoch": 3.21,
      "eval_accuracy": 0.9086353122590594,
      "eval_loss": 0.37549787759780884,
      "eval_runtime": 41.0263,
      "eval_samples_per_second": 63.228,
      "eval_steps_per_second": 7.922,
      "step": 4100
    },
    {
      "epoch": 3.22,
      "learning_rate": 3.9075959279561474e-05,
      "loss": 0.0172,
      "step": 4110
    },
    {
      "epoch": 3.23,
      "learning_rate": 3.8684416601409556e-05,
      "loss": 0.0282,
      "step": 4120
    },
    {
      "epoch": 3.23,
      "learning_rate": 3.829287392325764e-05,
      "loss": 0.0504,
      "step": 4130
    },
    {
      "epoch": 3.24,
      "learning_rate": 3.790133124510572e-05,
      "loss": 0.0555,
      "step": 4140
    },
    {
      "epoch": 3.25,
      "learning_rate": 3.7509788566953804e-05,
      "loss": 0.0247,
      "step": 4150
    },
    {
      "epoch": 3.26,
      "learning_rate": 3.7118245888801886e-05,
      "loss": 0.0661,
      "step": 4160
    },
    {
      "epoch": 3.27,
      "learning_rate": 3.672670321064996e-05,
      "loss": 0.0955,
      "step": 4170
    },
    {
      "epoch": 3.27,
      "learning_rate": 3.6335160532498045e-05,
      "loss": 0.0089,
      "step": 4180
    },
    {
      "epoch": 3.28,
      "learning_rate": 3.594361785434612e-05,
      "loss": 0.0259,
      "step": 4190
    },
    {
      "epoch": 3.29,
      "learning_rate": 3.55520751761942e-05,
      "loss": 0.023,
      "step": 4200
    },
    {
      "epoch": 3.29,
      "eval_accuracy": 0.9109483423284502,
      "eval_loss": 0.36847907304763794,
      "eval_runtime": 42.1365,
      "eval_samples_per_second": 61.562,
      "eval_steps_per_second": 7.713,
      "step": 4200
    },
    {
      "epoch": 3.3,
      "learning_rate": 3.5160532498042285e-05,
      "loss": 0.0313,
      "step": 4210
    },
    {
      "epoch": 3.3,
      "learning_rate": 3.476898981989037e-05,
      "loss": 0.0552,
      "step": 4220
    },
    {
      "epoch": 3.31,
      "learning_rate": 3.437744714173845e-05,
      "loss": 0.0277,
      "step": 4230
    },
    {
      "epoch": 3.32,
      "learning_rate": 3.398590446358653e-05,
      "loss": 0.0224,
      "step": 4240
    },
    {
      "epoch": 3.33,
      "learning_rate": 3.3594361785434616e-05,
      "loss": 0.0351,
      "step": 4250
    },
    {
      "epoch": 3.34,
      "learning_rate": 3.32028191072827e-05,
      "loss": 0.0287,
      "step": 4260
    },
    {
      "epoch": 3.34,
      "learning_rate": 3.281127642913078e-05,
      "loss": 0.0322,
      "step": 4270
    },
    {
      "epoch": 3.35,
      "learning_rate": 3.2419733750978856e-05,
      "loss": 0.013,
      "step": 4280
    },
    {
      "epoch": 3.36,
      "learning_rate": 3.202819107282694e-05,
      "loss": 0.1024,
      "step": 4290
    },
    {
      "epoch": 3.37,
      "learning_rate": 3.163664839467502e-05,
      "loss": 0.0337,
      "step": 4300
    },
    {
      "epoch": 3.37,
      "eval_accuracy": 0.9109483423284502,
      "eval_loss": 0.3623192310333252,
      "eval_runtime": 42.3406,
      "eval_samples_per_second": 61.265,
      "eval_steps_per_second": 7.676,
      "step": 4300
    },
    {
      "epoch": 3.38,
      "learning_rate": 3.1245105716523104e-05,
      "loss": 0.0564,
      "step": 4310
    },
    {
      "epoch": 3.38,
      "learning_rate": 3.0853563038371186e-05,
      "loss": 0.1042,
      "step": 4320
    },
    {
      "epoch": 3.39,
      "learning_rate": 3.046202036021927e-05,
      "loss": 0.0885,
      "step": 4330
    },
    {
      "epoch": 3.4,
      "learning_rate": 3.0070477682067348e-05,
      "loss": 0.027,
      "step": 4340
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.9678935003915427e-05,
      "loss": 0.0763,
      "step": 4350
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.9287392325763506e-05,
      "loss": 0.0634,
      "step": 4360
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.889584964761159e-05,
      "loss": 0.0143,
      "step": 4370
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.850430696945967e-05,
      "loss": 0.1023,
      "step": 4380
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.8112764291307754e-05,
      "loss": 0.0347,
      "step": 4390
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.7721221613155833e-05,
      "loss": 0.0123,
      "step": 4400
    },
    {
      "epoch": 3.45,
      "eval_accuracy": 0.9067077872012336,
      "eval_loss": 0.3647214472293854,
      "eval_runtime": 41.0472,
      "eval_samples_per_second": 63.196,
      "eval_steps_per_second": 7.918,
      "step": 4400
    },
    {
      "epoch": 3.45,
      "learning_rate": 2.7329678935003916e-05,
      "loss": 0.0451,
      "step": 4410
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.6938136256851998e-05,
      "loss": 0.0478,
      "step": 4420
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.654659357870008e-05,
      "loss": 0.0391,
      "step": 4430
    },
    {
      "epoch": 3.48,
      "learning_rate": 2.615505090054816e-05,
      "loss": 0.0161,
      "step": 4440
    },
    {
      "epoch": 3.48,
      "learning_rate": 2.5763508222396242e-05,
      "loss": 0.019,
      "step": 4450
    },
    {
      "epoch": 3.49,
      "learning_rate": 2.5371965544244325e-05,
      "loss": 0.0477,
      "step": 4460
    },
    {
      "epoch": 3.5,
      "learning_rate": 2.4980422866092404e-05,
      "loss": 0.0198,
      "step": 4470
    },
    {
      "epoch": 3.51,
      "learning_rate": 2.4588880187940486e-05,
      "loss": 0.0154,
      "step": 4480
    },
    {
      "epoch": 3.52,
      "learning_rate": 2.419733750978857e-05,
      "loss": 0.0088,
      "step": 4490
    },
    {
      "epoch": 3.52,
      "learning_rate": 2.3805794831636648e-05,
      "loss": 0.0159,
      "step": 4500
    },
    {
      "epoch": 3.52,
      "eval_accuracy": 0.9082498072474943,
      "eval_loss": 0.36297959089279175,
      "eval_runtime": 40.6048,
      "eval_samples_per_second": 63.884,
      "eval_steps_per_second": 8.004,
      "step": 4500
    },
    {
      "epoch": 3.53,
      "learning_rate": 2.341425215348473e-05,
      "loss": 0.057,
      "step": 4510
    },
    {
      "epoch": 3.54,
      "learning_rate": 2.3022709475332813e-05,
      "loss": 0.0259,
      "step": 4520
    },
    {
      "epoch": 3.55,
      "learning_rate": 2.2631166797180896e-05,
      "loss": 0.0109,
      "step": 4530
    },
    {
      "epoch": 3.56,
      "learning_rate": 2.2239624119028975e-05,
      "loss": 0.1792,
      "step": 4540
    },
    {
      "epoch": 3.56,
      "learning_rate": 2.1848081440877057e-05,
      "loss": 0.0274,
      "step": 4550
    },
    {
      "epoch": 3.57,
      "learning_rate": 2.1456538762725137e-05,
      "loss": 0.0116,
      "step": 4560
    },
    {
      "epoch": 3.58,
      "learning_rate": 2.106499608457322e-05,
      "loss": 0.0094,
      "step": 4570
    },
    {
      "epoch": 3.59,
      "learning_rate": 2.06734534064213e-05,
      "loss": 0.0752,
      "step": 4580
    },
    {
      "epoch": 3.59,
      "learning_rate": 2.028191072826938e-05,
      "loss": 0.0436,
      "step": 4590
    },
    {
      "epoch": 3.6,
      "learning_rate": 1.9890368050117463e-05,
      "loss": 0.0154,
      "step": 4600
    },
    {
      "epoch": 3.6,
      "eval_accuracy": 0.9094063222821896,
      "eval_loss": 0.352169930934906,
      "eval_runtime": 40.8221,
      "eval_samples_per_second": 63.544,
      "eval_steps_per_second": 7.961,
      "step": 4600
    },
    {
      "epoch": 3.61,
      "learning_rate": 1.9498825371965546e-05,
      "loss": 0.0358,
      "step": 4610
    },
    {
      "epoch": 3.62,
      "learning_rate": 1.9107282693813628e-05,
      "loss": 0.0385,
      "step": 4620
    },
    {
      "epoch": 3.63,
      "learning_rate": 1.8715740015661707e-05,
      "loss": 0.0168,
      "step": 4630
    },
    {
      "epoch": 3.63,
      "learning_rate": 1.8324197337509787e-05,
      "loss": 0.017,
      "step": 4640
    },
    {
      "epoch": 3.64,
      "learning_rate": 1.793265465935787e-05,
      "loss": 0.0657,
      "step": 4650
    },
    {
      "epoch": 3.65,
      "learning_rate": 1.754111198120595e-05,
      "loss": 0.092,
      "step": 4660
    },
    {
      "epoch": 3.66,
      "learning_rate": 1.7149569303054034e-05,
      "loss": 0.0428,
      "step": 4670
    },
    {
      "epoch": 3.66,
      "learning_rate": 1.6758026624902117e-05,
      "loss": 0.0487,
      "step": 4680
    },
    {
      "epoch": 3.67,
      "learning_rate": 1.6366483946750196e-05,
      "loss": 0.009,
      "step": 4690
    },
    {
      "epoch": 3.68,
      "learning_rate": 1.597494126859828e-05,
      "loss": 0.0112,
      "step": 4700
    },
    {
      "epoch": 3.68,
      "eval_accuracy": 0.9163454124903624,
      "eval_loss": 0.3439232110977173,
      "eval_runtime": 40.4277,
      "eval_samples_per_second": 64.164,
      "eval_steps_per_second": 8.039,
      "step": 4700
    },
    {
      "epoch": 3.69,
      "learning_rate": 1.558339859044636e-05,
      "loss": 0.0141,
      "step": 4710
    },
    {
      "epoch": 3.7,
      "learning_rate": 1.5191855912294442e-05,
      "loss": 0.022,
      "step": 4720
    },
    {
      "epoch": 3.7,
      "learning_rate": 1.480031323414252e-05,
      "loss": 0.0074,
      "step": 4730
    },
    {
      "epoch": 3.71,
      "learning_rate": 1.4408770555990603e-05,
      "loss": 0.11,
      "step": 4740
    },
    {
      "epoch": 3.72,
      "learning_rate": 1.4017227877838684e-05,
      "loss": 0.0327,
      "step": 4750
    },
    {
      "epoch": 3.73,
      "learning_rate": 1.3625685199686767e-05,
      "loss": 0.0086,
      "step": 4760
    },
    {
      "epoch": 3.74,
      "learning_rate": 1.3234142521534848e-05,
      "loss": 0.0178,
      "step": 4770
    },
    {
      "epoch": 3.74,
      "learning_rate": 1.284259984338293e-05,
      "loss": 0.0363,
      "step": 4780
    },
    {
      "epoch": 3.75,
      "learning_rate": 1.2451057165231011e-05,
      "loss": 0.0114,
      "step": 4790
    },
    {
      "epoch": 3.76,
      "learning_rate": 1.2059514487079092e-05,
      "loss": 0.0219,
      "step": 4800
    },
    {
      "epoch": 3.76,
      "eval_accuracy": 0.9194294525828836,
      "eval_loss": 0.34038111567497253,
      "eval_runtime": 40.3468,
      "eval_samples_per_second": 64.293,
      "eval_steps_per_second": 8.055,
      "step": 4800
    },
    {
      "epoch": 3.77,
      "learning_rate": 1.1667971808927174e-05,
      "loss": 0.0136,
      "step": 4810
    },
    {
      "epoch": 3.77,
      "learning_rate": 1.1276429130775255e-05,
      "loss": 0.0083,
      "step": 4820
    },
    {
      "epoch": 3.78,
      "learning_rate": 1.0884886452623338e-05,
      "loss": 0.0432,
      "step": 4830
    },
    {
      "epoch": 3.79,
      "learning_rate": 1.0493343774471417e-05,
      "loss": 0.0437,
      "step": 4840
    },
    {
      "epoch": 3.8,
      "learning_rate": 1.01018010963195e-05,
      "loss": 0.0196,
      "step": 4850
    },
    {
      "epoch": 3.81,
      "learning_rate": 9.710258418167582e-06,
      "loss": 0.0566,
      "step": 4860
    },
    {
      "epoch": 3.81,
      "learning_rate": 9.318715740015663e-06,
      "loss": 0.0105,
      "step": 4870
    },
    {
      "epoch": 3.82,
      "learning_rate": 8.927173061863743e-06,
      "loss": 0.0672,
      "step": 4880
    },
    {
      "epoch": 3.83,
      "learning_rate": 8.535630383711824e-06,
      "loss": 0.0211,
      "step": 4890
    },
    {
      "epoch": 3.84,
      "learning_rate": 8.144087705559907e-06,
      "loss": 0.0183,
      "step": 4900
    },
    {
      "epoch": 3.84,
      "eval_accuracy": 0.9182729375481882,
      "eval_loss": 0.3370583951473236,
      "eval_runtime": 40.5395,
      "eval_samples_per_second": 63.987,
      "eval_steps_per_second": 8.017,
      "step": 4900
    },
    {
      "epoch": 3.84,
      "learning_rate": 7.752545027407988e-06,
      "loss": 0.0176,
      "step": 4910
    },
    {
      "epoch": 3.85,
      "learning_rate": 7.3610023492560685e-06,
      "loss": 0.0167,
      "step": 4920
    },
    {
      "epoch": 3.86,
      "learning_rate": 6.96945967110415e-06,
      "loss": 0.0162,
      "step": 4930
    },
    {
      "epoch": 3.87,
      "learning_rate": 6.577916992952232e-06,
      "loss": 0.0271,
      "step": 4940
    },
    {
      "epoch": 3.88,
      "learning_rate": 6.1863743148003135e-06,
      "loss": 0.0448,
      "step": 4950
    },
    {
      "epoch": 3.88,
      "learning_rate": 5.794831636648395e-06,
      "loss": 0.0082,
      "step": 4960
    },
    {
      "epoch": 3.89,
      "learning_rate": 5.403288958496476e-06,
      "loss": 0.0126,
      "step": 4970
    },
    {
      "epoch": 3.9,
      "learning_rate": 5.011746280344558e-06,
      "loss": 0.0121,
      "step": 4980
    },
    {
      "epoch": 3.91,
      "learning_rate": 4.620203602192639e-06,
      "loss": 0.009,
      "step": 4990
    },
    {
      "epoch": 3.92,
      "learning_rate": 4.22866092404072e-06,
      "loss": 0.0103,
      "step": 5000
    },
    {
      "epoch": 3.92,
      "eval_accuracy": 0.9182729375481882,
      "eval_loss": 0.33618906140327454,
      "eval_runtime": 41.0648,
      "eval_samples_per_second": 63.168,
      "eval_steps_per_second": 7.914,
      "step": 5000
    },
    {
      "epoch": 3.92,
      "learning_rate": 3.837118245888802e-06,
      "loss": 0.0123,
      "step": 5010
    },
    {
      "epoch": 3.93,
      "learning_rate": 3.4455755677368836e-06,
      "loss": 0.0266,
      "step": 5020
    },
    {
      "epoch": 3.94,
      "learning_rate": 3.054032889584965e-06,
      "loss": 0.0339,
      "step": 5030
    },
    {
      "epoch": 3.95,
      "learning_rate": 2.662490211433046e-06,
      "loss": 0.0484,
      "step": 5040
    },
    {
      "epoch": 3.95,
      "learning_rate": 2.2709475332811278e-06,
      "loss": 0.0948,
      "step": 5050
    },
    {
      "epoch": 3.96,
      "learning_rate": 1.8794048551292093e-06,
      "loss": 0.0101,
      "step": 5060
    },
    {
      "epoch": 3.97,
      "learning_rate": 1.4878621769772905e-06,
      "loss": 0.0559,
      "step": 5070
    },
    {
      "epoch": 3.98,
      "learning_rate": 1.096319498825372e-06,
      "loss": 0.02,
      "step": 5080
    },
    {
      "epoch": 3.99,
      "learning_rate": 7.047768206734534e-07,
      "loss": 0.028,
      "step": 5090
    },
    {
      "epoch": 3.99,
      "learning_rate": 3.1323414252153486e-07,
      "loss": 0.0357,
      "step": 5100
    },
    {
      "epoch": 3.99,
      "eval_accuracy": 0.9182729375481882,
      "eval_loss": 0.3352562189102173,
      "eval_runtime": 41.0843,
      "eval_samples_per_second": 63.138,
      "eval_steps_per_second": 7.911,
      "step": 5100
    },
    {
      "epoch": 4.0,
      "step": 5108,
      "total_flos": 3.1689424899978854e+18,
      "train_loss": 0.839469795966391,
      "train_runtime": 4434.1172,
      "train_samples_per_second": 9.21,
      "train_steps_per_second": 1.152
    }
  ],
  "max_steps": 5108,
  "num_train_epochs": 4,
  "total_flos": 3.1689424899978854e+18,
  "trial_name": null,
  "trial_params": null
}