{
  "best_metric": 0.051459942013025284,
  "best_model_checkpoint": "./vit-base-letter/checkpoint-2900",
  "epoch": 4.0,
  "global_step": 3252,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 0.0001993849938499385,
      "loss": 3.1242,
      "step": 10
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.000198769987699877,
      "loss": 2.6937,
      "step": 20
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00019815498154981552,
      "loss": 2.1871,
      "step": 30
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.000197539975399754,
      "loss": 1.7584,
      "step": 40
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0001969249692496925,
      "loss": 1.428,
      "step": 50
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.000196309963099631,
      "loss": 1.1022,
      "step": 60
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.0001956949569495695,
      "loss": 0.862,
      "step": 70
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.000195079950799508,
      "loss": 0.7286,
      "step": 80
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.00019446494464944652,
      "loss": 0.6413,
      "step": 90
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.000193849938499385,
      "loss": 0.5539,
      "step": 100
    },
    {
      "epoch": 0.12,
      "eval_accuracy": 0.9307692307692308,
      "eval_loss": 0.5576168298721313,
      "eval_runtime": 20.0625,
      "eval_samples_per_second": 129.595,
      "eval_steps_per_second": 8.125,
      "step": 100
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0001932349323493235,
      "loss": 0.5329,
      "step": 110
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.000192619926199262,
      "loss": 0.4424,
      "step": 120
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0001920049200492005,
      "loss": 0.3445,
      "step": 130
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00019138991389913902,
      "loss": 0.448,
      "step": 140
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00019077490774907748,
      "loss": 0.3616,
      "step": 150
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.000190159901599016,
      "loss": 0.422,
      "step": 160
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0001895448954489545,
      "loss": 0.3273,
      "step": 170
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.000188929889298893,
      "loss": 0.2256,
      "step": 180
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0001883148831488315,
      "loss": 0.1926,
      "step": 190
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00018769987699876998,
      "loss": 0.2688,
      "step": 200
    },
    {
      "epoch": 0.25,
      "eval_accuracy": 0.9665384615384616,
      "eval_loss": 0.23705258965492249,
      "eval_runtime": 19.9301,
      "eval_samples_per_second": 130.456,
      "eval_steps_per_second": 8.179,
      "step": 200
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0001870848708487085,
      "loss": 0.2367,
      "step": 210
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.000186469864698647,
      "loss": 0.2625,
      "step": 220
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.0001858548585485855,
      "loss": 0.2304,
      "step": 230
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.000185239852398524,
      "loss": 0.2889,
      "step": 240
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0001846248462484625,
      "loss": 0.1779,
      "step": 250
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00018400984009840098,
      "loss": 0.1514,
      "step": 260
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0001833948339483395,
      "loss": 0.2225,
      "step": 270
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00018277982779827797,
      "loss": 0.1753,
      "step": 280
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0001821648216482165,
      "loss": 0.1848,
      "step": 290
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00018154981549815499,
      "loss": 0.1568,
      "step": 300
    },
    {
      "epoch": 0.37,
      "eval_accuracy": 0.9688461538461538,
      "eval_loss": 0.18294188380241394,
      "eval_runtime": 19.7383,
      "eval_samples_per_second": 131.724,
      "eval_steps_per_second": 8.258,
      "step": 300
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00018093480934809348,
      "loss": 0.1902,
      "step": 310
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00018031980319803198,
      "loss": 0.1647,
      "step": 320
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0001797047970479705,
      "loss": 0.1953,
      "step": 330
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.000179089790897909,
      "loss": 0.2202,
      "step": 340
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.0001784747847478475,
      "loss": 0.1543,
      "step": 350
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00017785977859778598,
      "loss": 0.1254,
      "step": 360
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00017724477244772448,
      "loss": 0.225,
      "step": 370
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.000176629766297663,
      "loss": 0.2148,
      "step": 380
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00017601476014760147,
      "loss": 0.2064,
      "step": 390
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00017539975399754,
      "loss": 0.1684,
      "step": 400
    },
    {
      "epoch": 0.49,
      "eval_accuracy": 0.9661538461538461,
      "eval_loss": 0.16110503673553467,
      "eval_runtime": 19.5434,
      "eval_samples_per_second": 133.037,
      "eval_steps_per_second": 8.34,
      "step": 400
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00017478474784747848,
      "loss": 0.1313,
      "step": 410
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00017416974169741698,
      "loss": 0.1437,
      "step": 420
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0001736162361623616,
      "loss": 0.1067,
      "step": 430
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00017300123001230013,
      "loss": 0.1696,
      "step": 440
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.00017238622386223863,
      "loss": 0.1877,
      "step": 450
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.00017177121771217712,
      "loss": 0.1683,
      "step": 460
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00017115621156211564,
      "loss": 0.1345,
      "step": 470
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.00017054120541205414,
      "loss": 0.1548,
      "step": 480
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.00016992619926199263,
      "loss": 0.1133,
      "step": 490
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00016931119311193113,
      "loss": 0.1584,
      "step": 500
    },
    {
      "epoch": 0.62,
      "eval_accuracy": 0.9673076923076923,
      "eval_loss": 0.134025439620018,
      "eval_runtime": 19.5947,
      "eval_samples_per_second": 132.689,
      "eval_steps_per_second": 8.319,
      "step": 500
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00016869618696186962,
      "loss": 0.0944,
      "step": 510
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00016808118081180812,
      "loss": 0.1339,
      "step": 520
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00016746617466174664,
      "loss": 0.1224,
      "step": 530
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0001668511685116851,
      "loss": 0.0884,
      "step": 540
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.00016623616236162363,
      "loss": 0.0793,
      "step": 550
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00016562115621156212,
      "loss": 0.1709,
      "step": 560
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00016500615006150062,
      "loss": 0.1247,
      "step": 570
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00016439114391143914,
      "loss": 0.0932,
      "step": 580
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0001637761377613776,
      "loss": 0.1186,
      "step": 590
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00016316113161131613,
      "loss": 0.1569,
      "step": 600
    },
    {
      "epoch": 0.74,
      "eval_accuracy": 0.953076923076923,
      "eval_loss": 0.19328586757183075,
      "eval_runtime": 19.7536,
      "eval_samples_per_second": 131.621,
      "eval_steps_per_second": 8.252,
      "step": 600
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.00016254612546125462,
      "loss": 0.1604,
      "step": 610
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00016193111931119312,
      "loss": 0.0663,
      "step": 620
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.00016131611316113161,
      "loss": 0.2167,
      "step": 630
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.00016070110701107014,
      "loss": 0.095,
      "step": 640
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.0001600861008610086,
      "loss": 0.0874,
      "step": 650
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.00015947109471094713,
      "loss": 0.1631,
      "step": 660
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00015885608856088562,
      "loss": 0.1154,
      "step": 670
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00015824108241082411,
      "loss": 0.1664,
      "step": 680
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.00015762607626076264,
      "loss": 0.0859,
      "step": 690
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.0001570110701107011,
      "loss": 0.0992,
      "step": 700
    },
    {
      "epoch": 0.86,
      "eval_accuracy": 0.9780769230769231,
      "eval_loss": 0.10314588993787766,
      "eval_runtime": 19.6708,
      "eval_samples_per_second": 132.176,
      "eval_steps_per_second": 8.286,
      "step": 700
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.00015639606396063963,
      "loss": 0.0465,
      "step": 710
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00015578105781057812,
      "loss": 0.0781,
      "step": 720
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00015516605166051662,
      "loss": 0.1373,
      "step": 730
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0001545510455104551,
      "loss": 0.1474,
      "step": 740
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0001539360393603936,
      "loss": 0.1383,
      "step": 750
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.0001533210332103321,
      "loss": 0.1276,
      "step": 760
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.00015270602706027062,
      "loss": 0.1304,
      "step": 770
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.00015209102091020912,
      "loss": 0.1041,
      "step": 780
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0001514760147601476,
      "loss": 0.0746,
      "step": 790
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0001508610086100861,
      "loss": 0.0573,
      "step": 800
    },
    {
      "epoch": 0.98,
      "eval_accuracy": 0.9780769230769231,
      "eval_loss": 0.1024000346660614,
      "eval_runtime": 19.7326,
      "eval_samples_per_second": 131.762,
      "eval_steps_per_second": 8.26,
      "step": 800
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0001502460024600246,
      "loss": 0.068,
      "step": 810
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.00014963099630996312,
      "loss": 0.0615,
      "step": 820
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0001490159901599016,
      "loss": 0.0928,
      "step": 830
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.0001484009840098401,
      "loss": 0.1008,
      "step": 840
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0001477859778597786,
      "loss": 0.0815,
      "step": 850
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0001471709717097171,
      "loss": 0.0865,
      "step": 860
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.0001465559655596556,
      "loss": 0.0468,
      "step": 870
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.00014594095940959412,
      "loss": 0.0811,
      "step": 880
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0001453259532595326,
      "loss": 0.0475,
      "step": 890
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0001447109471094711,
      "loss": 0.0359,
      "step": 900
    },
    {
      "epoch": 1.11,
      "eval_accuracy": 0.9803846153846154,
      "eval_loss": 0.09497525542974472,
      "eval_runtime": 19.8135,
      "eval_samples_per_second": 131.224,
      "eval_steps_per_second": 8.227,
      "step": 900
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.0001440959409594096,
      "loss": 0.1011,
      "step": 910
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.0001434809348093481,
      "loss": 0.1215,
      "step": 920
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.00014286592865928662,
      "loss": 0.1678,
      "step": 930
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00014225092250922509,
      "loss": 0.0816,
      "step": 940
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.0001416359163591636,
      "loss": 0.1048,
      "step": 950
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0001410209102091021,
      "loss": 0.0386,
      "step": 960
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.0001404059040590406,
      "loss": 0.0852,
      "step": 970
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.0001397908979089791,
      "loss": 0.0613,
      "step": 980
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.00013917589175891759,
      "loss": 0.0738,
      "step": 990
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0001385608856088561,
      "loss": 0.0961,
      "step": 1000
    },
    {
      "epoch": 1.23,
      "eval_accuracy": 0.9723076923076923,
      "eval_loss": 0.11998701840639114,
      "eval_runtime": 19.8193,
      "eval_samples_per_second": 131.185,
      "eval_steps_per_second": 8.224,
      "step": 1000
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.0001379458794587946,
      "loss": 0.0705,
      "step": 1010
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0001373308733087331,
      "loss": 0.0944,
      "step": 1020
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0001367158671586716,
      "loss": 0.0626,
      "step": 1030
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.0001361008610086101,
      "loss": 0.0442,
      "step": 1040
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00013548585485854858,
      "loss": 0.1229,
      "step": 1050
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0001348708487084871,
      "loss": 0.0466,
      "step": 1060
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00013425584255842557,
      "loss": 0.0718,
      "step": 1070
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0001336408364083641,
      "loss": 0.0657,
      "step": 1080
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0001330258302583026,
      "loss": 0.0509,
      "step": 1090
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.00013241082410824108,
      "loss": 0.0334,
      "step": 1100
    },
    {
      "epoch": 1.35,
      "eval_accuracy": 0.975,
      "eval_loss": 0.09949370473623276,
      "eval_runtime": 19.8958,
      "eval_samples_per_second": 130.681,
      "eval_steps_per_second": 8.193,
      "step": 1100
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0001317958179581796,
      "loss": 0.0408,
      "step": 1110
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0001311808118081181,
      "loss": 0.0814,
      "step": 1120
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.0001305658056580566,
      "loss": 0.0706,
      "step": 1130
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0001299507995079951,
      "loss": 0.0224,
      "step": 1140
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00012933579335793358,
      "loss": 0.1008,
      "step": 1150
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.00012872078720787208,
      "loss": 0.0706,
      "step": 1160
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0001281057810578106,
      "loss": 0.1359,
      "step": 1170
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.00012749077490774907,
      "loss": 0.1311,
      "step": 1180
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.0001268757687576876,
      "loss": 0.053,
      "step": 1190
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.00012626076260762608,
      "loss": 0.0855,
      "step": 1200
    },
    {
      "epoch": 1.48,
      "eval_accuracy": 0.9815384615384616,
      "eval_loss": 0.07914509624242783,
      "eval_runtime": 20.4038,
      "eval_samples_per_second": 127.427,
      "eval_steps_per_second": 7.989,
      "step": 1200
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.00012564575645756458,
      "loss": 0.0763,
      "step": 1210
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0001250307503075031,
      "loss": 0.0159,
      "step": 1220
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.00012441574415744157,
      "loss": 0.0918,
      "step": 1230
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.0001238007380073801,
      "loss": 0.0489,
      "step": 1240
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.00012318573185731858,
      "loss": 0.0499,
      "step": 1250
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.00012257072570725708,
      "loss": 0.0372,
      "step": 1260
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.00012195571955719559,
      "loss": 0.0526,
      "step": 1270
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.00012134071340713408,
      "loss": 0.0638,
      "step": 1280
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.00012072570725707258,
      "loss": 0.0662,
      "step": 1290
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.00012011070110701108,
      "loss": 0.0902,
      "step": 1300
    },
    {
      "epoch": 1.6,
      "eval_accuracy": 0.9765384615384616,
      "eval_loss": 0.09805391728878021,
      "eval_runtime": 20.4238,
      "eval_samples_per_second": 127.302,
      "eval_steps_per_second": 7.981,
      "step": 1300
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.00011949569495694957,
      "loss": 0.0736,
      "step": 1310
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.00011888068880688807,
      "loss": 0.0986,
      "step": 1320
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.00011826568265682658,
      "loss": 0.0817,
      "step": 1330
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.00011765067650676506,
      "loss": 0.0586,
      "step": 1340
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.00011703567035670357,
      "loss": 0.0788,
      "step": 1350
    },
    {
      "epoch": 1.67,
      "learning_rate": 0.00011642066420664208,
      "loss": 0.0452,
      "step": 1360
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.00011580565805658057,
      "loss": 0.0125,
      "step": 1370
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.00011519065190651908,
      "loss": 0.1138,
      "step": 1380
    },
    {
      "epoch": 1.71,
      "learning_rate": 0.00011457564575645756,
      "loss": 0.0824,
      "step": 1390
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.00011396063960639607,
      "loss": 0.0583,
      "step": 1400
    },
    {
      "epoch": 1.72,
      "eval_accuracy": 0.9711538461538461,
      "eval_loss": 0.11919466406106949,
      "eval_runtime": 20.5384,
      "eval_samples_per_second": 126.592,
      "eval_steps_per_second": 7.936,
      "step": 1400
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.00011334563345633458,
      "loss": 0.0746,
      "step": 1410
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.00011273062730627306,
      "loss": 0.1108,
      "step": 1420
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.00011211562115621157,
      "loss": 0.1226,
      "step": 1430
    },
    {
      "epoch": 1.77,
      "learning_rate": 0.00011150061500615008,
      "loss": 0.1026,
      "step": 1440
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.00011088560885608856,
      "loss": 0.1145,
      "step": 1450
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.00011027060270602707,
      "loss": 0.0374,
      "step": 1460
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.00010965559655596556,
      "loss": 0.0525,
      "step": 1470
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.00010904059040590407,
      "loss": 0.0879,
      "step": 1480
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.00010842558425584256,
      "loss": 0.0417,
      "step": 1490
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.00010781057810578106,
      "loss": 0.0683,
      "step": 1500
    },
    {
      "epoch": 1.85,
      "eval_accuracy": 0.9846153846153847,
      "eval_loss": 0.06915987282991409,
      "eval_runtime": 20.393,
      "eval_samples_per_second": 127.495,
      "eval_steps_per_second": 7.993,
      "step": 1500
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.00010719557195571957,
      "loss": 0.0444,
      "step": 1510
    },
    {
      "epoch": 1.87,
      "learning_rate": 0.00010658056580565808,
      "loss": 0.0272,
      "step": 1520
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.00010596555965559656,
      "loss": 0.0615,
      "step": 1530
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.00010535055350553507,
      "loss": 0.0254,
      "step": 1540
    },
    {
      "epoch": 1.91,
      "learning_rate": 0.00010473554735547355,
      "loss": 0.0911,
      "step": 1550
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.00010412054120541206,
      "loss": 0.0585,
      "step": 1560
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.00010350553505535056,
      "loss": 0.1168,
      "step": 1570
    },
    {
      "epoch": 1.94,
      "learning_rate": 0.00010289052890528906,
      "loss": 0.0898,
      "step": 1580
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.00010227552275522757,
      "loss": 0.0568,
      "step": 1590
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.00010166051660516606,
      "loss": 0.1188,
      "step": 1600
    },
    {
      "epoch": 1.97,
      "eval_accuracy": 0.9784615384615385,
      "eval_loss": 0.09314684569835663,
      "eval_runtime": 20.4297,
      "eval_samples_per_second": 127.266,
      "eval_steps_per_second": 7.979,
      "step": 1600
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.00010104551045510456,
      "loss": 0.0545,
      "step": 1610
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.00010043050430504306,
      "loss": 0.0571,
      "step": 1620
    },
    {
      "epoch": 2.0,
      "learning_rate": 9.981549815498156e-05,
      "loss": 0.0525,
      "step": 1630
    },
    {
      "epoch": 2.02,
      "learning_rate": 9.920049200492005e-05,
      "loss": 0.0253,
      "step": 1640
    },
    {
      "epoch": 2.03,
      "learning_rate": 9.858548585485855e-05,
      "loss": 0.0148,
      "step": 1650
    },
    {
      "epoch": 2.04,
      "learning_rate": 9.797047970479706e-05,
      "loss": 0.0338,
      "step": 1660
    },
    {
      "epoch": 2.05,
      "learning_rate": 9.735547355473555e-05,
      "loss": 0.0078,
      "step": 1670
    },
    {
      "epoch": 2.07,
      "learning_rate": 9.674046740467405e-05,
      "loss": 0.0407,
      "step": 1680
    },
    {
      "epoch": 2.08,
      "learning_rate": 9.612546125461255e-05,
      "loss": 0.0654,
      "step": 1690
    },
    {
      "epoch": 2.09,
      "learning_rate": 9.551045510455105e-05,
      "loss": 0.0366,
      "step": 1700
    },
    {
      "epoch": 2.09,
      "eval_accuracy": 0.9803846153846154,
      "eval_loss": 0.09191113710403442,
      "eval_runtime": 20.4659,
      "eval_samples_per_second": 127.041,
      "eval_steps_per_second": 7.964,
      "step": 1700
    },
    {
      "epoch": 2.1,
      "learning_rate": 9.489544895448956e-05,
      "loss": 0.0548,
      "step": 1710
    },
    {
      "epoch": 2.12,
      "learning_rate": 9.428044280442805e-05,
      "loss": 0.0473,
      "step": 1720
    },
    {
      "epoch": 2.13,
      "learning_rate": 9.366543665436655e-05,
      "loss": 0.0664,
      "step": 1730
    },
    {
      "epoch": 2.14,
      "learning_rate": 9.305043050430505e-05,
      "loss": 0.0333,
      "step": 1740
    },
    {
      "epoch": 2.15,
      "learning_rate": 9.243542435424355e-05,
      "loss": 0.0527,
      "step": 1750
    },
    {
      "epoch": 2.16,
      "learning_rate": 9.182041820418204e-05,
      "loss": 0.0532,
      "step": 1760
    },
    {
      "epoch": 2.18,
      "learning_rate": 9.120541205412054e-05,
      "loss": 0.0172,
      "step": 1770
    },
    {
      "epoch": 2.19,
      "learning_rate": 9.059040590405905e-05,
      "loss": 0.0396,
      "step": 1780
    },
    {
      "epoch": 2.2,
      "learning_rate": 8.997539975399754e-05,
      "loss": 0.0656,
      "step": 1790
    },
    {
      "epoch": 2.21,
      "learning_rate": 8.936039360393605e-05,
      "loss": 0.0276,
      "step": 1800
    },
    {
      "epoch": 2.21,
      "eval_accuracy": 0.9846153846153847,
      "eval_loss": 0.06672599911689758,
      "eval_runtime": 20.3533,
      "eval_samples_per_second": 127.743,
      "eval_steps_per_second": 8.009,
      "step": 1800
    },
    {
      "epoch": 2.23,
      "learning_rate": 8.874538745387454e-05,
      "loss": 0.043,
      "step": 1810
    },
    {
      "epoch": 2.24,
      "learning_rate": 8.813038130381305e-05,
      "loss": 0.0156,
      "step": 1820
    },
    {
      "epoch": 2.25,
      "learning_rate": 8.751537515375155e-05,
      "loss": 0.0472,
      "step": 1830
    },
    {
      "epoch": 2.26,
      "learning_rate": 8.690036900369004e-05,
      "loss": 0.0278,
      "step": 1840
    },
    {
      "epoch": 2.28,
      "learning_rate": 8.628536285362854e-05,
      "loss": 0.0376,
      "step": 1850
    },
    {
      "epoch": 2.29,
      "learning_rate": 8.567035670356705e-05,
      "loss": 0.0764,
      "step": 1860
    },
    {
      "epoch": 2.3,
      "learning_rate": 8.505535055350554e-05,
      "loss": 0.0819,
      "step": 1870
    },
    {
      "epoch": 2.31,
      "learning_rate": 8.444034440344403e-05,
      "loss": 0.0277,
      "step": 1880
    },
    {
      "epoch": 2.32,
      "learning_rate": 8.382533825338253e-05,
      "loss": 0.0471,
      "step": 1890
    },
    {
      "epoch": 2.34,
      "learning_rate": 8.321033210332104e-05,
      "loss": 0.0309,
      "step": 1900
    },
    {
      "epoch": 2.34,
      "eval_accuracy": 0.9857692307692307,
      "eval_loss": 0.059869736433029175,
      "eval_runtime": 20.2748,
      "eval_samples_per_second": 128.238,
      "eval_steps_per_second": 8.04,
      "step": 1900
    },
    {
      "epoch": 2.35,
      "learning_rate": 8.259532595325955e-05,
      "loss": 0.0217,
      "step": 1910
    },
    {
      "epoch": 2.36,
      "learning_rate": 8.198031980319804e-05,
      "loss": 0.02,
      "step": 1920
    },
    {
      "epoch": 2.37,
      "learning_rate": 8.136531365313654e-05,
      "loss": 0.0179,
      "step": 1930
    },
    {
      "epoch": 2.39,
      "learning_rate": 8.075030750307504e-05,
      "loss": 0.0511,
      "step": 1940
    },
    {
      "epoch": 2.4,
      "learning_rate": 8.013530135301354e-05,
      "loss": 0.037,
      "step": 1950
    },
    {
      "epoch": 2.41,
      "learning_rate": 7.952029520295203e-05,
      "loss": 0.0609,
      "step": 1960
    },
    {
      "epoch": 2.42,
      "learning_rate": 7.890528905289053e-05,
      "loss": 0.0392,
      "step": 1970
    },
    {
      "epoch": 2.44,
      "learning_rate": 7.829028290282904e-05,
      "loss": 0.0322,
      "step": 1980
    },
    {
      "epoch": 2.45,
      "learning_rate": 7.767527675276753e-05,
      "loss": 0.0366,
      "step": 1990
    },
    {
      "epoch": 2.46,
      "learning_rate": 7.706027060270603e-05,
      "loss": 0.0183,
      "step": 2000
    },
    {
      "epoch": 2.46,
      "eval_accuracy": 0.9769230769230769,
      "eval_loss": 0.08916281163692474,
      "eval_runtime": 20.1987,
      "eval_samples_per_second": 128.721,
      "eval_steps_per_second": 8.07,
      "step": 2000
    },
    {
      "epoch": 2.47,
      "learning_rate": 7.644526445264453e-05,
      "loss": 0.0555,
      "step": 2010
    },
    {
      "epoch": 2.48,
      "learning_rate": 7.583025830258303e-05,
      "loss": 0.0125,
      "step": 2020
    },
    {
      "epoch": 2.5,
      "learning_rate": 7.521525215252154e-05,
      "loss": 0.0155,
      "step": 2030
    },
    {
      "epoch": 2.51,
      "learning_rate": 7.460024600246003e-05,
      "loss": 0.005,
      "step": 2040
    },
    {
      "epoch": 2.52,
      "learning_rate": 7.398523985239853e-05,
      "loss": 0.0193,
      "step": 2050
    },
    {
      "epoch": 2.53,
      "learning_rate": 7.337023370233703e-05,
      "loss": 0.0513,
      "step": 2060
    },
    {
      "epoch": 2.55,
      "learning_rate": 7.275522755227553e-05,
      "loss": 0.0618,
      "step": 2070
    },
    {
      "epoch": 2.56,
      "learning_rate": 7.214022140221402e-05,
      "loss": 0.0601,
      "step": 2080
    },
    {
      "epoch": 2.57,
      "learning_rate": 7.152521525215252e-05,
      "loss": 0.02,
      "step": 2090
    },
    {
      "epoch": 2.58,
      "learning_rate": 7.091020910209103e-05,
      "loss": 0.0431,
      "step": 2100
    },
    {
      "epoch": 2.58,
      "eval_accuracy": 0.985,
      "eval_loss": 0.06632479280233383,
      "eval_runtime": 19.6556,
      "eval_samples_per_second": 132.278,
      "eval_steps_per_second": 8.293,
      "step": 2100
    },
    {
      "epoch": 2.6,
      "learning_rate": 7.029520295202952e-05,
      "loss": 0.0327,
      "step": 2110
    },
    {
      "epoch": 2.61,
      "learning_rate": 6.968019680196803e-05,
      "loss": 0.0119,
      "step": 2120
    },
    {
      "epoch": 2.62,
      "learning_rate": 6.906519065190652e-05,
      "loss": 0.0558,
      "step": 2130
    },
    {
      "epoch": 2.63,
      "learning_rate": 6.845018450184503e-05,
      "loss": 0.0283,
      "step": 2140
    },
    {
      "epoch": 2.64,
      "learning_rate": 6.783517835178353e-05,
      "loss": 0.0451,
      "step": 2150
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.722017220172202e-05,
      "loss": 0.0484,
      "step": 2160
    },
    {
      "epoch": 2.67,
      "learning_rate": 6.660516605166052e-05,
      "loss": 0.0319,
      "step": 2170
    },
    {
      "epoch": 2.68,
      "learning_rate": 6.599015990159902e-05,
      "loss": 0.0496,
      "step": 2180
    },
    {
      "epoch": 2.69,
      "learning_rate": 6.537515375153752e-05,
      "loss": 0.0128,
      "step": 2190
    },
    {
      "epoch": 2.71,
      "learning_rate": 6.476014760147601e-05,
      "loss": 0.0424,
      "step": 2200
    },
    {
      "epoch": 2.71,
      "eval_accuracy": 0.9861538461538462,
      "eval_loss": 0.0642935037612915,
      "eval_runtime": 20.0743,
      "eval_samples_per_second": 129.519,
      "eval_steps_per_second": 8.12,
      "step": 2200
    },
    {
      "epoch": 2.72,
      "learning_rate": 6.414514145141451e-05,
      "loss": 0.063,
      "step": 2210
    },
    {
      "epoch": 2.73,
      "learning_rate": 6.353013530135302e-05,
      "loss": 0.0555,
      "step": 2220
    },
    {
      "epoch": 2.74,
      "learning_rate": 6.291512915129151e-05,
      "loss": 0.0488,
      "step": 2230
    },
    {
      "epoch": 2.76,
      "learning_rate": 6.230012300123002e-05,
      "loss": 0.0197,
      "step": 2240
    },
    {
      "epoch": 2.77,
      "learning_rate": 6.168511685116851e-05,
      "loss": 0.0157,
      "step": 2250
    },
    {
      "epoch": 2.78,
      "learning_rate": 6.107011070110702e-05,
      "loss": 0.0359,
      "step": 2260
    },
    {
      "epoch": 2.79,
      "learning_rate": 6.045510455104552e-05,
      "loss": 0.0218,
      "step": 2270
    },
    {
      "epoch": 2.8,
      "learning_rate": 5.984009840098401e-05,
      "loss": 0.0176,
      "step": 2280
    },
    {
      "epoch": 2.82,
      "learning_rate": 5.922509225092251e-05,
      "loss": 0.0146,
      "step": 2290
    },
    {
      "epoch": 2.83,
      "learning_rate": 5.8610086100861016e-05,
      "loss": 0.0453,
      "step": 2300
    },
    {
      "epoch": 2.83,
      "eval_accuracy": 0.9861538461538462,
      "eval_loss": 0.06456407159566879,
      "eval_runtime": 19.4018,
      "eval_samples_per_second": 134.008,
      "eval_steps_per_second": 8.401,
      "step": 2300
    },
    {
      "epoch": 2.84,
      "learning_rate": 5.799507995079951e-05,
      "loss": 0.0617,
      "step": 2310
    },
    {
      "epoch": 2.85,
      "learning_rate": 5.738007380073801e-05,
      "loss": 0.046,
      "step": 2320
    },
    {
      "epoch": 2.87,
      "learning_rate": 5.6765067650676506e-05,
      "loss": 0.0127,
      "step": 2330
    },
    {
      "epoch": 2.88,
      "learning_rate": 5.6150061500615015e-05,
      "loss": 0.0337,
      "step": 2340
    },
    {
      "epoch": 2.89,
      "learning_rate": 5.553505535055351e-05,
      "loss": 0.0428,
      "step": 2350
    },
    {
      "epoch": 2.9,
      "learning_rate": 5.492004920049201e-05,
      "loss": 0.0395,
      "step": 2360
    },
    {
      "epoch": 2.92,
      "learning_rate": 5.4305043050430506e-05,
      "loss": 0.0145,
      "step": 2370
    },
    {
      "epoch": 2.93,
      "learning_rate": 5.3690036900369014e-05,
      "loss": 0.0131,
      "step": 2380
    },
    {
      "epoch": 2.94,
      "learning_rate": 5.307503075030751e-05,
      "loss": 0.0316,
      "step": 2390
    },
    {
      "epoch": 2.95,
      "learning_rate": 5.2460024600246e-05,
      "loss": 0.0528,
      "step": 2400
    },
    {
      "epoch": 2.95,
      "eval_accuracy": 0.985,
      "eval_loss": 0.05503399297595024,
      "eval_runtime": 19.633,
      "eval_samples_per_second": 132.43,
      "eval_steps_per_second": 8.302,
      "step": 2400
    },
    {
      "epoch": 2.96,
      "learning_rate": 5.1845018450184505e-05,
      "loss": 0.0179,
      "step": 2410
    },
    {
      "epoch": 2.98,
      "learning_rate": 5.1230012300123006e-05,
      "loss": 0.0185,
      "step": 2420
    },
    {
      "epoch": 2.99,
      "learning_rate": 5.061500615006151e-05,
      "loss": 0.0151,
      "step": 2430
    },
    {
      "epoch": 3.0,
      "learning_rate": 5e-05,
      "loss": 0.0501,
      "step": 2440
    },
    {
      "epoch": 3.01,
      "learning_rate": 4.93849938499385e-05,
      "loss": 0.0043,
      "step": 2450
    },
    {
      "epoch": 3.03,
      "learning_rate": 4.8769987699877e-05,
      "loss": 0.0085,
      "step": 2460
    },
    {
      "epoch": 3.04,
      "learning_rate": 4.81549815498155e-05,
      "loss": 0.0068,
      "step": 2470
    },
    {
      "epoch": 3.05,
      "learning_rate": 4.7539975399754e-05,
      "loss": 0.0222,
      "step": 2480
    },
    {
      "epoch": 3.06,
      "learning_rate": 4.6924969249692496e-05,
      "loss": 0.022,
      "step": 2490
    },
    {
      "epoch": 3.08,
      "learning_rate": 4.6309963099631e-05,
      "loss": 0.0045,
      "step": 2500
    },
    {
      "epoch": 3.08,
      "eval_accuracy": 0.9846153846153847,
      "eval_loss": 0.05792374908924103,
      "eval_runtime": 19.8321,
      "eval_samples_per_second": 131.101,
      "eval_steps_per_second": 8.219,
      "step": 2500
    },
    {
      "epoch": 3.09,
      "learning_rate": 4.569495694956949e-05,
      "loss": 0.0161,
      "step": 2510
    },
    {
      "epoch": 3.1,
      "learning_rate": 4.5079950799507994e-05,
      "loss": 0.0409,
      "step": 2520
    },
    {
      "epoch": 3.11,
      "learning_rate": 4.4464944649446495e-05,
      "loss": 0.0364,
      "step": 2530
    },
    {
      "epoch": 3.12,
      "learning_rate": 4.3849938499385e-05,
      "loss": 0.0069,
      "step": 2540
    },
    {
      "epoch": 3.14,
      "learning_rate": 4.323493234932349e-05,
      "loss": 0.0091,
      "step": 2550
    },
    {
      "epoch": 3.15,
      "learning_rate": 4.261992619926199e-05,
      "loss": 0.0037,
      "step": 2560
    },
    {
      "epoch": 3.16,
      "learning_rate": 4.2004920049200495e-05,
      "loss": 0.045,
      "step": 2570
    },
    {
      "epoch": 3.17,
      "learning_rate": 4.1389913899138996e-05,
      "loss": 0.0038,
      "step": 2580
    },
    {
      "epoch": 3.19,
      "learning_rate": 4.077490774907749e-05,
      "loss": 0.0035,
      "step": 2590
    },
    {
      "epoch": 3.2,
      "learning_rate": 4.015990159901599e-05,
      "loss": 0.007,
      "step": 2600
    },
    {
      "epoch": 3.2,
      "eval_accuracy": 0.9884615384615385,
      "eval_loss": 0.05168744549155235,
      "eval_runtime": 20.5897,
      "eval_samples_per_second": 126.277,
      "eval_steps_per_second": 7.917,
      "step": 2600
    },
    {
      "epoch": 3.21,
      "learning_rate": 3.954489544895449e-05,
      "loss": 0.0043,
      "step": 2610
    },
    {
      "epoch": 3.22,
      "learning_rate": 3.892988929889299e-05,
      "loss": 0.0466,
      "step": 2620
    },
    {
      "epoch": 3.23,
      "learning_rate": 3.831488314883149e-05,
      "loss": 0.0055,
      "step": 2630
    },
    {
      "epoch": 3.25,
      "learning_rate": 3.769987699876999e-05,
      "loss": 0.0037,
      "step": 2640
    },
    {
      "epoch": 3.26,
      "learning_rate": 3.7084870848708486e-05,
      "loss": 0.005,
      "step": 2650
    },
    {
      "epoch": 3.27,
      "learning_rate": 3.646986469864699e-05,
      "loss": 0.0082,
      "step": 2660
    },
    {
      "epoch": 3.28,
      "learning_rate": 3.585485854858548e-05,
      "loss": 0.0223,
      "step": 2670
    },
    {
      "epoch": 3.3,
      "learning_rate": 3.5239852398523984e-05,
      "loss": 0.0034,
      "step": 2680
    },
    {
      "epoch": 3.31,
      "learning_rate": 3.4624846248462485e-05,
      "loss": 0.0053,
      "step": 2690
    },
    {
      "epoch": 3.32,
      "learning_rate": 3.400984009840099e-05,
      "loss": 0.0048,
      "step": 2700
    },
    {
      "epoch": 3.32,
      "eval_accuracy": 0.9865384615384616,
      "eval_loss": 0.058377496898174286,
      "eval_runtime": 20.5674,
      "eval_samples_per_second": 126.413,
      "eval_steps_per_second": 7.925,
      "step": 2700
    },
    {
      "epoch": 3.33,
      "learning_rate": 3.339483394833948e-05,
      "loss": 0.0075,
      "step": 2710
    },
    {
      "epoch": 3.35,
      "learning_rate": 3.277982779827798e-05,
      "loss": 0.0092,
      "step": 2720
    },
    {
      "epoch": 3.36,
      "learning_rate": 3.2164821648216484e-05,
      "loss": 0.0164,
      "step": 2730
    },
    {
      "epoch": 3.37,
      "learning_rate": 3.1549815498154986e-05,
      "loss": 0.0063,
      "step": 2740
    },
    {
      "epoch": 3.38,
      "learning_rate": 3.093480934809348e-05,
      "loss": 0.0195,
      "step": 2750
    },
    {
      "epoch": 3.39,
      "learning_rate": 3.0319803198031982e-05,
      "loss": 0.0563,
      "step": 2760
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.970479704797048e-05,
      "loss": 0.003,
      "step": 2770
    },
    {
      "epoch": 3.42,
      "learning_rate": 2.908979089790898e-05,
      "loss": 0.0423,
      "step": 2780
    },
    {
      "epoch": 3.43,
      "learning_rate": 2.8474784747847476e-05,
      "loss": 0.0035,
      "step": 2790
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.7859778597785978e-05,
      "loss": 0.019,
      "step": 2800
    },
    {
      "epoch": 3.44,
      "eval_accuracy": 0.9873076923076923,
      "eval_loss": 0.05602681264281273,
      "eval_runtime": 20.8189,
      "eval_samples_per_second": 124.886,
      "eval_steps_per_second": 7.829,
      "step": 2800
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.7244772447724476e-05,
      "loss": 0.0066,
      "step": 2810
    },
    {
      "epoch": 3.47,
      "learning_rate": 2.6629766297662977e-05,
      "loss": 0.0059,
      "step": 2820
    },
    {
      "epoch": 3.48,
      "learning_rate": 2.6014760147601475e-05,
      "loss": 0.0035,
      "step": 2830
    },
    {
      "epoch": 3.49,
      "learning_rate": 2.5399753997539977e-05,
      "loss": 0.0276,
      "step": 2840
    },
    {
      "epoch": 3.51,
      "learning_rate": 2.4784747847478475e-05,
      "loss": 0.0174,
      "step": 2850
    },
    {
      "epoch": 3.52,
      "learning_rate": 2.4169741697416977e-05,
      "loss": 0.0041,
      "step": 2860
    },
    {
      "epoch": 3.53,
      "learning_rate": 2.3554735547355475e-05,
      "loss": 0.0174,
      "step": 2870
    },
    {
      "epoch": 3.54,
      "learning_rate": 2.2939729397293973e-05,
      "loss": 0.0059,
      "step": 2880
    },
    {
      "epoch": 3.55,
      "learning_rate": 2.2324723247232474e-05,
      "loss": 0.0141,
      "step": 2890
    },
    {
      "epoch": 3.57,
      "learning_rate": 2.1709717097170972e-05,
      "loss": 0.0038,
      "step": 2900
    },
    {
      "epoch": 3.57,
      "eval_accuracy": 0.9880769230769231,
      "eval_loss": 0.051459942013025284,
      "eval_runtime": 20.7979,
      "eval_samples_per_second": 125.013,
      "eval_steps_per_second": 7.837,
      "step": 2900
    },
    {
      "epoch": 3.58,
      "learning_rate": 2.1094710947109474e-05,
      "loss": 0.0105,
      "step": 2910
    },
    {
      "epoch": 3.59,
      "learning_rate": 2.0479704797047972e-05,
      "loss": 0.0036,
      "step": 2920
    },
    {
      "epoch": 3.6,
      "learning_rate": 1.986469864698647e-05,
      "loss": 0.035,
      "step": 2930
    },
    {
      "epoch": 3.62,
      "learning_rate": 1.924969249692497e-05,
      "loss": 0.027,
      "step": 2940
    },
    {
      "epoch": 3.63,
      "learning_rate": 1.863468634686347e-05,
      "loss": 0.0062,
      "step": 2950
    },
    {
      "epoch": 3.64,
      "learning_rate": 1.8019680196801968e-05,
      "loss": 0.0183,
      "step": 2960
    },
    {
      "epoch": 3.65,
      "learning_rate": 1.740467404674047e-05,
      "loss": 0.0031,
      "step": 2970
    },
    {
      "epoch": 3.67,
      "learning_rate": 1.6789667896678967e-05,
      "loss": 0.0027,
      "step": 2980
    },
    {
      "epoch": 3.68,
      "learning_rate": 1.617466174661747e-05,
      "loss": 0.0037,
      "step": 2990
    },
    {
      "epoch": 3.69,
      "learning_rate": 1.5559655596555967e-05,
      "loss": 0.0219,
      "step": 3000
    },
    {
      "epoch": 3.69,
      "eval_accuracy": 0.9880769230769231,
      "eval_loss": 0.05274398624897003,
      "eval_runtime": 20.6528,
      "eval_samples_per_second": 125.891,
      "eval_steps_per_second": 7.892,
      "step": 3000
    },
    {
      "epoch": 3.7,
      "learning_rate": 1.4944649446494467e-05,
      "loss": 0.0038,
      "step": 3010
    },
    {
      "epoch": 3.71,
      "learning_rate": 1.4329643296432965e-05,
      "loss": 0.0445,
      "step": 3020
    },
    {
      "epoch": 3.73,
      "learning_rate": 1.3714637146371464e-05,
      "loss": 0.004,
      "step": 3030
    },
    {
      "epoch": 3.74,
      "learning_rate": 1.3099630996309964e-05,
      "loss": 0.0065,
      "step": 3040
    },
    {
      "epoch": 3.75,
      "learning_rate": 1.2484624846248464e-05,
      "loss": 0.0029,
      "step": 3050
    },
    {
      "epoch": 3.76,
      "learning_rate": 1.1869618696186962e-05,
      "loss": 0.0055,
      "step": 3060
    },
    {
      "epoch": 3.78,
      "learning_rate": 1.1254612546125462e-05,
      "loss": 0.007,
      "step": 3070
    },
    {
      "epoch": 3.79,
      "learning_rate": 1.0639606396063962e-05,
      "loss": 0.0099,
      "step": 3080
    },
    {
      "epoch": 3.8,
      "learning_rate": 1.0024600246002461e-05,
      "loss": 0.0067,
      "step": 3090
    },
    {
      "epoch": 3.81,
      "learning_rate": 9.40959409594096e-06,
      "loss": 0.0117,
      "step": 3100
    },
    {
      "epoch": 3.81,
      "eval_accuracy": 0.9888461538461538,
      "eval_loss": 0.05230804532766342,
      "eval_runtime": 20.2347,
      "eval_samples_per_second": 128.492,
      "eval_steps_per_second": 8.055,
      "step": 3100
    },
    {
      "epoch": 3.83,
      "learning_rate": 8.79458794587946e-06,
      "loss": 0.0271,
      "step": 3110
    },
    {
      "epoch": 3.84,
      "learning_rate": 8.179581795817959e-06,
      "loss": 0.0219,
      "step": 3120
    },
    {
      "epoch": 3.85,
      "learning_rate": 7.564575645756458e-06,
      "loss": 0.0055,
      "step": 3130
    },
    {
      "epoch": 3.86,
      "learning_rate": 6.949569495694958e-06,
      "loss": 0.0031,
      "step": 3140
    },
    {
      "epoch": 3.87,
      "learning_rate": 6.334563345633457e-06,
      "loss": 0.0115,
      "step": 3150
    },
    {
      "epoch": 3.89,
      "learning_rate": 5.7195571955719566e-06,
      "loss": 0.006,
      "step": 3160
    },
    {
      "epoch": 3.9,
      "learning_rate": 5.1045510455104555e-06,
      "loss": 0.0101,
      "step": 3170
    },
    {
      "epoch": 3.91,
      "learning_rate": 4.489544895448955e-06,
      "loss": 0.0197,
      "step": 3180
    },
    {
      "epoch": 3.92,
      "learning_rate": 3.874538745387454e-06,
      "loss": 0.0201,
      "step": 3190
    },
    {
      "epoch": 3.94,
      "learning_rate": 3.2595325953259536e-06,
      "loss": 0.0035,
      "step": 3200
    },
    {
      "epoch": 3.94,
      "eval_accuracy": 0.9865384615384616,
      "eval_loss": 0.05585360527038574,
      "eval_runtime": 20.2068,
      "eval_samples_per_second": 128.67,
      "eval_steps_per_second": 8.067,
      "step": 3200
    },
    {
      "epoch": 3.95,
      "learning_rate": 2.6445264452644525e-06,
      "loss": 0.0056,
      "step": 3210
    },
    {
      "epoch": 3.96,
      "learning_rate": 2.029520295202952e-06,
      "loss": 0.0221,
      "step": 3220
    },
    {
      "epoch": 3.97,
      "learning_rate": 1.4145141451414515e-06,
      "loss": 0.0238,
      "step": 3230
    },
    {
      "epoch": 3.99,
      "learning_rate": 7.995079950799507e-07,
      "loss": 0.0264,
      "step": 3240
    },
    {
      "epoch": 4.0,
      "learning_rate": 1.845018450184502e-07,
      "loss": 0.0255,
      "step": 3250
    },
    {
      "epoch": 4.0,
      "step": 3252,
      "total_flos": 8.0609004822528e+18,
      "train_loss": 0.11576663036481334,
      "train_runtime": 3043.19,
      "train_samples_per_second": 34.175,
      "train_steps_per_second": 1.069
    }
  ],
  "max_steps": 3252,
  "num_train_epochs": 4,
  "total_flos": 8.0609004822528e+18,
  "trial_name": null,
  "trial_params": null
}