{
  "best_metric": 0.9945371961698993,
  "best_model_checkpoint": "swin-tiny-patch4-window7-224-plant-disease-new/checkpoint-2196",
  "epoch": 5.987730061349693,
  "eval_steps": 500,
  "global_step": 2196,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.03,
      "learning_rate": 2.2727272727272728e-06,
      "loss": 3.7243,
      "step": 10
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.5454545454545455e-06,
      "loss": 3.6642,
      "step": 20
    },
    {
      "epoch": 0.08,
      "learning_rate": 6.818181818181818e-06,
      "loss": 3.5713,
      "step": 30
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.090909090909091e-06,
      "loss": 3.4156,
      "step": 40
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.1363636363636365e-05,
      "loss": 3.2078,
      "step": 50
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.3636363636363637e-05,
      "loss": 2.9585,
      "step": 60
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.590909090909091e-05,
      "loss": 2.5413,
      "step": 70
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.8181818181818182e-05,
      "loss": 2.1267,
      "step": 80
    },
    {
      "epoch": 0.25,
      "learning_rate": 2.0454545454545457e-05,
      "loss": 1.6404,
      "step": 90
    },
    {
      "epoch": 0.27,
      "learning_rate": 2.272727272727273e-05,
      "loss": 1.24,
      "step": 100
    },
    {
      "epoch": 0.3,
      "learning_rate": 2.5e-05,
      "loss": 0.9935,
      "step": 110
    },
    {
      "epoch": 0.33,
      "learning_rate": 2.7272727272727273e-05,
      "loss": 0.7996,
      "step": 120
    },
    {
      "epoch": 0.35,
      "learning_rate": 2.954545454545455e-05,
      "loss": 0.6445,
      "step": 130
    },
    {
      "epoch": 0.38,
      "learning_rate": 3.181818181818182e-05,
      "loss": 0.5497,
      "step": 140
    },
    {
      "epoch": 0.41,
      "learning_rate": 3.409090909090909e-05,
      "loss": 0.4696,
      "step": 150
    },
    {
      "epoch": 0.44,
      "learning_rate": 3.6363636363636364e-05,
      "loss": 0.4347,
      "step": 160
    },
    {
      "epoch": 0.46,
      "learning_rate": 3.8636363636363636e-05,
      "loss": 0.3752,
      "step": 170
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.0909090909090915e-05,
      "loss": 0.3534,
      "step": 180
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.318181818181819e-05,
      "loss": 0.3373,
      "step": 190
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.545454545454546e-05,
      "loss": 0.3061,
      "step": 200
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.772727272727273e-05,
      "loss": 0.2794,
      "step": 210
    },
    {
      "epoch": 0.6,
      "learning_rate": 5e-05,
      "loss": 0.2683,
      "step": 220
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.974696356275304e-05,
      "loss": 0.2566,
      "step": 230
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.9493927125506076e-05,
      "loss": 0.2368,
      "step": 240
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.924089068825911e-05,
      "loss": 0.2404,
      "step": 250
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.898785425101215e-05,
      "loss": 0.2411,
      "step": 260
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.8734817813765186e-05,
      "loss": 0.2121,
      "step": 270
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.848178137651822e-05,
      "loss": 0.2067,
      "step": 280
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.822874493927126e-05,
      "loss": 0.2171,
      "step": 290
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.797570850202429e-05,
      "loss": 0.1911,
      "step": 300
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.772267206477733e-05,
      "loss": 0.1926,
      "step": 310
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.746963562753037e-05,
      "loss": 0.1872,
      "step": 320
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.72165991902834e-05,
      "loss": 0.1808,
      "step": 330
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.6963562753036435e-05,
      "loss": 0.1596,
      "step": 340
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.671052631578948e-05,
      "loss": 0.1731,
      "step": 350
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.6457489878542516e-05,
      "loss": 0.1728,
      "step": 360
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9750184139454947,
      "eval_loss": 0.07616888731718063,
      "eval_runtime": 118.4949,
      "eval_samples_per_second": 137.491,
      "eval_steps_per_second": 1.376,
      "step": 366
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.6204453441295545e-05,
      "loss": 0.1655,
      "step": 370
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.595141700404859e-05,
      "loss": 0.1495,
      "step": 380
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.5698380566801625e-05,
      "loss": 0.1614,
      "step": 390
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.5445344129554655e-05,
      "loss": 0.1435,
      "step": 400
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.519230769230769e-05,
      "loss": 0.1485,
      "step": 410
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.4939271255060735e-05,
      "loss": 0.1526,
      "step": 420
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.4686234817813765e-05,
      "loss": 0.1461,
      "step": 430
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.44331983805668e-05,
      "loss": 0.1534,
      "step": 440
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.4180161943319845e-05,
      "loss": 0.144,
      "step": 450
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.3927125506072875e-05,
      "loss": 0.1539,
      "step": 460
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.367408906882591e-05,
      "loss": 0.1198,
      "step": 470
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.342105263157895e-05,
      "loss": 0.1312,
      "step": 480
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.3168016194331985e-05,
      "loss": 0.133,
      "step": 490
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.291497975708502e-05,
      "loss": 0.1274,
      "step": 500
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.266194331983806e-05,
      "loss": 0.1306,
      "step": 510
    },
    {
      "epoch": 1.42,
      "learning_rate": 4.2408906882591095e-05,
      "loss": 0.1442,
      "step": 520
    },
    {
      "epoch": 1.45,
      "learning_rate": 4.215587044534413e-05,
      "loss": 0.1272,
      "step": 530
    },
    {
      "epoch": 1.47,
      "learning_rate": 4.190283400809717e-05,
      "loss": 0.1244,
      "step": 540
    },
    {
      "epoch": 1.5,
      "learning_rate": 4.1649797570850205e-05,
      "loss": 0.1138,
      "step": 550
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.139676113360324e-05,
      "loss": 0.1235,
      "step": 560
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.114372469635628e-05,
      "loss": 0.1349,
      "step": 570
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.089068825910931e-05,
      "loss": 0.1329,
      "step": 580
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.063765182186235e-05,
      "loss": 0.1173,
      "step": 590
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.038461538461539e-05,
      "loss": 0.1101,
      "step": 600
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.0131578947368425e-05,
      "loss": 0.1145,
      "step": 610
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.9878542510121455e-05,
      "loss": 0.1136,
      "step": 620
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.96255060728745e-05,
      "loss": 0.1141,
      "step": 630
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.9372469635627535e-05,
      "loss": 0.1168,
      "step": 640
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.9119433198380565e-05,
      "loss": 0.1034,
      "step": 650
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.886639676113361e-05,
      "loss": 0.1227,
      "step": 660
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.8613360323886645e-05,
      "loss": 0.1155,
      "step": 670
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.8360323886639675e-05,
      "loss": 0.0965,
      "step": 680
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.810728744939271e-05,
      "loss": 0.1192,
      "step": 690
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.7854251012145755e-05,
      "loss": 0.0992,
      "step": 700
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.7601214574898785e-05,
      "loss": 0.1018,
      "step": 710
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.734817813765182e-05,
      "loss": 0.1134,
      "step": 720
    },
    {
      "epoch": 1.99,
      "learning_rate": 3.7095141700404865e-05,
      "loss": 0.1022,
      "step": 730
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9855143628774858,
      "eval_loss": 0.042967669665813446,
      "eval_runtime": 117.5684,
      "eval_samples_per_second": 138.575,
      "eval_steps_per_second": 1.386,
      "step": 733
    },
    {
      "epoch": 2.02,
      "learning_rate": 3.6842105263157895e-05,
      "loss": 0.094,
      "step": 740
    },
    {
      "epoch": 2.04,
      "learning_rate": 3.658906882591093e-05,
      "loss": 0.1053,
      "step": 750
    },
    {
      "epoch": 2.07,
      "learning_rate": 3.633603238866397e-05,
      "loss": 0.0836,
      "step": 760
    },
    {
      "epoch": 2.1,
      "learning_rate": 3.6082995951417005e-05,
      "loss": 0.1101,
      "step": 770
    },
    {
      "epoch": 2.13,
      "learning_rate": 3.582995951417004e-05,
      "loss": 0.0958,
      "step": 780
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.557692307692308e-05,
      "loss": 0.0855,
      "step": 790
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.5323886639676115e-05,
      "loss": 0.0926,
      "step": 800
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.507085020242915e-05,
      "loss": 0.0971,
      "step": 810
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.481781376518219e-05,
      "loss": 0.0965,
      "step": 820
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.4564777327935225e-05,
      "loss": 0.094,
      "step": 830
    },
    {
      "epoch": 2.29,
      "learning_rate": 3.431174089068826e-05,
      "loss": 0.0762,
      "step": 840
    },
    {
      "epoch": 2.32,
      "learning_rate": 3.40587044534413e-05,
      "loss": 0.1017,
      "step": 850
    },
    {
      "epoch": 2.34,
      "learning_rate": 3.3805668016194335e-05,
      "loss": 0.0767,
      "step": 860
    },
    {
      "epoch": 2.37,
      "learning_rate": 3.355263157894737e-05,
      "loss": 0.0871,
      "step": 870
    },
    {
      "epoch": 2.4,
      "learning_rate": 3.329959514170041e-05,
      "loss": 0.0968,
      "step": 880
    },
    {
      "epoch": 2.43,
      "learning_rate": 3.3046558704453444e-05,
      "loss": 0.0881,
      "step": 890
    },
    {
      "epoch": 2.45,
      "learning_rate": 3.279352226720648e-05,
      "loss": 0.0911,
      "step": 900
    },
    {
      "epoch": 2.48,
      "learning_rate": 3.254048582995952e-05,
      "loss": 0.0912,
      "step": 910
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.2287449392712554e-05,
      "loss": 0.0937,
      "step": 920
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.2034412955465584e-05,
      "loss": 0.0884,
      "step": 930
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.178137651821863e-05,
      "loss": 0.0975,
      "step": 940
    },
    {
      "epoch": 2.59,
      "learning_rate": 3.1528340080971664e-05,
      "loss": 0.0827,
      "step": 950
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.1275303643724694e-05,
      "loss": 0.0873,
      "step": 960
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.102226720647773e-05,
      "loss": 0.0835,
      "step": 970
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.0769230769230774e-05,
      "loss": 0.0833,
      "step": 980
    },
    {
      "epoch": 2.7,
      "learning_rate": 3.0516194331983804e-05,
      "loss": 0.0728,
      "step": 990
    },
    {
      "epoch": 2.73,
      "learning_rate": 3.0263157894736844e-05,
      "loss": 0.0901,
      "step": 1000
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.001012145748988e-05,
      "loss": 0.0836,
      "step": 1010
    },
    {
      "epoch": 2.78,
      "learning_rate": 2.9757085020242914e-05,
      "loss": 0.0822,
      "step": 1020
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.950404858299595e-05,
      "loss": 0.072,
      "step": 1030
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.925101214574899e-05,
      "loss": 0.0844,
      "step": 1040
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.8997975708502024e-05,
      "loss": 0.0781,
      "step": 1050
    },
    {
      "epoch": 2.89,
      "learning_rate": 2.874493927125506e-05,
      "loss": 0.0809,
      "step": 1060
    },
    {
      "epoch": 2.92,
      "learning_rate": 2.84919028340081e-05,
      "loss": 0.0907,
      "step": 1070
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.823886639676113e-05,
      "loss": 0.0987,
      "step": 1080
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.798582995951417e-05,
      "loss": 0.0882,
      "step": 1090
    },
    {
      "epoch": 3.0,
      "learning_rate": 2.7732793522267207e-05,
      "loss": 0.0806,
      "step": 1100
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9912226859808495,
      "eval_loss": 0.02648293785750866,
      "eval_runtime": 118.1333,
      "eval_samples_per_second": 137.912,
      "eval_steps_per_second": 1.38,
      "step": 1100
    },
    {
      "epoch": 3.03,
      "learning_rate": 2.7479757085020247e-05,
      "loss": 0.0795,
      "step": 1110
    },
    {
      "epoch": 3.05,
      "learning_rate": 2.722672064777328e-05,
      "loss": 0.07,
      "step": 1120
    },
    {
      "epoch": 3.08,
      "learning_rate": 2.6973684210526317e-05,
      "loss": 0.0783,
      "step": 1130
    },
    {
      "epoch": 3.11,
      "learning_rate": 2.6720647773279357e-05,
      "loss": 0.0709,
      "step": 1140
    },
    {
      "epoch": 3.14,
      "learning_rate": 2.6467611336032387e-05,
      "loss": 0.0728,
      "step": 1150
    },
    {
      "epoch": 3.16,
      "learning_rate": 2.6214574898785427e-05,
      "loss": 0.0779,
      "step": 1160
    },
    {
      "epoch": 3.19,
      "learning_rate": 2.5961538461538464e-05,
      "loss": 0.0713,
      "step": 1170
    },
    {
      "epoch": 3.22,
      "learning_rate": 2.5708502024291497e-05,
      "loss": 0.0873,
      "step": 1180
    },
    {
      "epoch": 3.24,
      "learning_rate": 2.5455465587044537e-05,
      "loss": 0.0582,
      "step": 1190
    },
    {
      "epoch": 3.27,
      "learning_rate": 2.5202429149797574e-05,
      "loss": 0.0717,
      "step": 1200
    },
    {
      "epoch": 3.3,
      "learning_rate": 2.494939271255061e-05,
      "loss": 0.061,
      "step": 1210
    },
    {
      "epoch": 3.33,
      "learning_rate": 2.4696356275303644e-05,
      "loss": 0.0777,
      "step": 1220
    },
    {
      "epoch": 3.35,
      "learning_rate": 2.444331983805668e-05,
      "loss": 0.0801,
      "step": 1230
    },
    {
      "epoch": 3.38,
      "learning_rate": 2.4190283400809717e-05,
      "loss": 0.0713,
      "step": 1240
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.3937246963562754e-05,
      "loss": 0.0777,
      "step": 1250
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.368421052631579e-05,
      "loss": 0.0569,
      "step": 1260
    },
    {
      "epoch": 3.46,
      "learning_rate": 2.3431174089068827e-05,
      "loss": 0.0731,
      "step": 1270
    },
    {
      "epoch": 3.49,
      "learning_rate": 2.3178137651821864e-05,
      "loss": 0.0715,
      "step": 1280
    },
    {
      "epoch": 3.52,
      "learning_rate": 2.29251012145749e-05,
      "loss": 0.0755,
      "step": 1290
    },
    {
      "epoch": 3.54,
      "learning_rate": 2.2672064777327937e-05,
      "loss": 0.0685,
      "step": 1300
    },
    {
      "epoch": 3.57,
      "learning_rate": 2.241902834008097e-05,
      "loss": 0.0634,
      "step": 1310
    },
    {
      "epoch": 3.6,
      "learning_rate": 2.216599190283401e-05,
      "loss": 0.073,
      "step": 1320
    },
    {
      "epoch": 3.63,
      "learning_rate": 2.1912955465587047e-05,
      "loss": 0.0736,
      "step": 1330
    },
    {
      "epoch": 3.65,
      "learning_rate": 2.165991902834008e-05,
      "loss": 0.0604,
      "step": 1340
    },
    {
      "epoch": 3.68,
      "learning_rate": 2.140688259109312e-05,
      "loss": 0.0824,
      "step": 1350
    },
    {
      "epoch": 3.71,
      "learning_rate": 2.1153846153846154e-05,
      "loss": 0.0566,
      "step": 1360
    },
    {
      "epoch": 3.74,
      "learning_rate": 2.090080971659919e-05,
      "loss": 0.0576,
      "step": 1370
    },
    {
      "epoch": 3.76,
      "learning_rate": 2.0647773279352227e-05,
      "loss": 0.0668,
      "step": 1380
    },
    {
      "epoch": 3.79,
      "learning_rate": 2.0394736842105264e-05,
      "loss": 0.0726,
      "step": 1390
    },
    {
      "epoch": 3.82,
      "learning_rate": 2.0141700404858304e-05,
      "loss": 0.0606,
      "step": 1400
    },
    {
      "epoch": 3.84,
      "learning_rate": 1.9888663967611337e-05,
      "loss": 0.0676,
      "step": 1410
    },
    {
      "epoch": 3.87,
      "learning_rate": 1.9635627530364373e-05,
      "loss": 0.0657,
      "step": 1420
    },
    {
      "epoch": 3.9,
      "learning_rate": 1.938259109311741e-05,
      "loss": 0.0644,
      "step": 1430
    },
    {
      "epoch": 3.93,
      "learning_rate": 1.9129554655870447e-05,
      "loss": 0.0568,
      "step": 1440
    },
    {
      "epoch": 3.95,
      "learning_rate": 1.8876518218623483e-05,
      "loss": 0.0758,
      "step": 1450
    },
    {
      "epoch": 3.98,
      "learning_rate": 1.862348178137652e-05,
      "loss": 0.0623,
      "step": 1460
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9930640805303216,
      "eval_loss": 0.022850118577480316,
      "eval_runtime": 117.8576,
      "eval_samples_per_second": 138.235,
      "eval_steps_per_second": 1.383,
      "step": 1467
    },
    {
      "epoch": 4.01,
      "learning_rate": 1.8370445344129557e-05,
      "loss": 0.0697,
      "step": 1470
    },
    {
      "epoch": 4.04,
      "learning_rate": 1.811740890688259e-05,
      "loss": 0.0646,
      "step": 1480
    },
    {
      "epoch": 4.06,
      "learning_rate": 1.786437246963563e-05,
      "loss": 0.0639,
      "step": 1490
    },
    {
      "epoch": 4.09,
      "learning_rate": 1.7611336032388663e-05,
      "loss": 0.0683,
      "step": 1500
    },
    {
      "epoch": 4.12,
      "learning_rate": 1.73582995951417e-05,
      "loss": 0.0627,
      "step": 1510
    },
    {
      "epoch": 4.14,
      "learning_rate": 1.7105263157894737e-05,
      "loss": 0.0641,
      "step": 1520
    },
    {
      "epoch": 4.17,
      "learning_rate": 1.6852226720647773e-05,
      "loss": 0.062,
      "step": 1530
    },
    {
      "epoch": 4.2,
      "learning_rate": 1.6599190283400813e-05,
      "loss": 0.0625,
      "step": 1540
    },
    {
      "epoch": 4.23,
      "learning_rate": 1.6346153846153847e-05,
      "loss": 0.0588,
      "step": 1550
    },
    {
      "epoch": 4.25,
      "learning_rate": 1.6093117408906883e-05,
      "loss": 0.0575,
      "step": 1560
    },
    {
      "epoch": 4.28,
      "learning_rate": 1.584008097165992e-05,
      "loss": 0.066,
      "step": 1570
    },
    {
      "epoch": 4.31,
      "learning_rate": 1.5587044534412957e-05,
      "loss": 0.0536,
      "step": 1580
    },
    {
      "epoch": 4.34,
      "learning_rate": 1.5334008097165993e-05,
      "loss": 0.0618,
      "step": 1590
    },
    {
      "epoch": 4.36,
      "learning_rate": 1.508097165991903e-05,
      "loss": 0.0502,
      "step": 1600
    },
    {
      "epoch": 4.39,
      "learning_rate": 1.4827935222672065e-05,
      "loss": 0.0472,
      "step": 1610
    },
    {
      "epoch": 4.42,
      "learning_rate": 1.4574898785425101e-05,
      "loss": 0.0644,
      "step": 1620
    },
    {
      "epoch": 4.44,
      "learning_rate": 1.4321862348178138e-05,
      "loss": 0.0577,
      "step": 1630
    },
    {
      "epoch": 4.47,
      "learning_rate": 1.4068825910931175e-05,
      "loss": 0.0618,
      "step": 1640
    },
    {
      "epoch": 4.5,
      "learning_rate": 1.3815789473684213e-05,
      "loss": 0.0566,
      "step": 1650
    },
    {
      "epoch": 4.53,
      "learning_rate": 1.3562753036437248e-05,
      "loss": 0.0576,
      "step": 1660
    },
    {
      "epoch": 4.55,
      "learning_rate": 1.3309716599190283e-05,
      "loss": 0.05,
      "step": 1670
    },
    {
      "epoch": 4.58,
      "learning_rate": 1.3056680161943321e-05,
      "loss": 0.0692,
      "step": 1680
    },
    {
      "epoch": 4.61,
      "learning_rate": 1.2803643724696356e-05,
      "loss": 0.0621,
      "step": 1690
    },
    {
      "epoch": 4.64,
      "learning_rate": 1.2550607287449393e-05,
      "loss": 0.0545,
      "step": 1700
    },
    {
      "epoch": 4.66,
      "learning_rate": 1.229757085020243e-05,
      "loss": 0.0592,
      "step": 1710
    },
    {
      "epoch": 4.69,
      "learning_rate": 1.2044534412955466e-05,
      "loss": 0.0544,
      "step": 1720
    },
    {
      "epoch": 4.72,
      "learning_rate": 1.1791497975708503e-05,
      "loss": 0.0662,
      "step": 1730
    },
    {
      "epoch": 4.74,
      "learning_rate": 1.153846153846154e-05,
      "loss": 0.0548,
      "step": 1740
    },
    {
      "epoch": 4.77,
      "learning_rate": 1.1285425101214574e-05,
      "loss": 0.0512,
      "step": 1750
    },
    {
      "epoch": 4.8,
      "learning_rate": 1.1032388663967611e-05,
      "loss": 0.0529,
      "step": 1760
    },
    {
      "epoch": 4.83,
      "learning_rate": 1.077935222672065e-05,
      "loss": 0.05,
      "step": 1770
    },
    {
      "epoch": 4.85,
      "learning_rate": 1.0526315789473684e-05,
      "loss": 0.0505,
      "step": 1780
    },
    {
      "epoch": 4.88,
      "learning_rate": 1.0273279352226721e-05,
      "loss": 0.0491,
      "step": 1790
    },
    {
      "epoch": 4.91,
      "learning_rate": 1.0020242914979758e-05,
      "loss": 0.0597,
      "step": 1800
    },
    {
      "epoch": 4.94,
      "learning_rate": 9.767206477732794e-06,
      "loss": 0.0594,
      "step": 1810
    },
    {
      "epoch": 4.96,
      "learning_rate": 9.51417004048583e-06,
      "loss": 0.0486,
      "step": 1820
    },
    {
      "epoch": 4.99,
      "learning_rate": 9.261133603238866e-06,
      "loss": 0.046,
      "step": 1830
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.9937392585317948,
      "eval_loss": 0.018908195197582245,
      "eval_runtime": 117.9858,
      "eval_samples_per_second": 138.084,
      "eval_steps_per_second": 1.382,
      "step": 1833
    },
    {
      "epoch": 5.02,
      "learning_rate": 9.008097165991904e-06,
      "loss": 0.0566,
      "step": 1840
    },
    {
      "epoch": 5.04,
      "learning_rate": 8.75506072874494e-06,
      "loss": 0.0488,
      "step": 1850
    },
    {
      "epoch": 5.07,
      "learning_rate": 8.502024291497976e-06,
      "loss": 0.0577,
      "step": 1860
    },
    {
      "epoch": 5.1,
      "learning_rate": 8.248987854251013e-06,
      "loss": 0.0557,
      "step": 1870
    },
    {
      "epoch": 5.13,
      "learning_rate": 7.99595141700405e-06,
      "loss": 0.0452,
      "step": 1880
    },
    {
      "epoch": 5.15,
      "learning_rate": 7.742914979757084e-06,
      "loss": 0.0551,
      "step": 1890
    },
    {
      "epoch": 5.18,
      "learning_rate": 7.489878542510122e-06,
      "loss": 0.0529,
      "step": 1900
    },
    {
      "epoch": 5.21,
      "learning_rate": 7.236842105263158e-06,
      "loss": 0.0535,
      "step": 1910
    },
    {
      "epoch": 5.24,
      "learning_rate": 6.983805668016195e-06,
      "loss": 0.0493,
      "step": 1920
    },
    {
      "epoch": 5.26,
      "learning_rate": 6.730769230769231e-06,
      "loss": 0.0488,
      "step": 1930
    },
    {
      "epoch": 5.29,
      "learning_rate": 6.4777327935222675e-06,
      "loss": 0.0515,
      "step": 1940
    },
    {
      "epoch": 5.32,
      "learning_rate": 6.224696356275303e-06,
      "loss": 0.0469,
      "step": 1950
    },
    {
      "epoch": 5.34,
      "learning_rate": 5.971659919028341e-06,
      "loss": 0.0546,
      "step": 1960
    },
    {
      "epoch": 5.37,
      "learning_rate": 5.718623481781377e-06,
      "loss": 0.0567,
      "step": 1970
    },
    {
      "epoch": 5.4,
      "learning_rate": 5.465587044534413e-06,
      "loss": 0.0427,
      "step": 1980
    },
    {
      "epoch": 5.43,
      "learning_rate": 5.21255060728745e-06,
      "loss": 0.0479,
      "step": 1990
    },
    {
      "epoch": 5.45,
      "learning_rate": 4.9595141700404865e-06,
      "loss": 0.0505,
      "step": 2000
    },
    {
      "epoch": 5.48,
      "learning_rate": 4.706477732793522e-06,
      "loss": 0.0589,
      "step": 2010
    },
    {
      "epoch": 5.51,
      "learning_rate": 4.453441295546559e-06,
      "loss": 0.0487,
      "step": 2020
    },
    {
      "epoch": 5.54,
      "learning_rate": 4.200404858299596e-06,
      "loss": 0.0492,
      "step": 2030
    },
    {
      "epoch": 5.56,
      "learning_rate": 3.9473684210526315e-06,
      "loss": 0.0543,
      "step": 2040
    },
    {
      "epoch": 5.59,
      "learning_rate": 3.6943319838056685e-06,
      "loss": 0.0445,
      "step": 2050
    },
    {
      "epoch": 5.62,
      "learning_rate": 3.4412955465587043e-06,
      "loss": 0.0477,
      "step": 2060
    },
    {
      "epoch": 5.64,
      "learning_rate": 3.1882591093117414e-06,
      "loss": 0.0491,
      "step": 2070
    },
    {
      "epoch": 5.67,
      "learning_rate": 2.9352226720647772e-06,
      "loss": 0.0355,
      "step": 2080
    },
    {
      "epoch": 5.7,
      "learning_rate": 2.682186234817814e-06,
      "loss": 0.043,
      "step": 2090
    },
    {
      "epoch": 5.73,
      "learning_rate": 2.4291497975708505e-06,
      "loss": 0.0448,
      "step": 2100
    },
    {
      "epoch": 5.75,
      "learning_rate": 2.1761133603238867e-06,
      "loss": 0.0389,
      "step": 2110
    },
    {
      "epoch": 5.78,
      "learning_rate": 1.9230769230769234e-06,
      "loss": 0.0537,
      "step": 2120
    },
    {
      "epoch": 5.81,
      "learning_rate": 1.6700404858299598e-06,
      "loss": 0.0471,
      "step": 2130
    },
    {
      "epoch": 5.84,
      "learning_rate": 1.417004048582996e-06,
      "loss": 0.0503,
      "step": 2140
    },
    {
      "epoch": 5.86,
      "learning_rate": 1.1639676113360325e-06,
      "loss": 0.0391,
      "step": 2150
    },
    {
      "epoch": 5.89,
      "learning_rate": 9.109311740890688e-07,
      "loss": 0.0491,
      "step": 2160
    },
    {
      "epoch": 5.92,
      "learning_rate": 6.578947368421053e-07,
      "loss": 0.0438,
      "step": 2170
    },
    {
      "epoch": 5.94,
      "learning_rate": 4.048582995951417e-07,
      "loss": 0.058,
      "step": 2180
    },
    {
      "epoch": 5.97,
      "learning_rate": 1.5182186234817816e-07,
      "loss": 0.0471,
      "step": 2190
    },
    {
      "epoch": 5.99,
      "eval_accuracy": 0.9945371961698993,
      "eval_loss": 0.016341065987944603,
      "eval_runtime": 118.4482,
      "eval_samples_per_second": 137.545,
      "eval_steps_per_second": 1.376,
      "step": 2196
    },
    {
      "epoch": 5.99,
      "step": 2196,
      "total_flos": 2.184600884938408e+19,
      "train_loss": 0.2333228224803624,
      "train_runtime": 14131.7358,
      "train_samples_per_second": 62.253,
      "train_steps_per_second": 0.155
    }
  ],
  "logging_steps": 10,
  "max_steps": 2196,
  "num_train_epochs": 6,
  "save_steps": 500,
  "total_flos": 2.184600884938408e+19,
  "trial_name": null,
  "trial_params": null
}