{
  "best_metric": 0.9634788241962912,
  "best_model_checkpoint": "swinv2-tiny-patch4-window8-256-finetuned-og_dataset_5e/checkpoint-2730",
  "epoch": 4.998628257887518,
  "global_step": 2730,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 1.8315018315018316e-06,
      "loss": 1.4501,
      "step": 10
    },
    {
      "epoch": 0.04,
      "learning_rate": 3.663003663003663e-06,
      "loss": 1.3654,
      "step": 20
    },
    {
      "epoch": 0.05,
      "learning_rate": 5.494505494505494e-06,
      "loss": 1.2803,
      "step": 30
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.326007326007326e-06,
      "loss": 1.1795,
      "step": 40
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.157509157509158e-06,
      "loss": 1.038,
      "step": 50
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.0989010989010989e-05,
      "loss": 0.9104,
      "step": 60
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.282051282051282e-05,
      "loss": 0.841,
      "step": 70
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.4652014652014653e-05,
      "loss": 0.8229,
      "step": 80
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.6483516483516486e-05,
      "loss": 0.8016,
      "step": 90
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.8315018315018315e-05,
      "loss": 0.7767,
      "step": 100
    },
    {
      "epoch": 0.2,
      "learning_rate": 2.0146520146520148e-05,
      "loss": 0.768,
      "step": 110
    },
    {
      "epoch": 0.22,
      "learning_rate": 2.1978021978021977e-05,
      "loss": 0.732,
      "step": 120
    },
    {
      "epoch": 0.24,
      "learning_rate": 2.380952380952381e-05,
      "loss": 0.7581,
      "step": 130
    },
    {
      "epoch": 0.26,
      "learning_rate": 2.564102564102564e-05,
      "loss": 0.7529,
      "step": 140
    },
    {
      "epoch": 0.27,
      "learning_rate": 2.7472527472527476e-05,
      "loss": 0.7273,
      "step": 150
    },
    {
      "epoch": 0.29,
      "learning_rate": 2.9304029304029305e-05,
      "loss": 0.7354,
      "step": 160
    },
    {
      "epoch": 0.31,
      "learning_rate": 3.113553113553114e-05,
      "loss": 0.6923,
      "step": 170
    },
    {
      "epoch": 0.33,
      "learning_rate": 3.296703296703297e-05,
      "loss": 0.7213,
      "step": 180
    },
    {
      "epoch": 0.35,
      "learning_rate": 3.47985347985348e-05,
      "loss": 0.6918,
      "step": 190
    },
    {
      "epoch": 0.37,
      "learning_rate": 3.663003663003663e-05,
      "loss": 0.6491,
      "step": 200
    },
    {
      "epoch": 0.38,
      "learning_rate": 3.846153846153846e-05,
      "loss": 0.6922,
      "step": 210
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.0293040293040296e-05,
      "loss": 0.6825,
      "step": 220
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.212454212454213e-05,
      "loss": 0.696,
      "step": 230
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.3956043956043955e-05,
      "loss": 0.6681,
      "step": 240
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.578754578754579e-05,
      "loss": 0.6535,
      "step": 250
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.761904761904762e-05,
      "loss": 0.6698,
      "step": 260
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.945054945054945e-05,
      "loss": 0.654,
      "step": 270
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.985754985754986e-05,
      "loss": 0.6673,
      "step": 280
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.965404965404965e-05,
      "loss": 0.6415,
      "step": 290
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.945054945054945e-05,
      "loss": 0.6207,
      "step": 300
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.9247049247049247e-05,
      "loss": 0.6549,
      "step": 310
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.904354904354905e-05,
      "loss": 0.5926,
      "step": 320
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.884004884004884e-05,
      "loss": 0.5947,
      "step": 330
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.863654863654864e-05,
      "loss": 0.641,
      "step": 340
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.8433048433048433e-05,
      "loss": 0.5779,
      "step": 350
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.8229548229548234e-05,
      "loss": 0.5551,
      "step": 360
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.802604802604803e-05,
      "loss": 0.5699,
      "step": 370
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.782254782254783e-05,
      "loss": 0.5725,
      "step": 380
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.761904761904762e-05,
      "loss": 0.6402,
      "step": 390
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.741554741554742e-05,
      "loss": 0.5886,
      "step": 400
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.7212047212047214e-05,
      "loss": 0.5955,
      "step": 410
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.700854700854701e-05,
      "loss": 0.5773,
      "step": 420
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.680504680504681e-05,
      "loss": 0.5897,
      "step": 430
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.66015466015466e-05,
      "loss": 0.565,
      "step": 440
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.63980463980464e-05,
      "loss": 0.5424,
      "step": 450
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.6194546194546194e-05,
      "loss": 0.545,
      "step": 460
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.5991045991045994e-05,
      "loss": 0.53,
      "step": 470
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.578754578754579e-05,
      "loss": 0.5327,
      "step": 480
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.558404558404559e-05,
      "loss": 0.5538,
      "step": 490
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.538054538054538e-05,
      "loss": 0.5104,
      "step": 500
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.517704517704518e-05,
      "loss": 0.5277,
      "step": 510
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.4973544973544974e-05,
      "loss": 0.5374,
      "step": 520
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.4770044770044775e-05,
      "loss": 0.5149,
      "step": 530
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.4566544566544575e-05,
      "loss": 0.5003,
      "step": 540
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8383674791481092,
      "eval_loss": 0.38816481828689575,
      "eval_runtime": 218.2336,
      "eval_samples_per_second": 56.586,
      "eval_steps_per_second": 1.769,
      "step": 546
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.436304436304437e-05,
      "loss": 0.5277,
      "step": 550
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.415954415954416e-05,
      "loss": 0.4732,
      "step": 560
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.3956043956043955e-05,
      "loss": 0.4788,
      "step": 570
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.3752543752543755e-05,
      "loss": 0.5161,
      "step": 580
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.354904354904355e-05,
      "loss": 0.5127,
      "step": 590
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.334554334554335e-05,
      "loss": 0.5108,
      "step": 600
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.314204314204314e-05,
      "loss": 0.4743,
      "step": 610
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.293854293854294e-05,
      "loss": 0.4935,
      "step": 620
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.2735042735042735e-05,
      "loss": 0.5038,
      "step": 630
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.2531542531542535e-05,
      "loss": 0.4891,
      "step": 640
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.232804232804233e-05,
      "loss": 0.4793,
      "step": 650
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.212454212454213e-05,
      "loss": 0.4392,
      "step": 660
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.192104192104192e-05,
      "loss": 0.5051,
      "step": 670
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.171754171754172e-05,
      "loss": 0.4831,
      "step": 680
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.1514041514041515e-05,
      "loss": 0.5037,
      "step": 690
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.131054131054131e-05,
      "loss": 0.4316,
      "step": 700
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.110704110704111e-05,
      "loss": 0.4364,
      "step": 710
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.09035409035409e-05,
      "loss": 0.4346,
      "step": 720
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.07000407000407e-05,
      "loss": 0.4619,
      "step": 730
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.0496540496540496e-05,
      "loss": 0.4526,
      "step": 740
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.0293040293040296e-05,
      "loss": 0.4701,
      "step": 750
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.008954008954009e-05,
      "loss": 0.457,
      "step": 760
    },
    {
      "epoch": 1.41,
      "learning_rate": 3.988603988603989e-05,
      "loss": 0.4425,
      "step": 770
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.968253968253968e-05,
      "loss": 0.4597,
      "step": 780
    },
    {
      "epoch": 1.45,
      "learning_rate": 3.947903947903948e-05,
      "loss": 0.4755,
      "step": 790
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.9275539275539276e-05,
      "loss": 0.4405,
      "step": 800
    },
    {
      "epoch": 1.48,
      "learning_rate": 3.9072039072039076e-05,
      "loss": 0.414,
      "step": 810
    },
    {
      "epoch": 1.5,
      "learning_rate": 3.8868538868538876e-05,
      "loss": 0.4158,
      "step": 820
    },
    {
      "epoch": 1.52,
      "learning_rate": 3.866503866503867e-05,
      "loss": 0.4228,
      "step": 830
    },
    {
      "epoch": 1.54,
      "learning_rate": 3.846153846153846e-05,
      "loss": 0.4061,
      "step": 840
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.8258038258038256e-05,
      "loss": 0.3796,
      "step": 850
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.8054538054538056e-05,
      "loss": 0.4045,
      "step": 860
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.785103785103785e-05,
      "loss": 0.4146,
      "step": 870
    },
    {
      "epoch": 1.61,
      "learning_rate": 3.764753764753765e-05,
      "loss": 0.4309,
      "step": 880
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.744403744403744e-05,
      "loss": 0.4049,
      "step": 890
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.724053724053724e-05,
      "loss": 0.4353,
      "step": 900
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.7037037037037037e-05,
      "loss": 0.4114,
      "step": 910
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.683353683353684e-05,
      "loss": 0.3968,
      "step": 920
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.663003663003663e-05,
      "loss": 0.382,
      "step": 930
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.642653642653643e-05,
      "loss": 0.3718,
      "step": 940
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.622303622303623e-05,
      "loss": 0.4103,
      "step": 950
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.6019536019536024e-05,
      "loss": 0.4087,
      "step": 960
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.581603581603582e-05,
      "loss": 0.4003,
      "step": 970
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.561253561253561e-05,
      "loss": 0.3942,
      "step": 980
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.540903540903541e-05,
      "loss": 0.3723,
      "step": 990
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.5205535205535204e-05,
      "loss": 0.4053,
      "step": 1000
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.5002035002035004e-05,
      "loss": 0.3905,
      "step": 1010
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.47985347985348e-05,
      "loss": 0.4082,
      "step": 1020
    },
    {
      "epoch": 1.89,
      "learning_rate": 3.45950345950346e-05,
      "loss": 0.4056,
      "step": 1030
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.439153439153439e-05,
      "loss": 0.3755,
      "step": 1040
    },
    {
      "epoch": 1.92,
      "learning_rate": 3.418803418803419e-05,
      "loss": 0.3843,
      "step": 1050
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.3984533984533984e-05,
      "loss": 0.3725,
      "step": 1060
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.3781033781033784e-05,
      "loss": 0.3775,
      "step": 1070
    },
    {
      "epoch": 1.98,
      "learning_rate": 3.357753357753358e-05,
      "loss": 0.3923,
      "step": 1080
    },
    {
      "epoch": 1.99,
      "learning_rate": 3.337403337403338e-05,
      "loss": 0.3697,
      "step": 1090
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9235565632844764,
      "eval_loss": 0.20286566019058228,
      "eval_runtime": 225.6548,
      "eval_samples_per_second": 54.725,
      "eval_steps_per_second": 1.711,
      "step": 1092
    },
    {
      "epoch": 2.01,
      "learning_rate": 3.317053317053318e-05,
      "loss": 0.3883,
      "step": 1100
    },
    {
      "epoch": 2.03,
      "learning_rate": 3.296703296703297e-05,
      "loss": 0.3725,
      "step": 1110
    },
    {
      "epoch": 2.05,
      "learning_rate": 3.2763532763532764e-05,
      "loss": 0.3655,
      "step": 1120
    },
    {
      "epoch": 2.07,
      "learning_rate": 3.256003256003256e-05,
      "loss": 0.3505,
      "step": 1130
    },
    {
      "epoch": 2.09,
      "learning_rate": 3.235653235653236e-05,
      "loss": 0.3157,
      "step": 1140
    },
    {
      "epoch": 2.11,
      "learning_rate": 3.215303215303215e-05,
      "loss": 0.3418,
      "step": 1150
    },
    {
      "epoch": 2.12,
      "learning_rate": 3.194953194953195e-05,
      "loss": 0.3528,
      "step": 1160
    },
    {
      "epoch": 2.14,
      "learning_rate": 3.1746031746031745e-05,
      "loss": 0.383,
      "step": 1170
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.1542531542531545e-05,
      "loss": 0.3488,
      "step": 1180
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.133903133903134e-05,
      "loss": 0.3756,
      "step": 1190
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.113553113553114e-05,
      "loss": 0.3616,
      "step": 1200
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.093203093203093e-05,
      "loss": 0.3339,
      "step": 1210
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.072853072853073e-05,
      "loss": 0.3781,
      "step": 1220
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.052503052503053e-05,
      "loss": 0.3685,
      "step": 1230
    },
    {
      "epoch": 2.27,
      "learning_rate": 3.0321530321530322e-05,
      "loss": 0.332,
      "step": 1240
    },
    {
      "epoch": 2.29,
      "learning_rate": 3.0118030118030122e-05,
      "loss": 0.3916,
      "step": 1250
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.9914529914529915e-05,
      "loss": 0.3249,
      "step": 1260
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.9711029711029715e-05,
      "loss": 0.3511,
      "step": 1270
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.950752950752951e-05,
      "loss": 0.3667,
      "step": 1280
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.9304029304029305e-05,
      "loss": 0.3453,
      "step": 1290
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.91005291005291e-05,
      "loss": 0.3488,
      "step": 1300
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.88970288970289e-05,
      "loss": 0.3281,
      "step": 1310
    },
    {
      "epoch": 2.42,
      "learning_rate": 2.8693528693528692e-05,
      "loss": 0.3332,
      "step": 1320
    },
    {
      "epoch": 2.44,
      "learning_rate": 2.8490028490028492e-05,
      "loss": 0.3452,
      "step": 1330
    },
    {
      "epoch": 2.45,
      "learning_rate": 2.8286528286528286e-05,
      "loss": 0.3349,
      "step": 1340
    },
    {
      "epoch": 2.47,
      "learning_rate": 2.8083028083028086e-05,
      "loss": 0.3503,
      "step": 1350
    },
    {
      "epoch": 2.49,
      "learning_rate": 2.7879527879527882e-05,
      "loss": 0.3475,
      "step": 1360
    },
    {
      "epoch": 2.51,
      "learning_rate": 2.7676027676027676e-05,
      "loss": 0.3137,
      "step": 1370
    },
    {
      "epoch": 2.53,
      "learning_rate": 2.7472527472527476e-05,
      "loss": 0.2904,
      "step": 1380
    },
    {
      "epoch": 2.55,
      "learning_rate": 2.726902726902727e-05,
      "loss": 0.324,
      "step": 1390
    },
    {
      "epoch": 2.56,
      "learning_rate": 2.706552706552707e-05,
      "loss": 0.3297,
      "step": 1400
    },
    {
      "epoch": 2.58,
      "learning_rate": 2.6862026862026863e-05,
      "loss": 0.3152,
      "step": 1410
    },
    {
      "epoch": 2.6,
      "learning_rate": 2.6658526658526663e-05,
      "loss": 0.3395,
      "step": 1420
    },
    {
      "epoch": 2.62,
      "learning_rate": 2.6455026455026456e-05,
      "loss": 0.2879,
      "step": 1430
    },
    {
      "epoch": 2.64,
      "learning_rate": 2.6251526251526253e-05,
      "loss": 0.3319,
      "step": 1440
    },
    {
      "epoch": 2.65,
      "learning_rate": 2.6048026048026046e-05,
      "loss": 0.3046,
      "step": 1450
    },
    {
      "epoch": 2.67,
      "learning_rate": 2.5844525844525846e-05,
      "loss": 0.288,
      "step": 1460
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.564102564102564e-05,
      "loss": 0.3308,
      "step": 1470
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.543752543752544e-05,
      "loss": 0.343,
      "step": 1480
    },
    {
      "epoch": 2.73,
      "learning_rate": 2.5234025234025233e-05,
      "loss": 0.3114,
      "step": 1490
    },
    {
      "epoch": 2.75,
      "learning_rate": 2.5030525030525033e-05,
      "loss": 0.3302,
      "step": 1500
    },
    {
      "epoch": 2.76,
      "learning_rate": 2.4827024827024827e-05,
      "loss": 0.306,
      "step": 1510
    },
    {
      "epoch": 2.78,
      "learning_rate": 2.4623524623524623e-05,
      "loss": 0.3276,
      "step": 1520
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.442002442002442e-05,
      "loss": 0.3137,
      "step": 1530
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.4216524216524217e-05,
      "loss": 0.3048,
      "step": 1540
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.4013024013024013e-05,
      "loss": 0.2844,
      "step": 1550
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.380952380952381e-05,
      "loss": 0.3201,
      "step": 1560
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.3606023606023607e-05,
      "loss": 0.316,
      "step": 1570
    },
    {
      "epoch": 2.89,
      "learning_rate": 2.3402523402523404e-05,
      "loss": 0.3079,
      "step": 1580
    },
    {
      "epoch": 2.91,
      "learning_rate": 2.31990231990232e-05,
      "loss": 0.3029,
      "step": 1590
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.2995522995522997e-05,
      "loss": 0.3049,
      "step": 1600
    },
    {
      "epoch": 2.95,
      "learning_rate": 2.2792022792022794e-05,
      "loss": 0.3094,
      "step": 1610
    },
    {
      "epoch": 2.97,
      "learning_rate": 2.258852258852259e-05,
      "loss": 0.3074,
      "step": 1620
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.2385022385022387e-05,
      "loss": 0.3195,
      "step": 1630
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9380516641023565,
      "eval_loss": 0.15653181076049805,
      "eval_runtime": 218.5268,
      "eval_samples_per_second": 56.51,
      "eval_steps_per_second": 1.766,
      "step": 1638
    },
    {
      "epoch": 3.0,
      "learning_rate": 2.2181522181522184e-05,
      "loss": 0.3411,
      "step": 1640
    },
    {
      "epoch": 3.02,
      "learning_rate": 2.1978021978021977e-05,
      "loss": 0.2646,
      "step": 1650
    },
    {
      "epoch": 3.04,
      "learning_rate": 2.1774521774521774e-05,
      "loss": 0.2834,
      "step": 1660
    },
    {
      "epoch": 3.06,
      "learning_rate": 2.157102157102157e-05,
      "loss": 0.2824,
      "step": 1670
    },
    {
      "epoch": 3.08,
      "learning_rate": 2.1367521367521368e-05,
      "loss": 0.2962,
      "step": 1680
    },
    {
      "epoch": 3.1,
      "learning_rate": 2.1164021164021164e-05,
      "loss": 0.2941,
      "step": 1690
    },
    {
      "epoch": 3.11,
      "learning_rate": 2.096052096052096e-05,
      "loss": 0.2803,
      "step": 1700
    },
    {
      "epoch": 3.13,
      "learning_rate": 2.0757020757020758e-05,
      "loss": 0.2885,
      "step": 1710
    },
    {
      "epoch": 3.15,
      "learning_rate": 2.0553520553520554e-05,
      "loss": 0.2868,
      "step": 1720
    },
    {
      "epoch": 3.17,
      "learning_rate": 2.035002035002035e-05,
      "loss": 0.259,
      "step": 1730
    },
    {
      "epoch": 3.19,
      "learning_rate": 2.0146520146520148e-05,
      "loss": 0.3245,
      "step": 1740
    },
    {
      "epoch": 3.2,
      "learning_rate": 1.9943019943019945e-05,
      "loss": 0.2933,
      "step": 1750
    },
    {
      "epoch": 3.22,
      "learning_rate": 1.973951973951974e-05,
      "loss": 0.2959,
      "step": 1760
    },
    {
      "epoch": 3.24,
      "learning_rate": 1.9536019536019538e-05,
      "loss": 0.2795,
      "step": 1770
    },
    {
      "epoch": 3.26,
      "learning_rate": 1.9332519332519335e-05,
      "loss": 0.2735,
      "step": 1780
    },
    {
      "epoch": 3.28,
      "learning_rate": 1.9129019129019128e-05,
      "loss": 0.2925,
      "step": 1790
    },
    {
      "epoch": 3.3,
      "learning_rate": 1.8925518925518925e-05,
      "loss": 0.2705,
      "step": 1800
    },
    {
      "epoch": 3.31,
      "learning_rate": 1.872201872201872e-05,
      "loss": 0.2688,
      "step": 1810
    },
    {
      "epoch": 3.33,
      "learning_rate": 1.8518518518518518e-05,
      "loss": 0.2797,
      "step": 1820
    },
    {
      "epoch": 3.35,
      "learning_rate": 1.8315018315018315e-05,
      "loss": 0.2547,
      "step": 1830
    },
    {
      "epoch": 3.37,
      "learning_rate": 1.8111518111518115e-05,
      "loss": 0.2758,
      "step": 1840
    },
    {
      "epoch": 3.39,
      "learning_rate": 1.790801790801791e-05,
      "loss": 0.277,
      "step": 1850
    },
    {
      "epoch": 3.41,
      "learning_rate": 1.7704517704517705e-05,
      "loss": 0.2822,
      "step": 1860
    },
    {
      "epoch": 3.42,
      "learning_rate": 1.7501017501017502e-05,
      "loss": 0.235,
      "step": 1870
    },
    {
      "epoch": 3.44,
      "learning_rate": 1.72975172975173e-05,
      "loss": 0.264,
      "step": 1880
    },
    {
      "epoch": 3.46,
      "learning_rate": 1.7094017094017095e-05,
      "loss": 0.308,
      "step": 1890
    },
    {
      "epoch": 3.48,
      "learning_rate": 1.6890516890516892e-05,
      "loss": 0.2992,
      "step": 1900
    },
    {
      "epoch": 3.5,
      "learning_rate": 1.668701668701669e-05,
      "loss": 0.2528,
      "step": 1910
    },
    {
      "epoch": 3.52,
      "learning_rate": 1.6483516483516486e-05,
      "loss": 0.2778,
      "step": 1920
    },
    {
      "epoch": 3.53,
      "learning_rate": 1.628001628001628e-05,
      "loss": 0.2729,
      "step": 1930
    },
    {
      "epoch": 3.55,
      "learning_rate": 1.6076516076516076e-05,
      "loss": 0.2884,
      "step": 1940
    },
    {
      "epoch": 3.57,
      "learning_rate": 1.5873015873015872e-05,
      "loss": 0.269,
      "step": 1950
    },
    {
      "epoch": 3.59,
      "learning_rate": 1.566951566951567e-05,
      "loss": 0.2652,
      "step": 1960
    },
    {
      "epoch": 3.61,
      "learning_rate": 1.5466015466015466e-05,
      "loss": 0.2576,
      "step": 1970
    },
    {
      "epoch": 3.63,
      "learning_rate": 1.5262515262515266e-05,
      "loss": 0.261,
      "step": 1980
    },
    {
      "epoch": 3.64,
      "learning_rate": 1.5059015059015061e-05,
      "loss": 0.2784,
      "step": 1990
    },
    {
      "epoch": 3.66,
      "learning_rate": 1.4855514855514858e-05,
      "loss": 0.2698,
      "step": 2000
    },
    {
      "epoch": 3.68,
      "learning_rate": 1.4652014652014653e-05,
      "loss": 0.269,
      "step": 2010
    },
    {
      "epoch": 3.7,
      "learning_rate": 1.444851444851445e-05,
      "loss": 0.2805,
      "step": 2020
    },
    {
      "epoch": 3.72,
      "learning_rate": 1.4245014245014246e-05,
      "loss": 0.2754,
      "step": 2030
    },
    {
      "epoch": 3.74,
      "learning_rate": 1.4041514041514043e-05,
      "loss": 0.2834,
      "step": 2040
    },
    {
      "epoch": 3.75,
      "learning_rate": 1.3838013838013838e-05,
      "loss": 0.2787,
      "step": 2050
    },
    {
      "epoch": 3.77,
      "learning_rate": 1.3634513634513635e-05,
      "loss": 0.2538,
      "step": 2060
    },
    {
      "epoch": 3.79,
      "learning_rate": 1.3431013431013431e-05,
      "loss": 0.2758,
      "step": 2070
    },
    {
      "epoch": 3.81,
      "learning_rate": 1.3227513227513228e-05,
      "loss": 0.2772,
      "step": 2080
    },
    {
      "epoch": 3.83,
      "learning_rate": 1.3024013024013023e-05,
      "loss": 0.2552,
      "step": 2090
    },
    {
      "epoch": 3.84,
      "learning_rate": 1.282051282051282e-05,
      "loss": 0.2623,
      "step": 2100
    },
    {
      "epoch": 3.86,
      "learning_rate": 1.2617012617012617e-05,
      "loss": 0.2621,
      "step": 2110
    },
    {
      "epoch": 3.88,
      "learning_rate": 1.2413512413512413e-05,
      "loss": 0.263,
      "step": 2120
    },
    {
      "epoch": 3.9,
      "learning_rate": 1.221001221001221e-05,
      "loss": 0.2612,
      "step": 2130
    },
    {
      "epoch": 3.92,
      "learning_rate": 1.2006512006512007e-05,
      "loss": 0.27,
      "step": 2140
    },
    {
      "epoch": 3.94,
      "learning_rate": 1.1803011803011803e-05,
      "loss": 0.2289,
      "step": 2150
    },
    {
      "epoch": 3.95,
      "learning_rate": 1.15995115995116e-05,
      "loss": 0.2482,
      "step": 2160
    },
    {
      "epoch": 3.97,
      "learning_rate": 1.1396011396011397e-05,
      "loss": 0.2461,
      "step": 2170
    },
    {
      "epoch": 3.99,
      "learning_rate": 1.1192511192511194e-05,
      "loss": 0.2392,
      "step": 2180
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9516560045347802,
      "eval_loss": 0.12513846158981323,
      "eval_runtime": 206.4321,
      "eval_samples_per_second": 59.821,
      "eval_steps_per_second": 1.87,
      "step": 2184
    },
    {
      "epoch": 4.01,
      "learning_rate": 1.0989010989010989e-05,
      "loss": 0.3083,
      "step": 2190
    },
    {
      "epoch": 4.03,
      "learning_rate": 1.0785510785510785e-05,
      "loss": 0.2342,
      "step": 2200
    },
    {
      "epoch": 4.05,
      "learning_rate": 1.0582010582010582e-05,
      "loss": 0.2554,
      "step": 2210
    },
    {
      "epoch": 4.07,
      "learning_rate": 1.0378510378510379e-05,
      "loss": 0.2419,
      "step": 2220
    },
    {
      "epoch": 4.08,
      "learning_rate": 1.0175010175010176e-05,
      "loss": 0.2564,
      "step": 2230
    },
    {
      "epoch": 4.1,
      "learning_rate": 9.971509971509972e-06,
      "loss": 0.2548,
      "step": 2240
    },
    {
      "epoch": 4.12,
      "learning_rate": 9.768009768009769e-06,
      "loss": 0.2622,
      "step": 2250
    },
    {
      "epoch": 4.14,
      "learning_rate": 9.564509564509564e-06,
      "loss": 0.2294,
      "step": 2260
    },
    {
      "epoch": 4.16,
      "learning_rate": 9.36100936100936e-06,
      "loss": 0.2471,
      "step": 2270
    },
    {
      "epoch": 4.18,
      "learning_rate": 9.157509157509158e-06,
      "loss": 0.244,
      "step": 2280
    },
    {
      "epoch": 4.19,
      "learning_rate": 8.954008954008954e-06,
      "loss": 0.2529,
      "step": 2290
    },
    {
      "epoch": 4.21,
      "learning_rate": 8.750508750508751e-06,
      "loss": 0.2337,
      "step": 2300
    },
    {
      "epoch": 4.23,
      "learning_rate": 8.547008547008548e-06,
      "loss": 0.2255,
      "step": 2310
    },
    {
      "epoch": 4.25,
      "learning_rate": 8.343508343508344e-06,
      "loss": 0.2609,
      "step": 2320
    },
    {
      "epoch": 4.27,
      "learning_rate": 8.14000814000814e-06,
      "loss": 0.2577,
      "step": 2330
    },
    {
      "epoch": 4.29,
      "learning_rate": 7.936507936507936e-06,
      "loss": 0.254,
      "step": 2340
    },
    {
      "epoch": 4.3,
      "learning_rate": 7.733007733007733e-06,
      "loss": 0.2593,
      "step": 2350
    },
    {
      "epoch": 4.32,
      "learning_rate": 7.5295075295075305e-06,
      "loss": 0.2615,
      "step": 2360
    },
    {
      "epoch": 4.34,
      "learning_rate": 7.326007326007326e-06,
      "loss": 0.2452,
      "step": 2370
    },
    {
      "epoch": 4.36,
      "learning_rate": 7.122507122507123e-06,
      "loss": 0.2561,
      "step": 2380
    },
    {
      "epoch": 4.38,
      "learning_rate": 6.919006919006919e-06,
      "loss": 0.2304,
      "step": 2390
    },
    {
      "epoch": 4.4,
      "learning_rate": 6.715506715506716e-06,
      "loss": 0.2459,
      "step": 2400
    },
    {
      "epoch": 4.41,
      "learning_rate": 6.5120065120065115e-06,
      "loss": 0.2192,
      "step": 2410
    },
    {
      "epoch": 4.43,
      "learning_rate": 6.308506308506308e-06,
      "loss": 0.2282,
      "step": 2420
    },
    {
      "epoch": 4.45,
      "learning_rate": 6.105006105006105e-06,
      "loss": 0.2333,
      "step": 2430
    },
    {
      "epoch": 4.47,
      "learning_rate": 5.901505901505902e-06,
      "loss": 0.2499,
      "step": 2440
    },
    {
      "epoch": 4.49,
      "learning_rate": 5.6980056980056985e-06,
      "loss": 0.2374,
      "step": 2450
    },
    {
      "epoch": 4.5,
      "learning_rate": 5.494505494505494e-06,
      "loss": 0.2486,
      "step": 2460
    },
    {
      "epoch": 4.52,
      "learning_rate": 5.291005291005291e-06,
      "loss": 0.267,
      "step": 2470
    },
    {
      "epoch": 4.54,
      "learning_rate": 5.087505087505088e-06,
      "loss": 0.2624,
      "step": 2480
    },
    {
      "epoch": 4.56,
      "learning_rate": 4.8840048840048845e-06,
      "loss": 0.2389,
      "step": 2490
    },
    {
      "epoch": 4.58,
      "learning_rate": 4.68050468050468e-06,
      "loss": 0.2412,
      "step": 2500
    },
    {
      "epoch": 4.6,
      "learning_rate": 4.477004477004477e-06,
      "loss": 0.2394,
      "step": 2510
    },
    {
      "epoch": 4.61,
      "learning_rate": 4.273504273504274e-06,
      "loss": 0.2306,
      "step": 2520
    },
    {
      "epoch": 4.63,
      "learning_rate": 4.07000407000407e-06,
      "loss": 0.2339,
      "step": 2530
    },
    {
      "epoch": 4.65,
      "learning_rate": 3.8665038665038664e-06,
      "loss": 0.2555,
      "step": 2540
    },
    {
      "epoch": 4.67,
      "learning_rate": 3.663003663003663e-06,
      "loss": 0.2268,
      "step": 2550
    },
    {
      "epoch": 4.69,
      "learning_rate": 3.4595034595034595e-06,
      "loss": 0.2639,
      "step": 2560
    },
    {
      "epoch": 4.71,
      "learning_rate": 3.2560032560032558e-06,
      "loss": 0.2375,
      "step": 2570
    },
    {
      "epoch": 4.72,
      "learning_rate": 3.0525030525030525e-06,
      "loss": 0.2205,
      "step": 2580
    },
    {
      "epoch": 4.74,
      "learning_rate": 2.8490028490028492e-06,
      "loss": 0.2339,
      "step": 2590
    },
    {
      "epoch": 4.76,
      "learning_rate": 2.6455026455026455e-06,
      "loss": 0.2275,
      "step": 2600
    },
    {
      "epoch": 4.78,
      "learning_rate": 2.4420024420024423e-06,
      "loss": 0.249,
      "step": 2610
    },
    {
      "epoch": 4.8,
      "learning_rate": 2.2385022385022386e-06,
      "loss": 0.2265,
      "step": 2620
    },
    {
      "epoch": 4.82,
      "learning_rate": 2.035002035002035e-06,
      "loss": 0.2286,
      "step": 2630
    },
    {
      "epoch": 4.83,
      "learning_rate": 1.8315018315018316e-06,
      "loss": 0.225,
      "step": 2640
    },
    {
      "epoch": 4.85,
      "learning_rate": 1.6280016280016279e-06,
      "loss": 0.2364,
      "step": 2650
    },
    {
      "epoch": 4.87,
      "learning_rate": 1.4245014245014246e-06,
      "loss": 0.2364,
      "step": 2660
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.2210012210012211e-06,
      "loss": 0.245,
      "step": 2670
    },
    {
      "epoch": 4.91,
      "learning_rate": 1.0175010175010174e-06,
      "loss": 0.2308,
      "step": 2680
    },
    {
      "epoch": 4.93,
      "learning_rate": 8.140008140008139e-07,
      "loss": 0.227,
      "step": 2690
    },
    {
      "epoch": 4.94,
      "learning_rate": 6.105006105006106e-07,
      "loss": 0.2221,
      "step": 2700
    },
    {
      "epoch": 4.96,
      "learning_rate": 4.0700040700040697e-07,
      "loss": 0.2436,
      "step": 2710
    },
    {
      "epoch": 4.98,
      "learning_rate": 2.0350020350020349e-07,
      "loss": 0.2471,
      "step": 2720
    },
    {
      "epoch": 5.0,
      "learning_rate": 0.0,
      "loss": 0.2343,
      "step": 2730
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.9634788241962912,
      "eval_loss": 0.0957665666937828,
      "eval_runtime": 205.3993,
      "eval_samples_per_second": 60.122,
      "eval_steps_per_second": 1.879,
      "step": 2730
    },
    {
      "epoch": 5.0,
      "step": 2730,
      "total_flos": 1.1380361886475223e+19,
      "train_loss": 0.39669630824427904,
      "train_runtime": 11277.9253,
      "train_samples_per_second": 31.022,
      "train_steps_per_second": 0.242
    }
  ],
  "max_steps": 2730,
  "num_train_epochs": 5,
  "total_flos": 1.1380361886475223e+19,
  "trial_name": null,
  "trial_params": null
}