{
  "best_metric": 0.7513241767883301,
  "best_model_checkpoint": "videomae-base-finetuned-ucf101-nomixup/checkpoint-5500",
  "epoch": 36.91275167785235,
  "global_step": 5500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.07,
      "learning_rate": 6.7114093959731546e-06,
      "loss": 4.685,
      "step": 10
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.3422818791946309e-05,
      "loss": 4.6973,
      "step": 20
    },
    {
      "epoch": 0.2,
      "learning_rate": 2.013422818791946e-05,
      "loss": 4.6677,
      "step": 30
    },
    {
      "epoch": 0.27,
      "learning_rate": 2.6845637583892618e-05,
      "loss": 4.6252,
      "step": 40
    },
    {
      "epoch": 0.34,
      "learning_rate": 3.3557046979865775e-05,
      "loss": 4.6242,
      "step": 50
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.026845637583892e-05,
      "loss": 4.5581,
      "step": 60
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.697986577181208e-05,
      "loss": 4.5285,
      "step": 70
    },
    {
      "epoch": 0.54,
      "learning_rate": 5.3691275167785237e-05,
      "loss": 4.5188,
      "step": 80
    },
    {
      "epoch": 0.6,
      "learning_rate": 6.040268456375839e-05,
      "loss": 4.431,
      "step": 90
    },
    {
      "epoch": 0.67,
      "learning_rate": 6.711409395973155e-05,
      "loss": 4.3994,
      "step": 100
    },
    {
      "epoch": 0.74,
      "learning_rate": 7.38255033557047e-05,
      "loss": 4.3375,
      "step": 110
    },
    {
      "epoch": 0.81,
      "learning_rate": 8.053691275167784e-05,
      "loss": 4.2659,
      "step": 120
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.7248322147651e-05,
      "loss": 4.1503,
      "step": 130
    },
    {
      "epoch": 0.94,
      "learning_rate": 9.395973154362417e-05,
      "loss": 4.0464,
      "step": 140
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.00010067114093959731,
      "loss": 3.9801,
      "step": 150
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00010738255033557047,
      "loss": 3.9336,
      "step": 160
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.00011409395973154363,
      "loss": 3.8316,
      "step": 170
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.00012080536912751678,
      "loss": 3.7867,
      "step": 180
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00012751677852348994,
      "loss": 3.6977,
      "step": 190
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0001342281879194631,
      "loss": 3.5853,
      "step": 200
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00014093959731543624,
      "loss": 3.4901,
      "step": 210
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0001476510067114094,
      "loss": 3.4679,
      "step": 220
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.00015436241610738256,
      "loss": 3.4984,
      "step": 230
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.0001610738255033557,
      "loss": 3.3598,
      "step": 240
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.00016778523489932888,
      "loss": 3.2679,
      "step": 250
    },
    {
      "epoch": 1.74,
      "learning_rate": 0.000174496644295302,
      "loss": 3.2981,
      "step": 260
    },
    {
      "epoch": 1.81,
      "learning_rate": 0.00018120805369127517,
      "loss": 3.2664,
      "step": 270
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.00018791946308724833,
      "loss": 3.2929,
      "step": 280
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.00019463087248322146,
      "loss": 3.2039,
      "step": 290
    },
    {
      "epoch": 2.01,
      "learning_rate": 0.00020134228187919463,
      "loss": 3.1763,
      "step": 300
    },
    {
      "epoch": 2.08,
      "learning_rate": 0.00020805369127516779,
      "loss": 3.0554,
      "step": 310
    },
    {
      "epoch": 2.15,
      "learning_rate": 0.00021476510067114095,
      "loss": 2.9937,
      "step": 320
    },
    {
      "epoch": 2.21,
      "learning_rate": 0.0002214765100671141,
      "loss": 3.0402,
      "step": 330
    },
    {
      "epoch": 2.28,
      "learning_rate": 0.00022818791946308727,
      "loss": 3.0012,
      "step": 340
    },
    {
      "epoch": 2.35,
      "learning_rate": 0.0002348993288590604,
      "loss": 2.9892,
      "step": 350
    },
    {
      "epoch": 2.42,
      "learning_rate": 0.00024161073825503356,
      "loss": 3.0453,
      "step": 360
    },
    {
      "epoch": 2.48,
      "learning_rate": 0.0002483221476510067,
      "loss": 2.9497,
      "step": 370
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.0002550335570469799,
      "loss": 2.8924,
      "step": 380
    },
    {
      "epoch": 2.62,
      "learning_rate": 0.000261744966442953,
      "loss": 2.9597,
      "step": 390
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.0002684563758389262,
      "loss": 3.0828,
      "step": 400
    },
    {
      "epoch": 2.75,
      "learning_rate": 0.00027516778523489934,
      "loss": 3.0353,
      "step": 410
    },
    {
      "epoch": 2.82,
      "learning_rate": 0.00028187919463087247,
      "loss": 2.8786,
      "step": 420
    },
    {
      "epoch": 2.89,
      "learning_rate": 0.00028859060402684566,
      "loss": 2.9322,
      "step": 430
    },
    {
      "epoch": 2.95,
      "learning_rate": 0.0002953020134228188,
      "loss": 2.7467,
      "step": 440
    },
    {
      "epoch": 3.02,
      "learning_rate": 0.000302013422818792,
      "loss": 2.8574,
      "step": 450
    },
    {
      "epoch": 3.09,
      "learning_rate": 0.0003087248322147651,
      "loss": 2.7929,
      "step": 460
    },
    {
      "epoch": 3.15,
      "learning_rate": 0.00031543624161073825,
      "loss": 2.8987,
      "step": 470
    },
    {
      "epoch": 3.22,
      "learning_rate": 0.0003221476510067114,
      "loss": 2.8908,
      "step": 480
    },
    {
      "epoch": 3.29,
      "learning_rate": 0.0003288590604026846,
      "loss": 2.6509,
      "step": 490
    },
    {
      "epoch": 3.36,
      "learning_rate": 0.00033557046979865775,
      "loss": 2.8576,
      "step": 500
    },
    {
      "epoch": 3.36,
      "eval_accuracy": 0.5911017060279846,
      "eval_accuracy_top5": 0.727224588394165,
      "eval_loss": 1.8057622909545898,
      "eval_runtime": 104.4766,
      "eval_samples_per_second": 36.209,
      "eval_steps_per_second": 4.527,
      "step": 500
    },
    {
      "epoch": 3.42,
      "learning_rate": 0.0003422818791946309,
      "loss": 2.7528,
      "step": 510
    },
    {
      "epoch": 3.49,
      "learning_rate": 0.000348993288590604,
      "loss": 2.7427,
      "step": 520
    },
    {
      "epoch": 3.56,
      "learning_rate": 0.00035570469798657715,
      "loss": 2.8489,
      "step": 530
    },
    {
      "epoch": 3.62,
      "learning_rate": 0.00036241610738255034,
      "loss": 2.8167,
      "step": 540
    },
    {
      "epoch": 3.69,
      "learning_rate": 0.00036912751677852353,
      "loss": 2.7375,
      "step": 550
    },
    {
      "epoch": 3.76,
      "learning_rate": 0.00037583892617449666,
      "loss": 2.8108,
      "step": 560
    },
    {
      "epoch": 3.83,
      "learning_rate": 0.0003825503355704698,
      "loss": 2.8261,
      "step": 570
    },
    {
      "epoch": 3.89,
      "learning_rate": 0.00038926174496644293,
      "loss": 2.7504,
      "step": 580
    },
    {
      "epoch": 3.96,
      "learning_rate": 0.0003959731543624161,
      "loss": 2.802,
      "step": 590
    },
    {
      "epoch": 4.03,
      "learning_rate": 0.00040268456375838925,
      "loss": 2.7507,
      "step": 600
    },
    {
      "epoch": 4.09,
      "learning_rate": 0.00040939597315436244,
      "loss": 2.8335,
      "step": 610
    },
    {
      "epoch": 4.16,
      "learning_rate": 0.00041610738255033557,
      "loss": 2.6994,
      "step": 620
    },
    {
      "epoch": 4.23,
      "learning_rate": 0.00042281879194630876,
      "loss": 2.8312,
      "step": 630
    },
    {
      "epoch": 4.3,
      "learning_rate": 0.0004295302013422819,
      "loss": 2.8237,
      "step": 640
    },
    {
      "epoch": 4.36,
      "learning_rate": 0.000436241610738255,
      "loss": 2.849,
      "step": 650
    },
    {
      "epoch": 4.43,
      "learning_rate": 0.0004429530201342282,
      "loss": 2.8233,
      "step": 660
    },
    {
      "epoch": 4.5,
      "learning_rate": 0.00044966442953020135,
      "loss": 2.7256,
      "step": 670
    },
    {
      "epoch": 4.56,
      "learning_rate": 0.00045637583892617453,
      "loss": 2.8312,
      "step": 680
    },
    {
      "epoch": 4.63,
      "learning_rate": 0.00046308724832214767,
      "loss": 2.7241,
      "step": 690
    },
    {
      "epoch": 4.7,
      "learning_rate": 0.0004697986577181208,
      "loss": 2.6984,
      "step": 700
    },
    {
      "epoch": 4.77,
      "learning_rate": 0.00047651006711409394,
      "loss": 2.7826,
      "step": 710
    },
    {
      "epoch": 4.83,
      "learning_rate": 0.0004832214765100671,
      "loss": 2.781,
      "step": 720
    },
    {
      "epoch": 4.9,
      "learning_rate": 0.0004899328859060403,
      "loss": 2.7601,
      "step": 730
    },
    {
      "epoch": 4.97,
      "learning_rate": 0.0004966442953020134,
      "loss": 2.744,
      "step": 740
    },
    {
      "epoch": 5.03,
      "learning_rate": 0.0004999988659275162,
      "loss": 2.8382,
      "step": 750
    },
    {
      "epoch": 5.1,
      "learning_rate": 0.0004999897934093798,
      "loss": 2.8987,
      "step": 760
    },
    {
      "epoch": 5.17,
      "learning_rate": 0.0004999716487023506,
      "loss": 2.7493,
      "step": 770
    },
    {
      "epoch": 5.23,
      "learning_rate": 0.0004999444324649045,
      "loss": 2.8423,
      "step": 780
    },
    {
      "epoch": 5.3,
      "learning_rate": 0.0004999081456847252,
      "loss": 2.8108,
      "step": 790
    },
    {
      "epoch": 5.37,
      "learning_rate": 0.0004998627896786686,
      "loss": 2.7675,
      "step": 800
    },
    {
      "epoch": 5.44,
      "learning_rate": 0.0004998083660927148,
      "loss": 2.5845,
      "step": 810
    },
    {
      "epoch": 5.5,
      "learning_rate": 0.0004997448769019087,
      "loss": 2.7445,
      "step": 820
    },
    {
      "epoch": 5.57,
      "learning_rate": 0.0004996723244102881,
      "loss": 2.6608,
      "step": 830
    },
    {
      "epoch": 5.64,
      "learning_rate": 0.0004995907112508002,
      "loss": 2.6423,
      "step": 840
    },
    {
      "epoch": 5.7,
      "learning_rate": 0.0004995000403852056,
      "loss": 2.706,
      "step": 850
    },
    {
      "epoch": 5.77,
      "learning_rate": 0.0004994003151039719,
      "loss": 2.7687,
      "step": 860
    },
    {
      "epoch": 5.84,
      "learning_rate": 0.000499291539026153,
      "loss": 2.761,
      "step": 870
    },
    {
      "epoch": 5.91,
      "learning_rate": 0.0004991737160992588,
      "loss": 2.6858,
      "step": 880
    },
    {
      "epoch": 5.97,
      "learning_rate": 0.000499046850599111,
      "loss": 2.7172,
      "step": 890
    },
    {
      "epoch": 6.04,
      "learning_rate": 0.0004989109471296887,
      "loss": 2.8251,
      "step": 900
    },
    {
      "epoch": 6.11,
      "learning_rate": 0.000498766010622961,
      "loss": 2.6963,
      "step": 910
    },
    {
      "epoch": 6.17,
      "learning_rate": 0.0004986120463387084,
      "loss": 2.4729,
      "step": 920
    },
    {
      "epoch": 6.24,
      "learning_rate": 0.000498449059864331,
      "loss": 2.7372,
      "step": 930
    },
    {
      "epoch": 6.31,
      "learning_rate": 0.0004982770571146468,
      "loss": 2.8074,
      "step": 940
    },
    {
      "epoch": 6.38,
      "learning_rate": 0.0004980960443316762,
      "loss": 2.7279,
      "step": 950
    },
    {
      "epoch": 6.44,
      "learning_rate": 0.0004979060280844162,
      "loss": 2.6071,
      "step": 960
    },
    {
      "epoch": 6.51,
      "learning_rate": 0.0004977070152686013,
      "loss": 2.7364,
      "step": 970
    },
    {
      "epoch": 6.58,
      "learning_rate": 0.000497499013106454,
      "loss": 2.6095,
      "step": 980
    },
    {
      "epoch": 6.64,
      "learning_rate": 0.0004972820291464219,
      "loss": 2.574,
      "step": 990
    },
    {
      "epoch": 6.71,
      "learning_rate": 0.0004970560712629044,
      "loss": 2.6651,
      "step": 1000
    },
    {
      "epoch": 6.71,
      "eval_accuracy": 0.5632944703102112,
      "eval_accuracy_top5": 0.7002118825912476,
      "eval_loss": 1.7887717485427856,
      "eval_runtime": 1057.27,
      "eval_samples_per_second": 3.578,
      "eval_steps_per_second": 0.447,
      "step": 1000
    },
    {
      "epoch": 6.78,
      "learning_rate": 0.0004968211476559667,
      "loss": 2.7145,
      "step": 1010
    },
    {
      "epoch": 6.85,
      "learning_rate": 0.0004965772668510425,
      "loss": 2.6658,
      "step": 1020
    },
    {
      "epoch": 6.91,
      "learning_rate": 0.0004963244376986235,
      "loss": 2.7473,
      "step": 1030
    },
    {
      "epoch": 6.98,
      "learning_rate": 0.0004960626693739401,
      "loss": 2.8268,
      "step": 1040
    },
    {
      "epoch": 7.05,
      "learning_rate": 0.0004957919713766266,
      "loss": 2.5706,
      "step": 1050
    },
    {
      "epoch": 7.11,
      "learning_rate": 0.0004955123535303775,
      "loss": 2.645,
      "step": 1060
    },
    {
      "epoch": 7.18,
      "learning_rate": 0.0004952238259825911,
      "loss": 2.68,
      "step": 1070
    },
    {
      "epoch": 7.25,
      "learning_rate": 0.0004949263992040001,
      "loss": 2.6571,
      "step": 1080
    },
    {
      "epoch": 7.32,
      "learning_rate": 0.0004946200839882932,
      "loss": 2.4641,
      "step": 1090
    },
    {
      "epoch": 7.38,
      "learning_rate": 0.000494304891451722,
      "loss": 2.6196,
      "step": 1100
    },
    {
      "epoch": 7.45,
      "learning_rate": 0.0004939808330326985,
      "loss": 2.5998,
      "step": 1110
    },
    {
      "epoch": 7.52,
      "learning_rate": 0.0004936479204913795,
      "loss": 2.6489,
      "step": 1120
    },
    {
      "epoch": 7.58,
      "learning_rate": 0.0004933061659092401,
      "loss": 2.6613,
      "step": 1130
    },
    {
      "epoch": 7.65,
      "learning_rate": 0.0004929555816886352,
      "loss": 2.533,
      "step": 1140
    },
    {
      "epoch": 7.72,
      "learning_rate": 0.0004925961805523493,
      "loss": 2.6853,
      "step": 1150
    },
    {
      "epoch": 7.79,
      "learning_rate": 0.0004922279755431349,
      "loss": 2.6887,
      "step": 1160
    },
    {
      "epoch": 7.85,
      "learning_rate": 0.0004918509800232392,
      "loss": 2.6397,
      "step": 1170
    },
    {
      "epoch": 7.92,
      "learning_rate": 0.0004914652076739187,
      "loss": 2.4045,
      "step": 1180
    },
    {
      "epoch": 7.99,
      "learning_rate": 0.0004910706724949438,
      "loss": 2.5858,
      "step": 1190
    },
    {
      "epoch": 8.05,
      "learning_rate": 0.0004906673888040895,
      "loss": 2.5493,
      "step": 1200
    },
    {
      "epoch": 8.12,
      "learning_rate": 0.0004902553712366165,
      "loss": 2.5056,
      "step": 1210
    },
    {
      "epoch": 8.19,
      "learning_rate": 0.0004898346347447405,
      "loss": 2.6127,
      "step": 1220
    },
    {
      "epoch": 8.26,
      "learning_rate": 0.0004894051945970881,
      "loss": 2.519,
      "step": 1230
    },
    {
      "epoch": 8.32,
      "learning_rate": 0.0004889670663781443,
      "loss": 2.4318,
      "step": 1240
    },
    {
      "epoch": 8.39,
      "learning_rate": 0.0004885202659876861,
      "loss": 2.4674,
      "step": 1250
    },
    {
      "epoch": 8.46,
      "learning_rate": 0.00048806480964020554,
      "loss": 2.5532,
      "step": 1260
    },
    {
      "epoch": 8.52,
      "learning_rate": 0.0004876007138643216,
      "loss": 2.5892,
      "step": 1270
    },
    {
      "epoch": 8.59,
      "learning_rate": 0.00048712799550218,
      "loss": 2.5353,
      "step": 1280
    },
    {
      "epoch": 8.66,
      "learning_rate": 0.00048664667170884217,
      "loss": 2.4692,
      "step": 1290
    },
    {
      "epoch": 8.72,
      "learning_rate": 0.0004861567599516628,
      "loss": 2.5329,
      "step": 1300
    },
    {
      "epoch": 8.79,
      "learning_rate": 0.0004856582780096558,
      "loss": 2.5185,
      "step": 1310
    },
    {
      "epoch": 8.86,
      "learning_rate": 0.0004851512439728489,
      "loss": 2.509,
      "step": 1320
    },
    {
      "epoch": 8.93,
      "learning_rate": 0.0004846356762416276,
      "loss": 2.4121,
      "step": 1330
    },
    {
      "epoch": 8.99,
      "learning_rate": 0.00048411159352606734,
      "loss": 2.528,
      "step": 1340
    },
    {
      "epoch": 9.06,
      "learning_rate": 0.0004835790148452539,
      "loss": 2.57,
      "step": 1350
    },
    {
      "epoch": 9.13,
      "learning_rate": 0.000483037959526594,
      "loss": 2.3809,
      "step": 1360
    },
    {
      "epoch": 9.19,
      "learning_rate": 0.0004824884472051136,
      "loss": 2.4399,
      "step": 1370
    },
    {
      "epoch": 9.26,
      "learning_rate": 0.000481930497822745,
      "loss": 2.4777,
      "step": 1380
    },
    {
      "epoch": 9.33,
      "learning_rate": 0.00048136413162760375,
      "loss": 2.6037,
      "step": 1390
    },
    {
      "epoch": 9.4,
      "learning_rate": 0.00048078936917325335,
      "loss": 2.4826,
      "step": 1400
    },
    {
      "epoch": 9.46,
      "learning_rate": 0.0004802062313179595,
      "loss": 2.5202,
      "step": 1410
    },
    {
      "epoch": 9.53,
      "learning_rate": 0.0004796147392239334,
      "loss": 2.5156,
      "step": 1420
    },
    {
      "epoch": 9.6,
      "learning_rate": 0.00047901491435656333,
      "loss": 2.4923,
      "step": 1430
    },
    {
      "epoch": 9.66,
      "learning_rate": 0.000478406778483636,
      "loss": 2.4254,
      "step": 1440
    },
    {
      "epoch": 9.73,
      "learning_rate": 0.0004777903536745466,
      "loss": 2.5741,
      "step": 1450
    },
    {
      "epoch": 9.8,
      "learning_rate": 0.0004771656622994974,
      "loss": 2.5019,
      "step": 1460
    },
    {
      "epoch": 9.87,
      "learning_rate": 0.00047653272702868676,
      "loss": 2.5268,
      "step": 1470
    },
    {
      "epoch": 9.93,
      "learning_rate": 0.0004758915708314858,
      "loss": 2.4928,
      "step": 1480
    },
    {
      "epoch": 10.0,
      "learning_rate": 0.00047524221697560476,
      "loss": 2.5171,
      "step": 1490
    },
    {
      "epoch": 10.07,
      "learning_rate": 0.0004745846890262493,
      "loss": 2.3321,
      "step": 1500
    },
    {
      "epoch": 10.07,
      "eval_accuracy": 0.5985169410705566,
      "eval_accuracy_top5": 0.7327859997749329,
      "eval_loss": 1.7438325881958008,
      "eval_runtime": 103.1085,
      "eval_samples_per_second": 36.69,
      "eval_steps_per_second": 4.587,
      "step": 1500
    },
    {
      "epoch": 10.13,
      "learning_rate": 0.0004739190108452645,
      "loss": 2.4042,
      "step": 1510
    },
    {
      "epoch": 10.2,
      "learning_rate": 0.0004732452065902695,
      "loss": 2.4647,
      "step": 1520
    },
    {
      "epoch": 10.27,
      "learning_rate": 0.0004725633007137805,
      "loss": 2.4076,
      "step": 1530
    },
    {
      "epoch": 10.34,
      "learning_rate": 0.00047187331796232345,
      "loss": 2.2986,
      "step": 1540
    },
    {
      "epoch": 10.4,
      "learning_rate": 0.00047117528337553615,
      "loss": 2.5126,
      "step": 1550
    },
    {
      "epoch": 10.47,
      "learning_rate": 0.00047046922228525926,
      "loss": 2.4857,
      "step": 1560
    },
    {
      "epoch": 10.54,
      "learning_rate": 0.0004697551603146171,
      "loss": 2.4434,
      "step": 1570
    },
    {
      "epoch": 10.6,
      "learning_rate": 0.000469033123377088,
      "loss": 2.3844,
      "step": 1580
    },
    {
      "epoch": 10.67,
      "learning_rate": 0.00046830313767556365,
      "loss": 2.4232,
      "step": 1590
    },
    {
      "epoch": 10.74,
      "learning_rate": 0.0004675652297013984,
      "loss": 2.4638,
      "step": 1600
    },
    {
      "epoch": 10.81,
      "learning_rate": 0.0004668194262334475,
      "loss": 2.3976,
      "step": 1610
    },
    {
      "epoch": 10.87,
      "learning_rate": 0.0004660657543370958,
      "loss": 2.407,
      "step": 1620
    },
    {
      "epoch": 10.94,
      "learning_rate": 0.0004653042413632751,
      "loss": 2.42,
      "step": 1630
    },
    {
      "epoch": 11.01,
      "learning_rate": 0.0004645349149474718,
      "loss": 2.4782,
      "step": 1640
    },
    {
      "epoch": 11.07,
      "learning_rate": 0.00046375780300872405,
      "loss": 2.3456,
      "step": 1650
    },
    {
      "epoch": 11.14,
      "learning_rate": 0.00046297293374860846,
      "loss": 2.4317,
      "step": 1660
    },
    {
      "epoch": 11.21,
      "learning_rate": 0.00046218033565021647,
      "loss": 2.315,
      "step": 1670
    },
    {
      "epoch": 11.28,
      "learning_rate": 0.0004613800374771212,
      "loss": 2.3308,
      "step": 1680
    },
    {
      "epoch": 11.34,
      "learning_rate": 0.000460572068272333,
      "loss": 2.4006,
      "step": 1690
    },
    {
      "epoch": 11.41,
      "learning_rate": 0.0004597564573572461,
      "loss": 2.4491,
      "step": 1700
    },
    {
      "epoch": 11.48,
      "learning_rate": 0.000458933234330574,
      "loss": 2.4763,
      "step": 1710
    },
    {
      "epoch": 11.54,
      "learning_rate": 0.0004581024290672755,
      "loss": 2.4014,
      "step": 1720
    },
    {
      "epoch": 11.61,
      "learning_rate": 0.00045726407171747075,
      "loss": 2.5596,
      "step": 1730
    },
    {
      "epoch": 11.68,
      "learning_rate": 0.0004564181927053468,
      "loss": 2.4823,
      "step": 1740
    },
    {
      "epoch": 11.74,
      "learning_rate": 0.0004555648227280535,
      "loss": 2.4451,
      "step": 1750
    },
    {
      "epoch": 11.81,
      "learning_rate": 0.00045470399275458984,
      "loss": 2.4504,
      "step": 1760
    },
    {
      "epoch": 11.88,
      "learning_rate": 0.00045383573402467967,
      "loss": 2.3143,
      "step": 1770
    },
    {
      "epoch": 11.95,
      "learning_rate": 0.00045296007804763815,
      "loss": 2.2971,
      "step": 1780
    },
    {
      "epoch": 12.01,
      "learning_rate": 0.00045207705660122853,
      "loss": 2.2387,
      "step": 1790
    },
    {
      "epoch": 12.08,
      "learning_rate": 0.00045118670173050846,
      "loss": 2.386,
      "step": 1800
    },
    {
      "epoch": 12.15,
      "learning_rate": 0.0004502890457466673,
      "loss": 2.3928,
      "step": 1810
    },
    {
      "epoch": 12.21,
      "learning_rate": 0.00044938412122585373,
      "loss": 2.3766,
      "step": 1820
    },
    {
      "epoch": 12.28,
      "learning_rate": 0.00044847196100799306,
      "loss": 2.3229,
      "step": 1830
    },
    {
      "epoch": 12.35,
      "learning_rate": 0.000447552598195596,
      "loss": 2.1737,
      "step": 1840
    },
    {
      "epoch": 12.42,
      "learning_rate": 0.00044662606615255705,
      "loss": 2.3663,
      "step": 1850
    },
    {
      "epoch": 12.48,
      "learning_rate": 0.0004456923985029439,
      "loss": 2.3396,
      "step": 1860
    },
    {
      "epoch": 12.55,
      "learning_rate": 0.0004447516291297768,
      "loss": 2.5236,
      "step": 1870
    },
    {
      "epoch": 12.62,
      "learning_rate": 0.0004438037921737997,
      "loss": 2.242,
      "step": 1880
    },
    {
      "epoch": 12.68,
      "learning_rate": 0.00044284892203224024,
      "loss": 2.3837,
      "step": 1890
    },
    {
      "epoch": 12.75,
      "learning_rate": 0.00044188705335756253,
      "loss": 2.3464,
      "step": 1900
    },
    {
      "epoch": 12.82,
      "learning_rate": 0.0004409182210562087,
      "loss": 2.2084,
      "step": 1910
    },
    {
      "epoch": 12.89,
      "learning_rate": 0.00043994246028733296,
      "loss": 2.3124,
      "step": 1920
    },
    {
      "epoch": 12.95,
      "learning_rate": 0.0004389598064615249,
      "loss": 2.1847,
      "step": 1930
    },
    {
      "epoch": 13.02,
      "learning_rate": 0.0004379702952395249,
      "loss": 2.2371,
      "step": 1940
    },
    {
      "epoch": 13.09,
      "learning_rate": 0.00043697396253093,
      "loss": 2.383,
      "step": 1950
    },
    {
      "epoch": 13.15,
      "learning_rate": 0.0004359708444928904,
      "loss": 2.2143,
      "step": 1960
    },
    {
      "epoch": 13.22,
      "learning_rate": 0.0004349609775287977,
      "loss": 2.2602,
      "step": 1970
    },
    {
      "epoch": 13.29,
      "learning_rate": 0.0004339443982869634,
      "loss": 2.2835,
      "step": 1980
    },
    {
      "epoch": 13.36,
      "learning_rate": 0.00043292114365928934,
      "loss": 2.2528,
      "step": 1990
    },
    {
      "epoch": 13.42,
      "learning_rate": 0.00043189125077992867,
      "loss": 2.3348,
      "step": 2000
    },
    {
      "epoch": 13.42,
      "eval_accuracy": 0.6295021176338196,
      "eval_accuracy_top5": 0.7356991767883301,
      "eval_loss": 1.62944495677948,
      "eval_runtime": 99.1269,
      "eval_samples_per_second": 38.163,
      "eval_steps_per_second": 4.772,
      "step": 2000
    },
    {
      "epoch": 13.49,
      "learning_rate": 0.00043085475702393827,
      "loss": 2.3693,
      "step": 2010
    },
    {
      "epoch": 13.56,
      "learning_rate": 0.0004298117000059223,
      "loss": 2.2296,
      "step": 2020
    },
    {
      "epoch": 13.62,
      "learning_rate": 0.0004287621175786674,
      "loss": 2.2473,
      "step": 2030
    },
    {
      "epoch": 13.69,
      "learning_rate": 0.0004277060478317687,
      "loss": 2.2507,
      "step": 2040
    },
    {
      "epoch": 13.76,
      "learning_rate": 0.00042664352909024784,
      "loss": 2.3107,
      "step": 2050
    },
    {
      "epoch": 13.83,
      "learning_rate": 0.00042557459991316197,
      "loss": 2.3385,
      "step": 2060
    },
    {
      "epoch": 13.89,
      "learning_rate": 0.0004244992990922043,
      "loss": 2.2102,
      "step": 2070
    },
    {
      "epoch": 13.96,
      "learning_rate": 0.0004234176656502968,
      "loss": 2.3437,
      "step": 2080
    },
    {
      "epoch": 14.03,
      "learning_rate": 0.00042232973884017355,
      "loss": 2.3511,
      "step": 2090
    },
    {
      "epoch": 14.09,
      "learning_rate": 0.00042123555814295655,
      "loss": 2.3496,
      "step": 2100
    },
    {
      "epoch": 14.16,
      "learning_rate": 0.00042013516326672276,
      "loss": 2.2043,
      "step": 2110
    },
    {
      "epoch": 14.23,
      "learning_rate": 0.0004190285941450632,
      "loss": 2.2971,
      "step": 2120
    },
    {
      "epoch": 14.3,
      "learning_rate": 0.00041791589093563385,
      "loss": 2.2116,
      "step": 2130
    },
    {
      "epoch": 14.36,
      "learning_rate": 0.000416797094018698,
      "loss": 2.2991,
      "step": 2140
    },
    {
      "epoch": 14.43,
      "learning_rate": 0.00041567224399566104,
      "loss": 2.2796,
      "step": 2150
    },
    {
      "epoch": 14.5,
      "learning_rate": 0.00041454138168759726,
      "loss": 2.3402,
      "step": 2160
    },
    {
      "epoch": 14.56,
      "learning_rate": 0.00041340454813376784,
      "loss": 2.3073,
      "step": 2170
    },
    {
      "epoch": 14.63,
      "learning_rate": 0.0004122617845901322,
      "loss": 2.2733,
      "step": 2180
    },
    {
      "epoch": 14.7,
      "learning_rate": 0.0004111131325278502,
      "loss": 2.2013,
      "step": 2190
    },
    {
      "epoch": 14.77,
      "learning_rate": 0.0004099586336317779,
      "loss": 2.2275,
      "step": 2200
    },
    {
      "epoch": 14.83,
      "learning_rate": 0.00040879832979895376,
      "loss": 2.2008,
      "step": 2210
    },
    {
      "epoch": 14.9,
      "learning_rate": 0.00040763226313707924,
      "loss": 2.1913,
      "step": 2220
    },
    {
      "epoch": 14.97,
      "learning_rate": 0.00040646047596299,
      "loss": 2.2507,
      "step": 2230
    },
    {
      "epoch": 15.03,
      "learning_rate": 0.0004052830108011204,
      "loss": 2.138,
      "step": 2240
    },
    {
      "epoch": 15.1,
      "learning_rate": 0.00040409991038196057,
      "loss": 2.2183,
      "step": 2250
    },
    {
      "epoch": 15.17,
      "learning_rate": 0.00040291121764050533,
      "loss": 2.2204,
      "step": 2260
    },
    {
      "epoch": 15.23,
      "learning_rate": 0.000401716975714696,
      "loss": 2.1662,
      "step": 2270
    },
    {
      "epoch": 15.3,
      "learning_rate": 0.0004005172279438555,
      "loss": 2.3825,
      "step": 2280
    },
    {
      "epoch": 15.37,
      "learning_rate": 0.00039931201786711493,
      "loss": 2.2073,
      "step": 2290
    },
    {
      "epoch": 15.44,
      "learning_rate": 0.0003981013892218336,
      "loss": 2.1114,
      "step": 2300
    },
    {
      "epoch": 15.5,
      "learning_rate": 0.0003968853859420125,
      "loss": 2.2326,
      "step": 2310
    },
    {
      "epoch": 15.57,
      "learning_rate": 0.0003956640521566989,
      "loss": 2.2042,
      "step": 2320
    },
    {
      "epoch": 15.64,
      "learning_rate": 0.00039443743218838553,
      "loss": 2.1792,
      "step": 2330
    },
    {
      "epoch": 15.7,
      "learning_rate": 0.00039320557055140195,
      "loss": 2.1393,
      "step": 2340
    },
    {
      "epoch": 15.77,
      "learning_rate": 0.00039196851195029913,
      "loss": 2.1661,
      "step": 2350
    },
    {
      "epoch": 15.84,
      "learning_rate": 0.00039072630127822707,
      "loss": 2.2561,
      "step": 2360
    },
    {
      "epoch": 15.91,
      "learning_rate": 0.0003894789836153058,
      "loss": 2.1827,
      "step": 2370
    },
    {
      "epoch": 15.97,
      "learning_rate": 0.00038822660422698895,
      "loss": 2.2583,
      "step": 2380
    },
    {
      "epoch": 16.04,
      "learning_rate": 0.00038696920856242176,
      "loss": 2.1149,
      "step": 2390
    },
    {
      "epoch": 16.11,
      "learning_rate": 0.0003857068422527908,
      "loss": 2.1045,
      "step": 2400
    },
    {
      "epoch": 16.17,
      "learning_rate": 0.0003844395511096692,
      "loss": 2.2076,
      "step": 2410
    },
    {
      "epoch": 16.24,
      "learning_rate": 0.00038316738112335294,
      "loss": 2.2332,
      "step": 2420
    },
    {
      "epoch": 16.31,
      "learning_rate": 0.00038189037846119247,
      "loss": 2.1815,
      "step": 2430
    },
    {
      "epoch": 16.38,
      "learning_rate": 0.0003806085894659174,
      "loss": 2.139,
      "step": 2440
    },
    {
      "epoch": 16.44,
      "learning_rate": 0.00037932206065395427,
      "loss": 2.0076,
      "step": 2450
    },
    {
      "epoch": 16.51,
      "learning_rate": 0.0003780308387137387,
      "loss": 2.2418,
      "step": 2460
    },
    {
      "epoch": 16.58,
      "learning_rate": 0.00037673497050402143,
      "loss": 2.1668,
      "step": 2470
    },
    {
      "epoch": 16.64,
      "learning_rate": 0.0003754345030521669,
      "loss": 2.1749,
      "step": 2480
    },
    {
      "epoch": 16.71,
      "learning_rate": 0.0003741294835524475,
      "loss": 2.1023,
      "step": 2490
    },
    {
      "epoch": 16.78,
      "learning_rate": 0.00037281995936433026,
      "loss": 2.0315,
      "step": 2500
    },
    {
      "epoch": 16.78,
      "eval_accuracy": 0.6353284120559692,
      "eval_accuracy_top5": 0.7513241767883301,
      "eval_loss": 1.620951771736145,
      "eval_runtime": 106.1983,
      "eval_samples_per_second": 35.622,
      "eval_steps_per_second": 4.454,
      "step": 2500
    },
    {
      "epoch": 16.85,
      "learning_rate": 0.0003715059780107586,
      "loss": 2.1777,
      "step": 2510
    },
    {
      "epoch": 16.91,
      "learning_rate": 0.0003701875871764275,
      "loss": 2.0452,
      "step": 2520
    },
    {
      "epoch": 16.98,
      "learning_rate": 0.0003688648347060529,
      "loss": 2.2884,
      "step": 2530
    },
    {
      "epoch": 17.05,
      "learning_rate": 0.00036753776860263594,
      "loss": 2.1402,
      "step": 2540
    },
    {
      "epoch": 17.11,
      "learning_rate": 0.00036620643702571997,
      "loss": 2.0759,
      "step": 2550
    },
    {
      "epoch": 17.18,
      "learning_rate": 0.00036487088828964377,
      "loss": 2.0357,
      "step": 2560
    },
    {
      "epoch": 17.25,
      "learning_rate": 0.00036353117086178777,
      "loss": 2.0975,
      "step": 2570
    },
    {
      "epoch": 17.32,
      "learning_rate": 0.00036218733336081506,
      "loss": 2.2448,
      "step": 2580
    },
    {
      "epoch": 17.38,
      "learning_rate": 0.00036083942455490735,
      "loss": 2.1251,
      "step": 2590
    },
    {
      "epoch": 17.45,
      "learning_rate": 0.0003594874933599949,
      "loss": 2.1602,
      "step": 2600
    },
    {
      "epoch": 17.52,
      "learning_rate": 0.00035813158883798134,
      "loss": 2.1744,
      "step": 2610
    },
    {
      "epoch": 17.58,
      "learning_rate": 0.00035677176019496343,
      "loss": 2.1631,
      "step": 2620
    },
    {
      "epoch": 17.65,
      "learning_rate": 0.0003554080567794451,
      "loss": 2.1669,
      "step": 2630
    },
    {
      "epoch": 17.72,
      "learning_rate": 0.0003540405280805467,
      "loss": 2.1181,
      "step": 2640
    },
    {
      "epoch": 17.79,
      "learning_rate": 0.00035266922372620906,
      "loss": 2.0952,
      "step": 2650
    },
    {
      "epoch": 17.85,
      "learning_rate": 0.00035129419348139246,
      "loss": 2.1549,
      "step": 2660
    },
    {
      "epoch": 17.92,
      "learning_rate": 0.0003499154872462705,
      "loss": 2.1738,
      "step": 2670
    },
    {
      "epoch": 17.99,
      "learning_rate": 0.0003485331550544197,
      "loss": 2.1793,
      "step": 2680
    },
    {
      "epoch": 18.05,
      "learning_rate": 0.00034714724707100283,
      "loss": 2.1329,
      "step": 2690
    },
    {
      "epoch": 18.12,
      "learning_rate": 0.00034575781359094947,
      "loss": 2.0711,
      "step": 2700
    },
    {
      "epoch": 18.19,
      "learning_rate": 0.0003443649050371301,
      "loss": 2.0466,
      "step": 2710
    },
    {
      "epoch": 18.26,
      "learning_rate": 0.00034296857195852647,
      "loss": 2.1244,
      "step": 2720
    },
    {
      "epoch": 18.32,
      "learning_rate": 0.0003415688650283973,
      "loss": 2.1518,
      "step": 2730
    },
    {
      "epoch": 18.39,
      "learning_rate": 0.0003401658350424389,
      "loss": 2.1096,
      "step": 2740
    },
    {
      "epoch": 18.46,
      "learning_rate": 0.0003387595329169423,
      "loss": 2.0938,
      "step": 2750
    },
    {
      "epoch": 18.52,
      "learning_rate": 0.0003373500096869451,
      "loss": 2.1085,
      "step": 2760
    },
    {
      "epoch": 18.59,
      "learning_rate": 0.0003359373165043796,
      "loss": 2.0453,
      "step": 2770
    },
    {
      "epoch": 18.66,
      "learning_rate": 0.0003345215046362165,
      "loss": 2.2149,
      "step": 2780
    },
    {
      "epoch": 18.72,
      "learning_rate": 0.00033310262546260434,
      "loss": 2.1468,
      "step": 2790
    },
    {
      "epoch": 18.79,
      "learning_rate": 0.0003316807304750047,
      "loss": 2.0384,
      "step": 2800
    },
    {
      "epoch": 18.86,
      "learning_rate": 0.0003302558712743241,
      "loss": 2.0865,
      "step": 2810
    },
    {
      "epoch": 18.93,
      "learning_rate": 0.00032882809956904065,
      "loss": 2.1671,
      "step": 2820
    },
    {
      "epoch": 18.99,
      "learning_rate": 0.0003273974671733283,
      "loss": 2.0147,
      "step": 2830
    },
    {
      "epoch": 19.06,
      "learning_rate": 0.0003259640260051759,
      "loss": 1.9943,
      "step": 2840
    },
    {
      "epoch": 19.13,
      "learning_rate": 0.00032452782808450355,
      "loss": 1.9631,
      "step": 2850
    },
    {
      "epoch": 19.19,
      "learning_rate": 0.00032308892553127425,
      "loss": 2.1063,
      "step": 2860
    },
    {
      "epoch": 19.26,
      "learning_rate": 0.00032164737056360307,
      "loss": 1.9979,
      "step": 2870
    },
    {
      "epoch": 19.33,
      "learning_rate": 0.0003202032154958615,
      "loss": 1.9259,
      "step": 2880
    },
    {
      "epoch": 19.4,
      "learning_rate": 0.0003187565127367797,
      "loss": 1.9233,
      "step": 2890
    },
    {
      "epoch": 19.46,
      "learning_rate": 0.00031730731478754365,
      "loss": 2.0203,
      "step": 2900
    },
    {
      "epoch": 19.53,
      "learning_rate": 0.00031585567423989084,
      "loss": 2.0238,
      "step": 2910
    },
    {
      "epoch": 19.6,
      "learning_rate": 0.00031440164377420085,
      "loss": 2.0497,
      "step": 2920
    },
    {
      "epoch": 19.66,
      "learning_rate": 0.0003129452761575843,
      "loss": 1.9399,
      "step": 2930
    },
    {
      "epoch": 19.73,
      "learning_rate": 0.00031148662424196723,
      "loss": 2.0291,
      "step": 2940
    },
    {
      "epoch": 19.8,
      "learning_rate": 0.0003100257409621738,
      "loss": 2.0845,
      "step": 2950
    },
    {
      "epoch": 19.87,
      "learning_rate": 0.0003085626793340045,
      "loss": 1.9692,
      "step": 2960
    },
    {
      "epoch": 19.93,
      "learning_rate": 0.00030709749245231297,
      "loss": 2.0479,
      "step": 2970
    },
    {
      "epoch": 20.0,
      "learning_rate": 0.0003056302334890786,
      "loss": 2.0145,
      "step": 2980
    },
    {
      "epoch": 20.07,
      "learning_rate": 0.00030416095569147717,
      "loss": 1.9802,
      "step": 2990
    },
    {
      "epoch": 20.13,
      "learning_rate": 0.00030268971237994835,
      "loss": 2.0705,
      "step": 3000
    },
    {
      "epoch": 20.13,
      "eval_accuracy": 0.6464512944221497,
      "eval_accuracy_top5": 0.7627118825912476,
      "eval_loss": 1.5695319175720215,
      "eval_runtime": 101.6469,
      "eval_samples_per_second": 37.217,
      "eval_steps_per_second": 4.653,
      "step": 3000
    },
    {
      "epoch": 20.2,
      "learning_rate": 0.00030121655694626096,
      "loss": 1.9999,
      "step": 3010
    },
    {
      "epoch": 20.27,
      "learning_rate": 0.00029974154285157496,
      "loss": 2.0053,
      "step": 3020
    },
    {
      "epoch": 20.34,
      "learning_rate": 0.0002982647236245017,
      "loss": 1.9484,
      "step": 3030
    },
    {
      "epoch": 20.4,
      "learning_rate": 0.00029678615285916126,
      "loss": 1.9829,
      "step": 3040
    },
    {
      "epoch": 20.47,
      "learning_rate": 0.0002953058842132373,
      "loss": 1.9263,
      "step": 3050
    },
    {
      "epoch": 20.54,
      "learning_rate": 0.00029382397140603013,
      "loss": 1.955,
      "step": 3060
    },
    {
      "epoch": 20.6,
      "learning_rate": 0.00029234046821650694,
      "loss": 2.0098,
      "step": 3070
    },
    {
      "epoch": 20.67,
      "learning_rate": 0.0002908554284813503,
      "loss": 1.9278,
      "step": 3080
    },
    {
      "epoch": 20.74,
      "learning_rate": 0.00028936890609300447,
      "loss": 2.0939,
      "step": 3090
    },
    {
      "epoch": 20.81,
      "learning_rate": 0.00028788095499771943,
      "loss": 1.9715,
      "step": 3100
    },
    {
      "epoch": 20.87,
      "learning_rate": 0.0002863916291935933,
      "loss": 2.0042,
      "step": 3110
    },
    {
      "epoch": 20.94,
      "learning_rate": 0.00028490098272861283,
      "loss": 2.0696,
      "step": 3120
    },
    {
      "epoch": 21.01,
      "learning_rate": 0.0002834090696986916,
      "loss": 2.0059,
      "step": 3130
    },
    {
      "epoch": 21.07,
      "learning_rate": 0.00028191594424570754,
      "loss": 1.9319,
      "step": 3140
    },
    {
      "epoch": 21.14,
      "learning_rate": 0.00028042166055553723,
      "loss": 2.0266,
      "step": 3150
    },
    {
      "epoch": 21.21,
      "learning_rate": 0.00027892627285609035,
      "loss": 1.9712,
      "step": 3160
    },
    {
      "epoch": 21.28,
      "learning_rate": 0.0002774298354153411,
      "loss": 1.9637,
      "step": 3170
    },
    {
      "epoch": 21.34,
      "learning_rate": 0.0002759324025393591,
      "loss": 1.9696,
      "step": 3180
    },
    {
      "epoch": 21.41,
      "learning_rate": 0.00027443402857033864,
      "loss": 1.9457,
      "step": 3190
    },
    {
      "epoch": 21.48,
      "learning_rate": 0.00027293476788462623,
      "loss": 1.9282,
      "step": 3200
    },
    {
      "epoch": 21.54,
      "learning_rate": 0.000271434674890748,
      "loss": 2.0554,
      "step": 3210
    },
    {
      "epoch": 21.61,
      "learning_rate": 0.0002699338040274343,
      "loss": 1.9161,
      "step": 3220
    },
    {
      "epoch": 21.68,
      "learning_rate": 0.0002684322097616448,
      "loss": 1.9126,
      "step": 3230
    },
    {
      "epoch": 21.74,
      "learning_rate": 0.0002669299465865914,
      "loss": 2.0167,
      "step": 3240
    },
    {
      "epoch": 21.81,
      "learning_rate": 0.00026542706901976125,
      "loss": 1.9754,
      "step": 3250
    },
    {
      "epoch": 21.88,
      "learning_rate": 0.00026392363160093746,
      "loss": 1.9503,
      "step": 3260
    },
    {
      "epoch": 21.95,
      "learning_rate": 0.00026241968889022065,
      "loss": 2.0577,
      "step": 3270
    },
    {
      "epoch": 22.01,
      "learning_rate": 0.0002609152954660484,
      "loss": 1.9624,
      "step": 3280
    },
    {
      "epoch": 22.08,
      "learning_rate": 0.000259410505923215,
      "loss": 1.9243,
      "step": 3290
    },
    {
      "epoch": 22.15,
      "learning_rate": 0.0002579053748708897,
      "loss": 1.9068,
      "step": 3300
    },
    {
      "epoch": 22.21,
      "learning_rate": 0.0002563999569306355,
      "loss": 1.8909,
      "step": 3310
    },
    {
      "epoch": 22.28,
      "learning_rate": 0.00025489430673442635,
      "loss": 1.9244,
      "step": 3320
    },
    {
      "epoch": 22.35,
      "learning_rate": 0.0002533884789226651,
      "loss": 1.9617,
      "step": 3330
    },
    {
      "epoch": 22.42,
      "learning_rate": 0.00025188252814220004,
      "loss": 1.8954,
      "step": 3340
    },
    {
      "epoch": 22.48,
      "learning_rate": 0.0002503765090443422,
      "loss": 1.95,
      "step": 3350
    },
    {
      "epoch": 22.55,
      "learning_rate": 0.0002488704762828819,
      "loss": 1.9047,
      "step": 3360
    },
    {
      "epoch": 22.62,
      "learning_rate": 0.0002473644845121051,
      "loss": 2.0102,
      "step": 3370
    },
    {
      "epoch": 22.68,
      "learning_rate": 0.00024585858838481055,
      "loss": 1.9811,
      "step": 3380
    },
    {
      "epoch": 22.75,
      "learning_rate": 0.0002443528425503256,
      "loss": 1.8512,
      "step": 3390
    },
    {
      "epoch": 22.82,
      "learning_rate": 0.00024284730165252387,
      "loss": 1.929,
      "step": 3400
    },
    {
      "epoch": 22.89,
      "learning_rate": 0.00024134202032784156,
      "loss": 1.8707,
      "step": 3410
    },
    {
      "epoch": 22.95,
      "learning_rate": 0.00023983705320329508,
      "loss": 1.8647,
      "step": 3420
    },
    {
      "epoch": 23.02,
      "learning_rate": 0.00023833245489449805,
      "loss": 1.9206,
      "step": 3430
    },
    {
      "epoch": 23.09,
      "learning_rate": 0.00023682828000368012,
      "loss": 1.8319,
      "step": 3440
    },
    {
      "epoch": 23.15,
      "learning_rate": 0.00023532458311770486,
      "loss": 1.8411,
      "step": 3450
    },
    {
      "epoch": 23.22,
      "learning_rate": 0.0002338214188060887,
      "loss": 1.7666,
      "step": 3460
    },
    {
      "epoch": 23.29,
      "learning_rate": 0.0002323188416190211,
      "loss": 1.8131,
      "step": 3470
    },
    {
      "epoch": 23.36,
      "learning_rate": 0.00023081690608538463,
      "loss": 1.9416,
      "step": 3480
    },
    {
      "epoch": 23.42,
      "learning_rate": 0.00022931566671077603,
      "loss": 1.8075,
      "step": 3490
    },
    {
      "epoch": 23.49,
      "learning_rate": 0.0002278151779755281,
      "loss": 1.9087,
      "step": 3500
    },
    {
      "epoch": 23.49,
      "eval_accuracy": 0.6922669410705566,
      "eval_accuracy_top5": 0.7862817645072937,
      "eval_loss": 1.3863152265548706,
      "eval_runtime": 99.8777,
      "eval_samples_per_second": 37.876,
      "eval_steps_per_second": 4.736,
      "step": 3500
    },
    {
      "epoch": 23.56,
      "learning_rate": 0.00022631549433273294,
      "loss": 1.9428,
      "step": 3510
    },
    {
      "epoch": 23.62,
      "learning_rate": 0.0002248166702062657,
      "loss": 1.8847,
      "step": 3520
    },
    {
      "epoch": 23.69,
      "learning_rate": 0.00022331875998880943,
      "loss": 1.9183,
      "step": 3530
    },
    {
      "epoch": 23.76,
      "learning_rate": 0.0002218218180398811,
      "loss": 1.9658,
      "step": 3540
    },
    {
      "epoch": 23.83,
      "learning_rate": 0.0002203258986838591,
      "loss": 1.8596,
      "step": 3550
    },
    {
      "epoch": 23.89,
      "learning_rate": 0.00021883105620801164,
      "loss": 1.9158,
      "step": 3560
    },
    {
      "epoch": 23.96,
      "learning_rate": 0.0002173373448605267,
      "loss": 1.8477,
      "step": 3570
    },
    {
      "epoch": 24.03,
      "learning_rate": 0.0002158448188485433,
      "loss": 1.9931,
      "step": 3580
    },
    {
      "epoch": 24.09,
      "learning_rate": 0.00021435353233618446,
      "loss": 1.8643,
      "step": 3590
    },
    {
      "epoch": 24.16,
      "learning_rate": 0.00021286353944259138,
      "loss": 1.8867,
      "step": 3600
    },
    {
      "epoch": 24.23,
      "learning_rate": 0.00021137489423995963,
      "loss": 1.9012,
      "step": 3610
    },
    {
      "epoch": 24.3,
      "learning_rate": 0.00020988765075157648,
      "loss": 1.8438,
      "step": 3620
    },
    {
      "epoch": 24.36,
      "learning_rate": 0.00020840186294986097,
      "loss": 1.8712,
      "step": 3630
    },
    {
      "epoch": 24.43,
      "learning_rate": 0.00020691758475440482,
      "loss": 1.8504,
      "step": 3640
    },
    {
      "epoch": 24.5,
      "learning_rate": 0.0002054348700300158,
      "loss": 1.9375,
      "step": 3650
    },
    {
      "epoch": 24.56,
      "learning_rate": 0.00020395377258476278,
      "loss": 1.8488,
      "step": 3660
    },
    {
      "epoch": 24.63,
      "learning_rate": 0.0002024743461680234,
      "loss": 1.8815,
      "step": 3670
    },
    {
      "epoch": 24.7,
      "learning_rate": 0.00020099664446853328,
      "loss": 1.8098,
      "step": 3680
    },
    {
      "epoch": 24.77,
      "learning_rate": 0.00019952072111243738,
      "loss": 1.8489,
      "step": 3690
    },
    {
      "epoch": 24.83,
      "learning_rate": 0.00019804662966134442,
      "loss": 1.9232,
      "step": 3700
    },
    {
      "epoch": 24.9,
      "learning_rate": 0.0001965744236103828,
      "loss": 1.8177,
      "step": 3710
    },
    {
      "epoch": 24.97,
      "learning_rate": 0.0001951041563862593,
      "loss": 1.9149,
      "step": 3720
    },
    {
      "epoch": 25.03,
      "learning_rate": 0.00019363588134532007,
      "loss": 1.8291,
      "step": 3730
    },
    {
      "epoch": 25.1,
      "learning_rate": 0.0001921696517716147,
      "loss": 1.883,
      "step": 3740
    },
    {
      "epoch": 25.17,
      "learning_rate": 0.00019070552087496203,
      "loss": 1.8461,
      "step": 3750
    },
    {
      "epoch": 25.23,
      "learning_rate": 0.0001892435417890197,
      "loss": 1.8295,
      "step": 3760
    },
    {
      "epoch": 25.3,
      "learning_rate": 0.00018778376756935534,
      "loss": 1.887,
      "step": 3770
    },
    {
      "epoch": 25.37,
      "learning_rate": 0.00018632625119152163,
      "loss": 1.8155,
      "step": 3780
    },
    {
      "epoch": 25.44,
      "learning_rate": 0.0001848710455491336,
      "loss": 1.8593,
      "step": 3790
    },
    {
      "epoch": 25.5,
      "learning_rate": 0.00018341820345194932,
      "loss": 1.7904,
      "step": 3800
    },
    {
      "epoch": 25.57,
      "learning_rate": 0.00018196777762395275,
      "loss": 1.9035,
      "step": 3810
    },
    {
      "epoch": 25.64,
      "learning_rate": 0.00018051982070144135,
      "loss": 1.8704,
      "step": 3820
    },
    {
      "epoch": 25.7,
      "learning_rate": 0.00017907438523111528,
      "loss": 1.7725,
      "step": 3830
    },
    {
      "epoch": 25.77,
      "learning_rate": 0.0001776315236681706,
      "loss": 1.7824,
      "step": 3840
    },
    {
      "epoch": 25.84,
      "learning_rate": 0.00017619128837439546,
      "loss": 1.9004,
      "step": 3850
    },
    {
      "epoch": 25.91,
      "learning_rate": 0.00017475373161627035,
      "loss": 1.7393,
      "step": 3860
    },
    {
      "epoch": 25.97,
      "learning_rate": 0.00017331890556307095,
      "loss": 1.8055,
      "step": 3870
    },
    {
      "epoch": 26.04,
      "learning_rate": 0.0001718868622849752,
      "loss": 1.6775,
      "step": 3880
    },
    {
      "epoch": 26.11,
      "learning_rate": 0.00017045765375117299,
      "loss": 1.765,
      "step": 3890
    },
    {
      "epoch": 26.17,
      "learning_rate": 0.00016903133182798115,
      "loss": 1.7172,
      "step": 3900
    },
    {
      "epoch": 26.24,
      "learning_rate": 0.00016760794827696055,
      "loss": 1.7846,
      "step": 3910
    },
    {
      "epoch": 26.31,
      "learning_rate": 0.0001661875547530378,
      "loss": 1.8185,
      "step": 3920
    },
    {
      "epoch": 26.38,
      "learning_rate": 0.0001647702028026308,
      "loss": 1.7271,
      "step": 3930
    },
    {
      "epoch": 26.44,
      "learning_rate": 0.00016335594386177794,
      "loss": 1.7887,
      "step": 3940
    },
    {
      "epoch": 26.51,
      "learning_rate": 0.0001619448292542716,
      "loss": 1.7533,
      "step": 3950
    },
    {
      "epoch": 26.58,
      "learning_rate": 0.0001605369101897956,
      "loss": 1.8131,
      "step": 3960
    },
    {
      "epoch": 26.64,
      "learning_rate": 0.00015913223776206666,
      "loss": 1.7736,
      "step": 3970
    },
    {
      "epoch": 26.71,
      "learning_rate": 0.00015773086294698036,
      "loss": 1.7481,
      "step": 3980
    },
    {
      "epoch": 26.78,
      "learning_rate": 0.0001563328366007612,
      "loss": 1.8161,
      "step": 3990
    },
    {
      "epoch": 26.85,
      "learning_rate": 0.0001549382094581166,
      "loss": 1.816,
      "step": 4000
    },
    {
      "epoch": 26.85,
      "eval_accuracy": 0.7110699415206909,
      "eval_accuracy_top5": 0.7913135886192322,
      "eval_loss": 1.3816050291061401,
      "eval_runtime": 1281.5963,
      "eval_samples_per_second": 2.952,
      "eval_steps_per_second": 0.369,
      "step": 4000
    },
    {
      "epoch": 26.91,
      "learning_rate": 0.00015354703213039651,
      "loss": 1.7483,
      "step": 4010
    },
    {
      "epoch": 26.98,
      "learning_rate": 0.00015215935510375607,
      "loss": 1.7502,
      "step": 4020
    },
    {
      "epoch": 27.05,
      "learning_rate": 0.00015077522873732375,
      "loss": 1.7954,
      "step": 4030
    },
    {
      "epoch": 27.11,
      "learning_rate": 0.0001493947032613735,
      "loss": 1.7957,
      "step": 4040
    },
    {
      "epoch": 27.18,
      "learning_rate": 0.00014801782877550234,
      "loss": 1.7301,
      "step": 4050
    },
    {
      "epoch": 27.25,
      "learning_rate": 0.00014664465524681197,
      "loss": 1.7393,
      "step": 4060
    },
    {
      "epoch": 27.32,
      "learning_rate": 0.00014527523250809543,
      "loss": 1.7667,
      "step": 4070
    },
    {
      "epoch": 27.38,
      "learning_rate": 0.00014390961025602855,
      "loss": 1.8113,
      "step": 4080
    },
    {
      "epoch": 27.45,
      "learning_rate": 0.00014254783804936686,
      "loss": 1.7796,
      "step": 4090
    },
    {
      "epoch": 27.52,
      "learning_rate": 0.00014118996530714664,
      "loss": 1.7518,
      "step": 4100
    },
    {
      "epoch": 27.58,
      "learning_rate": 0.00013983604130689192,
      "loss": 1.7064,
      "step": 4110
    },
    {
      "epoch": 27.65,
      "learning_rate": 0.0001384861151828255,
      "loss": 1.754,
      "step": 4120
    },
    {
      "epoch": 27.72,
      "learning_rate": 0.00013714023592408678,
      "loss": 1.7321,
      "step": 4130
    },
    {
      "epoch": 27.79,
      "learning_rate": 0.0001357984523729533,
      "loss": 1.787,
      "step": 4140
    },
    {
      "epoch": 27.85,
      "learning_rate": 0.00013446081322306813,
      "loss": 1.7389,
      "step": 4150
    },
    {
      "epoch": 27.92,
      "learning_rate": 0.00013312736701767347,
      "loss": 1.7504,
      "step": 4160
    },
    {
      "epoch": 27.99,
      "learning_rate": 0.00013179816214784826,
      "loss": 1.7239,
      "step": 4170
    },
    {
      "epoch": 28.05,
      "learning_rate": 0.00013047324685075263,
      "loss": 1.7251,
      "step": 4180
    },
    {
      "epoch": 28.12,
      "learning_rate": 0.00012915266920787672,
      "loss": 1.7394,
      "step": 4190
    },
    {
      "epoch": 28.19,
      "learning_rate": 0.00012783647714329649,
      "loss": 1.7623,
      "step": 4200
    },
    {
      "epoch": 28.26,
      "learning_rate": 0.00012652471842193415,
      "loss": 1.68,
      "step": 4210
    },
    {
      "epoch": 28.32,
      "learning_rate": 0.00012521744064782476,
      "loss": 1.7476,
      "step": 4220
    },
    {
      "epoch": 28.39,
      "learning_rate": 0.00012391469126238884,
      "loss": 1.7888,
      "step": 4230
    },
    {
      "epoch": 28.46,
      "learning_rate": 0.0001226165175427105,
      "loss": 1.8063,
      "step": 4240
    },
    {
      "epoch": 28.52,
      "learning_rate": 0.00012132296659982206,
      "loss": 1.7794,
      "step": 4250
    },
    {
      "epoch": 28.59,
      "learning_rate": 0.00012003408537699384,
      "loss": 1.7629,
      "step": 4260
    },
    {
      "epoch": 28.66,
      "learning_rate": 0.00011874992064803114,
      "loss": 1.7939,
      "step": 4270
    },
    {
      "epoch": 28.72,
      "learning_rate": 0.0001174705190155766,
      "loss": 1.7379,
      "step": 4280
    },
    {
      "epoch": 28.79,
      "learning_rate": 0.00011619592690941886,
      "loss": 1.7218,
      "step": 4290
    },
    {
      "epoch": 28.86,
      "learning_rate": 0.00011492619058480783,
      "loss": 1.7113,
      "step": 4300
    },
    {
      "epoch": 28.93,
      "learning_rate": 0.00011366135612077571,
      "loss": 1.7242,
      "step": 4310
    },
    {
      "epoch": 28.99,
      "learning_rate": 0.00011240146941846526,
      "loss": 1.7338,
      "step": 4320
    },
    {
      "epoch": 29.06,
      "learning_rate": 0.00011114657619946372,
      "loss": 1.67,
      "step": 4330
    },
    {
      "epoch": 29.13,
      "learning_rate": 0.00010989672200414375,
      "loss": 1.7605,
      "step": 4340
    },
    {
      "epoch": 29.19,
      "learning_rate": 0.00010865195219001028,
      "loss": 1.682,
      "step": 4350
    },
    {
      "epoch": 29.26,
      "learning_rate": 0.00010741231193005521,
      "loss": 1.6429,
      "step": 4360
    },
    {
      "epoch": 29.33,
      "learning_rate": 0.00010617784621111767,
      "loss": 1.6666,
      "step": 4370
    },
    {
      "epoch": 29.4,
      "learning_rate": 0.0001049485998322512,
      "loss": 1.6439,
      "step": 4380
    },
    {
      "epoch": 29.46,
      "learning_rate": 0.00010372461740309849,
      "loss": 1.6682,
      "step": 4390
    },
    {
      "epoch": 29.53,
      "learning_rate": 0.00010250594334227223,
      "loss": 1.8048,
      "step": 4400
    },
    {
      "epoch": 29.6,
      "learning_rate": 0.00010129262187574318,
      "loss": 1.7627,
      "step": 4410
    },
    {
      "epoch": 29.66,
      "learning_rate": 0.00010008469703523492,
      "loss": 1.8278,
      "step": 4420
    },
    {
      "epoch": 29.73,
      "learning_rate": 9.888221265662655e-05,
      "loss": 1.7661,
      "step": 4430
    },
    {
      "epoch": 29.8,
      "learning_rate": 9.768521237836131e-05,
      "loss": 1.7062,
      "step": 4440
    },
    {
      "epoch": 29.87,
      "learning_rate": 9.64937396398633e-05,
      "loss": 1.6505,
      "step": 4450
    },
    {
      "epoch": 29.93,
      "learning_rate": 9.530783767996057e-05,
      "loss": 1.688,
      "step": 4460
    },
    {
      "epoch": 30.0,
      "learning_rate": 9.412754953531663e-05,
      "loss": 1.6097,
      "step": 4470
    },
    {
      "epoch": 30.07,
      "learning_rate": 9.295291803886818e-05,
      "loss": 1.6393,
      "step": 4480
    },
    {
      "epoch": 30.13,
      "learning_rate": 9.178398581827086e-05,
      "loss": 1.6005,
      "step": 4490
    },
    {
      "epoch": 30.2,
      "learning_rate": 9.062079529435204e-05,
      "loss": 1.7324,
      "step": 4500
    },
    {
      "epoch": 30.2,
      "eval_accuracy": 0.7243114113807678,
      "eval_accuracy_top5": 0.8164724707603455,
      "eval_loss": 1.2510910034179688,
      "eval_runtime": 99.3635,
      "eval_samples_per_second": 38.072,
      "eval_steps_per_second": 4.76,
      "step": 4500
    },
    {
      "epoch": 30.27,
      "learning_rate": 8.946338867957182e-05,
      "loss": 1.6926,
      "step": 4510
    },
    {
      "epoch": 30.34,
      "learning_rate": 8.831180797649071e-05,
      "loss": 1.6246,
      "step": 4520
    },
    {
      "epoch": 30.4,
      "learning_rate": 8.716609497624564e-05,
      "loss": 1.7316,
      "step": 4530
    },
    {
      "epoch": 30.47,
      "learning_rate": 8.602629125703296e-05,
      "loss": 1.6764,
      "step": 4540
    },
    {
      "epoch": 30.54,
      "learning_rate": 8.489243818260004e-05,
      "loss": 1.664,
      "step": 4550
    },
    {
      "epoch": 30.6,
      "learning_rate": 8.376457690074384e-05,
      "loss": 1.6697,
      "step": 4560
    },
    {
      "epoch": 30.67,
      "learning_rate": 8.264274834181793e-05,
      "loss": 1.6584,
      "step": 4570
    },
    {
      "epoch": 30.74,
      "learning_rate": 8.152699321724652e-05,
      "loss": 1.6844,
      "step": 4580
    },
    {
      "epoch": 30.81,
      "learning_rate": 8.041735201804783e-05,
      "loss": 1.7346,
      "step": 4590
    },
    {
      "epoch": 30.87,
      "learning_rate": 7.931386501336418e-05,
      "loss": 1.7664,
      "step": 4600
    },
    {
      "epoch": 30.94,
      "learning_rate": 7.821657224900064e-05,
      "loss": 1.7337,
      "step": 4610
    },
    {
      "epoch": 31.01,
      "learning_rate": 7.712551354597187e-05,
      "loss": 1.5892,
      "step": 4620
    },
    {
      "epoch": 31.07,
      "learning_rate": 7.604072849905708e-05,
      "loss": 1.6845,
      "step": 4630
    },
    {
      "epoch": 31.14,
      "learning_rate": 7.49622564753627e-05,
      "loss": 1.6969,
      "step": 4640
    },
    {
      "epoch": 31.21,
      "learning_rate": 7.38901366128944e-05,
      "loss": 1.6825,
      "step": 4650
    },
    {
      "epoch": 31.28,
      "learning_rate": 7.282440781913619e-05,
      "loss": 1.6149,
      "step": 4660
    },
    {
      "epoch": 31.34,
      "learning_rate": 7.176510876963876e-05,
      "loss": 1.6943,
      "step": 4670
    },
    {
      "epoch": 31.41,
      "learning_rate": 7.071227790661597e-05,
      "loss": 1.7003,
      "step": 4680
    },
    {
      "epoch": 31.48,
      "learning_rate": 6.966595343754936e-05,
      "loss": 1.754,
      "step": 4690
    },
    {
      "epoch": 31.54,
      "learning_rate": 6.862617333380214e-05,
      "loss": 1.7057,
      "step": 4700
    },
    {
      "epoch": 31.61,
      "learning_rate": 6.759297532924086e-05,
      "loss": 1.6623,
      "step": 4710
    },
    {
      "epoch": 31.68,
      "learning_rate": 6.656639691886629e-05,
      "loss": 1.5663,
      "step": 4720
    },
    {
      "epoch": 31.74,
      "learning_rate": 6.55464753574522e-05,
      "loss": 1.62,
      "step": 4730
    },
    {
      "epoch": 31.81,
      "learning_rate": 6.453324765819404e-05,
      "loss": 1.6554,
      "step": 4740
    },
    {
      "epoch": 31.88,
      "learning_rate": 6.352675059136531e-05,
      "loss": 1.5982,
      "step": 4750
    },
    {
      "epoch": 31.95,
      "learning_rate": 6.252702068298338e-05,
      "loss": 1.681,
      "step": 4760
    },
    {
      "epoch": 32.01,
      "learning_rate": 6.153409421348358e-05,
      "loss": 1.6682,
      "step": 4770
    },
    {
      "epoch": 32.08,
      "learning_rate": 6.054800721640305e-05,
      "loss": 1.6059,
      "step": 4780
    },
    {
      "epoch": 32.15,
      "learning_rate": 5.956879547707275e-05,
      "loss": 1.661,
      "step": 4790
    },
    {
      "epoch": 32.21,
      "learning_rate": 5.8596494531319045e-05,
      "loss": 1.6645,
      "step": 4800
    },
    {
      "epoch": 32.28,
      "learning_rate": 5.763113966417369e-05,
      "loss": 1.6481,
      "step": 4810
    },
    {
      "epoch": 32.35,
      "learning_rate": 5.667276590859385e-05,
      "loss": 1.6029,
      "step": 4820
    },
    {
      "epoch": 32.42,
      "learning_rate": 5.572140804419049e-05,
      "loss": 1.6251,
      "step": 4830
    },
    {
      "epoch": 32.48,
      "learning_rate": 5.477710059596599e-05,
      "loss": 1.676,
      "step": 4840
    },
    {
      "epoch": 32.55,
      "learning_rate": 5.3839877833061684e-05,
      "loss": 1.6486,
      "step": 4850
    },
    {
      "epoch": 32.62,
      "learning_rate": 5.2909773767513934e-05,
      "loss": 1.6394,
      "step": 4860
    },
    {
      "epoch": 32.68,
      "learning_rate": 5.198682215301989e-05,
      "loss": 1.5951,
      "step": 4870
    },
    {
      "epoch": 32.75,
      "learning_rate": 5.1071056483712435e-05,
      "loss": 1.6508,
      "step": 4880
    },
    {
      "epoch": 32.82,
      "learning_rate": 5.016250999294497e-05,
      "loss": 1.615,
      "step": 4890
    },
    {
      "epoch": 32.89,
      "learning_rate": 4.9261215652085105e-05,
      "loss": 1.6343,
      "step": 4900
    },
    {
      "epoch": 32.95,
      "learning_rate": 4.8367206169318305e-05,
      "loss": 1.608,
      "step": 4910
    },
    {
      "epoch": 33.02,
      "learning_rate": 4.7480513988460625e-05,
      "loss": 1.5985,
      "step": 4920
    },
    {
      "epoch": 33.09,
      "learning_rate": 4.660117128778163e-05,
      "loss": 1.6053,
      "step": 4930
    },
    {
      "epoch": 33.15,
      "learning_rate": 4.572920997883648e-05,
      "loss": 1.6875,
      "step": 4940
    },
    {
      "epoch": 33.22,
      "learning_rate": 4.486466170530798e-05,
      "loss": 1.6692,
      "step": 4950
    },
    {
      "epoch": 33.29,
      "learning_rate": 4.4007557841857865e-05,
      "loss": 1.7254,
      "step": 4960
    },
    {
      "epoch": 33.36,
      "learning_rate": 4.315792949298869e-05,
      "loss": 1.6377,
      "step": 4970
    },
    {
      "epoch": 33.42,
      "learning_rate": 4.231580749191474e-05,
      "loss": 1.6221,
      "step": 4980
    },
    {
      "epoch": 33.49,
      "learning_rate": 4.148122239944316e-05,
      "loss": 1.7078,
      "step": 4990
    },
    {
      "epoch": 33.56,
      "learning_rate": 4.0654204502864886e-05,
      "loss": 1.5947,
      "step": 5000
    },
    {
      "epoch": 33.56,
      "eval_accuracy": 0.7288135886192322,
      "eval_accuracy_top5": 0.8172669410705566,
      "eval_loss": 1.2728419303894043,
      "eval_runtime": 100.0466,
      "eval_samples_per_second": 37.812,
      "eval_steps_per_second": 4.728,
      "step": 5000
    },
    {
      "epoch": 33.62,
      "learning_rate": 3.983478381485558e-05,
      "loss": 1.6253,
      "step": 5010
    },
    {
      "epoch": 33.69,
      "learning_rate": 3.902299007238627e-05,
      "loss": 1.6119,
      "step": 5020
    },
    {
      "epoch": 33.76,
      "learning_rate": 3.8218852735644404e-05,
      "loss": 1.5703,
      "step": 5030
    },
    {
      "epoch": 33.83,
      "learning_rate": 3.7422400986964724e-05,
      "loss": 1.6635,
      "step": 5040
    },
    {
      "epoch": 33.89,
      "learning_rate": 3.6633663729770004e-05,
      "loss": 1.6116,
      "step": 5050
    },
    {
      "epoch": 33.96,
      "learning_rate": 3.585266958752248e-05,
      "loss": 1.5966,
      "step": 5060
    },
    {
      "epoch": 34.03,
      "learning_rate": 3.507944690268469e-05,
      "loss": 1.656,
      "step": 5070
    },
    {
      "epoch": 34.09,
      "learning_rate": 3.4314023735691286e-05,
      "loss": 1.6135,
      "step": 5080
    },
    {
      "epoch": 34.16,
      "learning_rate": 3.355642786393051e-05,
      "loss": 1.6592,
      "step": 5090
    },
    {
      "epoch": 34.23,
      "learning_rate": 3.2806686780736336e-05,
      "loss": 1.5753,
      "step": 5100
    },
    {
      "epoch": 34.3,
      "learning_rate": 3.2064827694390345e-05,
      "loss": 1.6342,
      "step": 5110
    },
    {
      "epoch": 34.36,
      "learning_rate": 3.1330877527134785e-05,
      "loss": 1.6055,
      "step": 5120
    },
    {
      "epoch": 34.43,
      "learning_rate": 3.060486291419531e-05,
      "loss": 1.6282,
      "step": 5130
    },
    {
      "epoch": 34.5,
      "learning_rate": 2.9886810202814447e-05,
      "loss": 1.6759,
      "step": 5140
    },
    {
      "epoch": 34.56,
      "learning_rate": 2.917674545129531e-05,
      "loss": 1.6517,
      "step": 5150
    },
    {
      "epoch": 34.63,
      "learning_rate": 2.847469442805614e-05,
      "loss": 1.5875,
      "step": 5160
    },
    {
      "epoch": 34.7,
      "learning_rate": 2.7780682610695136e-05,
      "loss": 1.6003,
      "step": 5170
    },
    {
      "epoch": 34.77,
      "learning_rate": 2.7094735185065778e-05,
      "loss": 1.5555,
      "step": 5180
    },
    {
      "epoch": 34.83,
      "learning_rate": 2.6416877044362685e-05,
      "loss": 1.6044,
      "step": 5190
    },
    {
      "epoch": 34.9,
      "learning_rate": 2.5747132788218663e-05,
      "loss": 1.6296,
      "step": 5200
    },
    {
      "epoch": 34.97,
      "learning_rate": 2.508552672181158e-05,
      "loss": 1.6314,
      "step": 5210
    },
    {
      "epoch": 35.03,
      "learning_rate": 2.4432082854982524e-05,
      "loss": 1.6526,
      "step": 5220
    },
    {
      "epoch": 35.1,
      "learning_rate": 2.3786824901364357e-05,
      "loss": 1.6304,
      "step": 5230
    },
    {
      "epoch": 35.17,
      "learning_rate": 2.3149776277521266e-05,
      "loss": 1.5172,
      "step": 5240
    },
    {
      "epoch": 35.23,
      "learning_rate": 2.2520960102098892e-05,
      "loss": 1.606,
      "step": 5250
    },
    {
      "epoch": 35.3,
      "learning_rate": 2.190039919498543e-05,
      "loss": 1.5952,
      "step": 5260
    },
    {
      "epoch": 35.37,
      "learning_rate": 2.1288116076483288e-05,
      "loss": 1.5936,
      "step": 5270
    },
    {
      "epoch": 35.44,
      "learning_rate": 2.0684132966492103e-05,
      "loss": 1.6147,
      "step": 5280
    },
    {
      "epoch": 35.5,
      "learning_rate": 2.008847178370221e-05,
      "loss": 1.6177,
      "step": 5290
    },
    {
      "epoch": 35.57,
      "learning_rate": 1.9501154144799137e-05,
      "loss": 1.6251,
      "step": 5300
    },
    {
      "epoch": 35.64,
      "learning_rate": 1.8922201363679338e-05,
      "loss": 1.6342,
      "step": 5310
    },
    {
      "epoch": 35.7,
      "learning_rate": 1.8351634450676502e-05,
      "loss": 1.6048,
      "step": 5320
    },
    {
      "epoch": 35.77,
      "learning_rate": 1.7789474111799318e-05,
      "loss": 1.5749,
      "step": 5330
    },
    {
      "epoch": 35.84,
      "learning_rate": 1.7235740747979733e-05,
      "loss": 1.6196,
      "step": 5340
    },
    {
      "epoch": 35.91,
      "learning_rate": 1.6690454454332843e-05,
      "loss": 1.5682,
      "step": 5350
    },
    {
      "epoch": 35.97,
      "learning_rate": 1.6153635019427598e-05,
      "loss": 1.6569,
      "step": 5360
    },
    {
      "epoch": 36.04,
      "learning_rate": 1.5625301924568626e-05,
      "loss": 1.6054,
      "step": 5370
    },
    {
      "epoch": 36.11,
      "learning_rate": 1.5105474343089205e-05,
      "loss": 1.6504,
      "step": 5380
    },
    {
      "epoch": 36.17,
      "learning_rate": 1.459417113965561e-05,
      "loss": 1.606,
      "step": 5390
    },
    {
      "epoch": 36.24,
      "learning_rate": 1.4091410869582266e-05,
      "loss": 1.627,
      "step": 5400
    },
    {
      "epoch": 36.31,
      "learning_rate": 1.3597211778158675e-05,
      "loss": 1.5914,
      "step": 5410
    },
    {
      "epoch": 36.38,
      "learning_rate": 1.3111591799987083e-05,
      "loss": 1.6279,
      "step": 5420
    },
    {
      "epoch": 36.44,
      "learning_rate": 1.2634568558331644e-05,
      "loss": 1.6583,
      "step": 5430
    },
    {
      "epoch": 36.51,
      "learning_rate": 1.216615936447893e-05,
      "loss": 1.567,
      "step": 5440
    },
    {
      "epoch": 36.58,
      "learning_rate": 1.1706381217109735e-05,
      "loss": 1.6408,
      "step": 5450
    },
    {
      "epoch": 36.64,
      "learning_rate": 1.1255250801681994e-05,
      "loss": 1.5914,
      "step": 5460
    },
    {
      "epoch": 36.71,
      "learning_rate": 1.0812784489825506e-05,
      "loss": 1.5906,
      "step": 5470
    },
    {
      "epoch": 36.78,
      "learning_rate": 1.0378998338747669e-05,
      "loss": 1.541,
      "step": 5480
    },
    {
      "epoch": 36.85,
      "learning_rate": 9.953908090650804e-06,
      "loss": 1.6335,
      "step": 5490
    },
    {
      "epoch": 36.91,
      "learning_rate": 9.53752917216083e-06,
      "loss": 1.6565,
      "step": 5500
    },
    {
      "epoch": 36.91,
      "eval_accuracy": 0.7513241767883301,
      "eval_accuracy_top5": 0.8336864113807678,
      "eval_loss": 1.1887362003326416,
      "eval_runtime": 101.4114,
      "eval_samples_per_second": 37.304,
      "eval_steps_per_second": 4.664,
      "step": 5500
    }
  ],
  "max_steps": 5960,
  "num_train_epochs": 40,
  "total_flos": 4.3900198664955494e+20,
  "trial_name": null,
  "trial_params": null
}