{
  "best_metric": 0.9943422913719944,
  "best_model_checkpoint": "swinv2-large-patch4-window12to16-192to256-22kto1k-ft-finetuned-BreastCancer-BreakHis-AH-60-20-20/checkpoint-1194",
  "epoch": 7.0,
  "global_step": 1393,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.03,
      "learning_rate": 3.586800573888092e-07,
      "loss": 0.7746,
      "step": 5
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.173601147776184e-07,
      "loss": 0.739,
      "step": 10
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.0760401721664278e-06,
      "loss": 0.7228,
      "step": 15
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.4347202295552369e-06,
      "loss": 0.7082,
      "step": 20
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.793400286944046e-06,
      "loss": 0.6802,
      "step": 25
    },
    {
      "epoch": 0.15,
      "learning_rate": 2.1520803443328555e-06,
      "loss": 0.6662,
      "step": 30
    },
    {
      "epoch": 0.18,
      "learning_rate": 2.5107604017216644e-06,
      "loss": 0.631,
      "step": 35
    },
    {
      "epoch": 0.2,
      "learning_rate": 2.8694404591104738e-06,
      "loss": 0.6168,
      "step": 40
    },
    {
      "epoch": 0.23,
      "learning_rate": 3.2281205164992827e-06,
      "loss": 0.5649,
      "step": 45
    },
    {
      "epoch": 0.25,
      "learning_rate": 3.586800573888092e-06,
      "loss": 0.5661,
      "step": 50
    },
    {
      "epoch": 0.28,
      "learning_rate": 3.945480631276901e-06,
      "loss": 0.4748,
      "step": 55
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.304160688665711e-06,
      "loss": 0.4842,
      "step": 60
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.66284074605452e-06,
      "loss": 0.4239,
      "step": 65
    },
    {
      "epoch": 0.35,
      "learning_rate": 5.021520803443329e-06,
      "loss": 0.4256,
      "step": 70
    },
    {
      "epoch": 0.38,
      "learning_rate": 5.380200860832138e-06,
      "loss": 0.3675,
      "step": 75
    },
    {
      "epoch": 0.4,
      "learning_rate": 5.7388809182209475e-06,
      "loss": 0.3646,
      "step": 80
    },
    {
      "epoch": 0.43,
      "learning_rate": 6.0975609756097564e-06,
      "loss": 0.3939,
      "step": 85
    },
    {
      "epoch": 0.45,
      "learning_rate": 6.456241032998565e-06,
      "loss": 0.3783,
      "step": 90
    },
    {
      "epoch": 0.48,
      "learning_rate": 6.814921090387374e-06,
      "loss": 0.4425,
      "step": 95
    },
    {
      "epoch": 0.5,
      "learning_rate": 7.173601147776184e-06,
      "loss": 0.348,
      "step": 100
    },
    {
      "epoch": 0.53,
      "learning_rate": 7.532281205164993e-06,
      "loss": 0.3337,
      "step": 105
    },
    {
      "epoch": 0.55,
      "learning_rate": 7.890961262553803e-06,
      "loss": 0.2808,
      "step": 110
    },
    {
      "epoch": 0.58,
      "learning_rate": 8.249641319942612e-06,
      "loss": 0.3154,
      "step": 115
    },
    {
      "epoch": 0.6,
      "learning_rate": 8.608321377331422e-06,
      "loss": 0.2637,
      "step": 120
    },
    {
      "epoch": 0.63,
      "learning_rate": 8.967001434720231e-06,
      "loss": 0.2857,
      "step": 125
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.32568149210904e-06,
      "loss": 0.2633,
      "step": 130
    },
    {
      "epoch": 0.68,
      "learning_rate": 9.684361549497849e-06,
      "loss": 0.3059,
      "step": 135
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.0043041606886658e-05,
      "loss": 0.2524,
      "step": 140
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.0401721664275467e-05,
      "loss": 0.2154,
      "step": 145
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.0760401721664276e-05,
      "loss": 0.3399,
      "step": 150
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.1119081779053084e-05,
      "loss": 0.2305,
      "step": 155
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.1477761836441895e-05,
      "loss": 0.2263,
      "step": 160
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.1836441893830704e-05,
      "loss": 0.1931,
      "step": 165
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.2195121951219513e-05,
      "loss": 0.2317,
      "step": 170
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.2553802008608323e-05,
      "loss": 0.2118,
      "step": 175
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.291248206599713e-05,
      "loss": 0.1408,
      "step": 180
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.3271162123385941e-05,
      "loss": 0.24,
      "step": 185
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.3629842180774748e-05,
      "loss": 0.2579,
      "step": 190
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.3988522238163559e-05,
      "loss": 0.2053,
      "step": 195
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9495520980669495,
      "eval_loss": 0.1227400153875351,
      "eval_runtime": 524.4475,
      "eval_samples_per_second": 4.044,
      "eval_steps_per_second": 0.507,
      "step": 199
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.4347202295552368e-05,
      "loss": 0.2,
      "step": 200
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.4705882352941177e-05,
      "loss": 0.2002,
      "step": 205
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.5064562410329986e-05,
      "loss": 0.2141,
      "step": 210
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.5423242467718796e-05,
      "loss": 0.2358,
      "step": 215
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.5781922525107605e-05,
      "loss": 0.1788,
      "step": 220
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.6140602582496414e-05,
      "loss": 0.0829,
      "step": 225
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.6499282639885223e-05,
      "loss": 0.205,
      "step": 230
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.6857962697274032e-05,
      "loss": 0.1433,
      "step": 235
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.7216642754662844e-05,
      "loss": 0.162,
      "step": 240
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.757532281205165e-05,
      "loss": 0.2635,
      "step": 245
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.7934002869440462e-05,
      "loss": 0.2703,
      "step": 250
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.8292682926829268e-05,
      "loss": 0.239,
      "step": 255
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.865136298421808e-05,
      "loss": 0.2087,
      "step": 260
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.9010043041606885e-05,
      "loss": 0.206,
      "step": 265
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.9368723098995698e-05,
      "loss": 0.1916,
      "step": 270
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.9727403156384507e-05,
      "loss": 0.2102,
      "step": 275
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.0086083213773316e-05,
      "loss": 0.1304,
      "step": 280
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.0444763271162124e-05,
      "loss": 0.1707,
      "step": 285
    },
    {
      "epoch": 1.46,
      "learning_rate": 2.0803443328550933e-05,
      "loss": 0.0719,
      "step": 290
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.1162123385939742e-05,
      "loss": 0.22,
      "step": 295
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.152080344332855e-05,
      "loss": 0.2759,
      "step": 300
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.187948350071736e-05,
      "loss": 0.2507,
      "step": 305
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.223816355810617e-05,
      "loss": 0.2151,
      "step": 310
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.259684361549498e-05,
      "loss": 0.1902,
      "step": 315
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.295552367288379e-05,
      "loss": 0.3037,
      "step": 320
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.33142037302726e-05,
      "loss": 0.2278,
      "step": 325
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.3672883787661408e-05,
      "loss": 0.2791,
      "step": 330
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.4031563845050217e-05,
      "loss": 0.1363,
      "step": 335
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.4390243902439026e-05,
      "loss": 0.2639,
      "step": 340
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.4748923959827835e-05,
      "loss": 0.108,
      "step": 345
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.5107604017216647e-05,
      "loss": 0.1188,
      "step": 350
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.5466284074605452e-05,
      "loss": 0.1908,
      "step": 355
    },
    {
      "epoch": 1.81,
      "learning_rate": 2.582496413199426e-05,
      "loss": 0.1316,
      "step": 360
    },
    {
      "epoch": 1.83,
      "learning_rate": 2.6183644189383074e-05,
      "loss": 0.2164,
      "step": 365
    },
    {
      "epoch": 1.86,
      "learning_rate": 2.6542324246771883e-05,
      "loss": 0.1956,
      "step": 370
    },
    {
      "epoch": 1.88,
      "learning_rate": 2.6901004304160688e-05,
      "loss": 0.1178,
      "step": 375
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.7259684361549497e-05,
      "loss": 0.2018,
      "step": 380
    },
    {
      "epoch": 1.93,
      "learning_rate": 2.761836441893831e-05,
      "loss": 0.1793,
      "step": 385
    },
    {
      "epoch": 1.96,
      "learning_rate": 2.7977044476327118e-05,
      "loss": 0.1478,
      "step": 390
    },
    {
      "epoch": 1.98,
      "learning_rate": 2.833572453371593e-05,
      "loss": 0.1302,
      "step": 395
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9735973597359736,
      "eval_loss": 0.06645132601261139,
      "eval_runtime": 102.5426,
      "eval_samples_per_second": 20.684,
      "eval_steps_per_second": 2.594,
      "step": 398
    },
    {
      "epoch": 2.01,
      "learning_rate": 2.8694404591104736e-05,
      "loss": 0.137,
      "step": 400
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.9053084648493545e-05,
      "loss": 0.1309,
      "step": 405
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.9411764705882354e-05,
      "loss": 0.1257,
      "step": 410
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.9770444763271166e-05,
      "loss": 0.1457,
      "step": 415
    },
    {
      "epoch": 2.11,
      "learning_rate": 3.012912482065997e-05,
      "loss": 0.259,
      "step": 420
    },
    {
      "epoch": 2.14,
      "learning_rate": 3.048780487804878e-05,
      "loss": 0.1984,
      "step": 425
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.084648493543759e-05,
      "loss": 0.1493,
      "step": 430
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.1205164992826405e-05,
      "loss": 0.1391,
      "step": 435
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.156384505021521e-05,
      "loss": 0.1312,
      "step": 440
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.1922525107604016e-05,
      "loss": 0.1164,
      "step": 445
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.228120516499283e-05,
      "loss": 0.1189,
      "step": 450
    },
    {
      "epoch": 2.29,
      "learning_rate": 3.263988522238164e-05,
      "loss": 0.1296,
      "step": 455
    },
    {
      "epoch": 2.31,
      "learning_rate": 3.2998565279770446e-05,
      "loss": 0.078,
      "step": 460
    },
    {
      "epoch": 2.34,
      "learning_rate": 3.335724533715925e-05,
      "loss": 0.0943,
      "step": 465
    },
    {
      "epoch": 2.36,
      "learning_rate": 3.3715925394548064e-05,
      "loss": 0.0932,
      "step": 470
    },
    {
      "epoch": 2.39,
      "learning_rate": 3.4074605451936876e-05,
      "loss": 0.1722,
      "step": 475
    },
    {
      "epoch": 2.41,
      "learning_rate": 3.443328550932569e-05,
      "loss": 0.2213,
      "step": 480
    },
    {
      "epoch": 2.44,
      "learning_rate": 3.479196556671449e-05,
      "loss": 0.1573,
      "step": 485
    },
    {
      "epoch": 2.46,
      "learning_rate": 3.51506456241033e-05,
      "loss": 0.2181,
      "step": 490
    },
    {
      "epoch": 2.49,
      "learning_rate": 3.550932568149211e-05,
      "loss": 0.1229,
      "step": 495
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.5868005738880924e-05,
      "loss": 0.0817,
      "step": 500
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.622668579626973e-05,
      "loss": 0.0617,
      "step": 505
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.6585365853658535e-05,
      "loss": 0.2366,
      "step": 510
    },
    {
      "epoch": 2.59,
      "learning_rate": 3.694404591104735e-05,
      "loss": 0.1142,
      "step": 515
    },
    {
      "epoch": 2.61,
      "learning_rate": 3.730272596843616e-05,
      "loss": 0.1394,
      "step": 520
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.7661406025824965e-05,
      "loss": 0.1163,
      "step": 525
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.802008608321377e-05,
      "loss": 0.1146,
      "step": 530
    },
    {
      "epoch": 2.69,
      "learning_rate": 3.837876614060258e-05,
      "loss": 0.1014,
      "step": 535
    },
    {
      "epoch": 2.71,
      "learning_rate": 3.8737446197991395e-05,
      "loss": 0.0707,
      "step": 540
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.909612625538021e-05,
      "loss": 0.1651,
      "step": 545
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.945480631276901e-05,
      "loss": 0.3627,
      "step": 550
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.981348637015782e-05,
      "loss": 0.1615,
      "step": 555
    },
    {
      "epoch": 2.81,
      "learning_rate": 4.017216642754663e-05,
      "loss": 0.2595,
      "step": 560
    },
    {
      "epoch": 2.84,
      "learning_rate": 4.053084648493544e-05,
      "loss": 0.1507,
      "step": 565
    },
    {
      "epoch": 2.86,
      "learning_rate": 4.088952654232425e-05,
      "loss": 0.1336,
      "step": 570
    },
    {
      "epoch": 2.89,
      "learning_rate": 4.1248206599713054e-05,
      "loss": 0.093,
      "step": 575
    },
    {
      "epoch": 2.91,
      "learning_rate": 4.160688665710187e-05,
      "loss": 0.1895,
      "step": 580
    },
    {
      "epoch": 2.94,
      "learning_rate": 4.196556671449068e-05,
      "loss": 0.1971,
      "step": 585
    },
    {
      "epoch": 2.96,
      "learning_rate": 4.2324246771879484e-05,
      "loss": 0.1536,
      "step": 590
    },
    {
      "epoch": 2.99,
      "learning_rate": 4.26829268292683e-05,
      "loss": 0.0784,
      "step": 595
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9778406412069779,
      "eval_loss": 0.05999346077442169,
      "eval_runtime": 102.3498,
      "eval_samples_per_second": 20.723,
      "eval_steps_per_second": 2.599,
      "step": 597
    },
    {
      "epoch": 3.02,
      "learning_rate": 4.30416068866571e-05,
      "loss": 0.1106,
      "step": 600
    },
    {
      "epoch": 3.04,
      "learning_rate": 4.3400286944045915e-05,
      "loss": 0.1983,
      "step": 605
    },
    {
      "epoch": 3.07,
      "learning_rate": 4.375896700143472e-05,
      "loss": 0.1241,
      "step": 610
    },
    {
      "epoch": 3.09,
      "learning_rate": 4.411764705882353e-05,
      "loss": 0.1141,
      "step": 615
    },
    {
      "epoch": 3.12,
      "learning_rate": 4.447632711621234e-05,
      "loss": 0.1416,
      "step": 620
    },
    {
      "epoch": 3.14,
      "learning_rate": 4.483500717360115e-05,
      "loss": 0.1494,
      "step": 625
    },
    {
      "epoch": 3.17,
      "learning_rate": 4.519368723098996e-05,
      "loss": 0.0762,
      "step": 630
    },
    {
      "epoch": 3.19,
      "learning_rate": 4.555236728837877e-05,
      "loss": 0.1448,
      "step": 635
    },
    {
      "epoch": 3.22,
      "learning_rate": 4.591104734576758e-05,
      "loss": 0.1665,
      "step": 640
    },
    {
      "epoch": 3.24,
      "learning_rate": 4.6269727403156386e-05,
      "loss": 0.0737,
      "step": 645
    },
    {
      "epoch": 3.27,
      "learning_rate": 4.66284074605452e-05,
      "loss": 0.0924,
      "step": 650
    },
    {
      "epoch": 3.29,
      "learning_rate": 4.6987087517934004e-05,
      "loss": 0.1181,
      "step": 655
    },
    {
      "epoch": 3.32,
      "learning_rate": 4.7345767575322816e-05,
      "loss": 0.2936,
      "step": 660
    },
    {
      "epoch": 3.34,
      "learning_rate": 4.770444763271162e-05,
      "loss": 0.1374,
      "step": 665
    },
    {
      "epoch": 3.37,
      "learning_rate": 4.8063127690100434e-05,
      "loss": 0.1504,
      "step": 670
    },
    {
      "epoch": 3.39,
      "learning_rate": 4.842180774748924e-05,
      "loss": 0.2001,
      "step": 675
    },
    {
      "epoch": 3.42,
      "learning_rate": 4.878048780487805e-05,
      "loss": 0.0628,
      "step": 680
    },
    {
      "epoch": 3.44,
      "learning_rate": 4.9139167862266864e-05,
      "loss": 0.116,
      "step": 685
    },
    {
      "epoch": 3.47,
      "learning_rate": 4.949784791965567e-05,
      "loss": 0.1345,
      "step": 690
    },
    {
      "epoch": 3.49,
      "learning_rate": 4.9856527977044475e-05,
      "loss": 0.119,
      "step": 695
    },
    {
      "epoch": 3.52,
      "learning_rate": 4.978448275862069e-05,
      "loss": 0.0882,
      "step": 700
    },
    {
      "epoch": 3.54,
      "learning_rate": 4.9425287356321845e-05,
      "loss": 0.0853,
      "step": 705
    },
    {
      "epoch": 3.57,
      "learning_rate": 4.906609195402299e-05,
      "loss": 0.2363,
      "step": 710
    },
    {
      "epoch": 3.59,
      "learning_rate": 4.870689655172414e-05,
      "loss": 0.1878,
      "step": 715
    },
    {
      "epoch": 3.62,
      "learning_rate": 4.834770114942529e-05,
      "loss": 0.0954,
      "step": 720
    },
    {
      "epoch": 3.64,
      "learning_rate": 4.798850574712644e-05,
      "loss": 0.1891,
      "step": 725
    },
    {
      "epoch": 3.67,
      "learning_rate": 4.762931034482759e-05,
      "loss": 0.1562,
      "step": 730
    },
    {
      "epoch": 3.69,
      "learning_rate": 4.7270114942528734e-05,
      "loss": 0.1207,
      "step": 735
    },
    {
      "epoch": 3.72,
      "learning_rate": 4.6910919540229886e-05,
      "loss": 0.096,
      "step": 740
    },
    {
      "epoch": 3.74,
      "learning_rate": 4.655172413793104e-05,
      "loss": 0.1072,
      "step": 745
    },
    {
      "epoch": 3.77,
      "learning_rate": 4.619252873563218e-05,
      "loss": 0.0718,
      "step": 750
    },
    {
      "epoch": 3.79,
      "learning_rate": 4.5833333333333334e-05,
      "loss": 0.0622,
      "step": 755
    },
    {
      "epoch": 3.82,
      "learning_rate": 4.5474137931034485e-05,
      "loss": 0.1661,
      "step": 760
    },
    {
      "epoch": 3.84,
      "learning_rate": 4.511494252873563e-05,
      "loss": 0.1559,
      "step": 765
    },
    {
      "epoch": 3.87,
      "learning_rate": 4.475574712643678e-05,
      "loss": 0.1195,
      "step": 770
    },
    {
      "epoch": 3.89,
      "learning_rate": 4.4396551724137933e-05,
      "loss": 0.105,
      "step": 775
    },
    {
      "epoch": 3.92,
      "learning_rate": 4.4037356321839085e-05,
      "loss": 0.0764,
      "step": 780
    },
    {
      "epoch": 3.94,
      "learning_rate": 4.367816091954024e-05,
      "loss": 0.1322,
      "step": 785
    },
    {
      "epoch": 3.97,
      "learning_rate": 4.331896551724138e-05,
      "loss": 0.0829,
      "step": 790
    },
    {
      "epoch": 3.99,
      "learning_rate": 4.295977011494253e-05,
      "loss": 0.1181,
      "step": 795
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9849127769919849,
      "eval_loss": 0.04494262859225273,
      "eval_runtime": 102.4081,
      "eval_samples_per_second": 20.711,
      "eval_steps_per_second": 2.597,
      "step": 796
    },
    {
      "epoch": 4.02,
      "learning_rate": 4.2600574712643685e-05,
      "loss": 0.1042,
      "step": 800
    },
    {
      "epoch": 4.05,
      "learning_rate": 4.224137931034483e-05,
      "loss": 0.0089,
      "step": 805
    },
    {
      "epoch": 4.07,
      "learning_rate": 4.188218390804598e-05,
      "loss": 0.1135,
      "step": 810
    },
    {
      "epoch": 4.1,
      "learning_rate": 4.1522988505747126e-05,
      "loss": 0.0473,
      "step": 815
    },
    {
      "epoch": 4.12,
      "learning_rate": 4.116379310344828e-05,
      "loss": 0.2587,
      "step": 820
    },
    {
      "epoch": 4.15,
      "learning_rate": 4.080459770114943e-05,
      "loss": 0.1912,
      "step": 825
    },
    {
      "epoch": 4.17,
      "learning_rate": 4.0445402298850574e-05,
      "loss": 0.1441,
      "step": 830
    },
    {
      "epoch": 4.2,
      "learning_rate": 4.0086206896551726e-05,
      "loss": 0.0334,
      "step": 835
    },
    {
      "epoch": 4.22,
      "learning_rate": 3.972701149425288e-05,
      "loss": 0.0537,
      "step": 840
    },
    {
      "epoch": 4.25,
      "learning_rate": 3.936781609195402e-05,
      "loss": 0.1477,
      "step": 845
    },
    {
      "epoch": 4.27,
      "learning_rate": 3.9008620689655174e-05,
      "loss": 0.0594,
      "step": 850
    },
    {
      "epoch": 4.3,
      "learning_rate": 3.8649425287356325e-05,
      "loss": 0.0833,
      "step": 855
    },
    {
      "epoch": 4.32,
      "learning_rate": 3.829022988505747e-05,
      "loss": 0.1532,
      "step": 860
    },
    {
      "epoch": 4.35,
      "learning_rate": 3.793103448275862e-05,
      "loss": 0.1087,
      "step": 865
    },
    {
      "epoch": 4.37,
      "learning_rate": 3.7571839080459766e-05,
      "loss": 0.1735,
      "step": 870
    },
    {
      "epoch": 4.4,
      "learning_rate": 3.721264367816092e-05,
      "loss": 0.1434,
      "step": 875
    },
    {
      "epoch": 4.42,
      "learning_rate": 3.685344827586207e-05,
      "loss": 0.1227,
      "step": 880
    },
    {
      "epoch": 4.45,
      "learning_rate": 3.649425287356322e-05,
      "loss": 0.1714,
      "step": 885
    },
    {
      "epoch": 4.47,
      "learning_rate": 3.613505747126437e-05,
      "loss": 0.1588,
      "step": 890
    },
    {
      "epoch": 4.5,
      "learning_rate": 3.5775862068965524e-05,
      "loss": 0.1283,
      "step": 895
    },
    {
      "epoch": 4.52,
      "learning_rate": 3.541666666666667e-05,
      "loss": 0.0578,
      "step": 900
    },
    {
      "epoch": 4.55,
      "learning_rate": 3.505747126436782e-05,
      "loss": 0.119,
      "step": 905
    },
    {
      "epoch": 4.57,
      "learning_rate": 3.4698275862068966e-05,
      "loss": 0.0986,
      "step": 910
    },
    {
      "epoch": 4.6,
      "learning_rate": 3.433908045977012e-05,
      "loss": 0.0577,
      "step": 915
    },
    {
      "epoch": 4.62,
      "learning_rate": 3.397988505747127e-05,
      "loss": 0.0148,
      "step": 920
    },
    {
      "epoch": 4.65,
      "learning_rate": 3.3620689655172414e-05,
      "loss": 0.0293,
      "step": 925
    },
    {
      "epoch": 4.67,
      "learning_rate": 3.3261494252873565e-05,
      "loss": 0.0276,
      "step": 930
    },
    {
      "epoch": 4.7,
      "learning_rate": 3.290229885057472e-05,
      "loss": 0.0033,
      "step": 935
    },
    {
      "epoch": 4.72,
      "learning_rate": 3.254310344827586e-05,
      "loss": 0.0977,
      "step": 940
    },
    {
      "epoch": 4.75,
      "learning_rate": 3.218390804597701e-05,
      "loss": 0.0749,
      "step": 945
    },
    {
      "epoch": 4.77,
      "learning_rate": 3.1824712643678165e-05,
      "loss": 0.0323,
      "step": 950
    },
    {
      "epoch": 4.8,
      "learning_rate": 3.146551724137931e-05,
      "loss": 0.1322,
      "step": 955
    },
    {
      "epoch": 4.82,
      "learning_rate": 3.110632183908046e-05,
      "loss": 0.081,
      "step": 960
    },
    {
      "epoch": 4.85,
      "learning_rate": 3.0747126436781606e-05,
      "loss": 0.1045,
      "step": 965
    },
    {
      "epoch": 4.87,
      "learning_rate": 3.0387931034482758e-05,
      "loss": 0.1323,
      "step": 970
    },
    {
      "epoch": 4.9,
      "learning_rate": 3.0028735632183906e-05,
      "loss": 0.2384,
      "step": 975
    },
    {
      "epoch": 4.92,
      "learning_rate": 2.9669540229885058e-05,
      "loss": 0.0663,
      "step": 980
    },
    {
      "epoch": 4.95,
      "learning_rate": 2.9310344827586206e-05,
      "loss": 0.0836,
      "step": 985
    },
    {
      "epoch": 4.97,
      "learning_rate": 2.8951149425287354e-05,
      "loss": 0.1856,
      "step": 990
    },
    {
      "epoch": 5.0,
      "learning_rate": 2.859195402298851e-05,
      "loss": 0.208,
      "step": 995
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.9886845827439887,
      "eval_loss": 0.03925405070185661,
      "eval_runtime": 102.5066,
      "eval_samples_per_second": 20.691,
      "eval_steps_per_second": 2.595,
      "step": 995
    },
    {
      "epoch": 5.03,
      "learning_rate": 2.8232758620689657e-05,
      "loss": 0.0379,
      "step": 1000
    },
    {
      "epoch": 5.05,
      "learning_rate": 2.787356321839081e-05,
      "loss": 0.1243,
      "step": 1005
    },
    {
      "epoch": 5.08,
      "learning_rate": 2.7514367816091957e-05,
      "loss": 0.0983,
      "step": 1010
    },
    {
      "epoch": 5.1,
      "learning_rate": 2.7155172413793105e-05,
      "loss": 0.0448,
      "step": 1015
    },
    {
      "epoch": 5.13,
      "learning_rate": 2.6795977011494257e-05,
      "loss": 0.1149,
      "step": 1020
    },
    {
      "epoch": 5.15,
      "learning_rate": 2.6436781609195405e-05,
      "loss": 0.0619,
      "step": 1025
    },
    {
      "epoch": 5.18,
      "learning_rate": 2.6077586206896553e-05,
      "loss": 0.1441,
      "step": 1030
    },
    {
      "epoch": 5.2,
      "learning_rate": 2.57183908045977e-05,
      "loss": 0.0575,
      "step": 1035
    },
    {
      "epoch": 5.23,
      "learning_rate": 2.5359195402298853e-05,
      "loss": 0.067,
      "step": 1040
    },
    {
      "epoch": 5.25,
      "learning_rate": 2.5e-05,
      "loss": 0.0616,
      "step": 1045
    },
    {
      "epoch": 5.28,
      "learning_rate": 2.464080459770115e-05,
      "loss": 0.0797,
      "step": 1050
    },
    {
      "epoch": 5.3,
      "learning_rate": 2.42816091954023e-05,
      "loss": 0.043,
      "step": 1055
    },
    {
      "epoch": 5.33,
      "learning_rate": 2.392241379310345e-05,
      "loss": 0.0809,
      "step": 1060
    },
    {
      "epoch": 5.35,
      "learning_rate": 2.3563218390804597e-05,
      "loss": 0.0548,
      "step": 1065
    },
    {
      "epoch": 5.38,
      "learning_rate": 2.3204022988505746e-05,
      "loss": 0.0662,
      "step": 1070
    },
    {
      "epoch": 5.4,
      "learning_rate": 2.2844827586206897e-05,
      "loss": 0.1148,
      "step": 1075
    },
    {
      "epoch": 5.43,
      "learning_rate": 2.248563218390805e-05,
      "loss": 0.0548,
      "step": 1080
    },
    {
      "epoch": 5.45,
      "learning_rate": 2.2126436781609197e-05,
      "loss": 0.0658,
      "step": 1085
    },
    {
      "epoch": 5.48,
      "learning_rate": 2.1767241379310345e-05,
      "loss": 0.0686,
      "step": 1090
    },
    {
      "epoch": 5.5,
      "learning_rate": 2.1408045977011497e-05,
      "loss": 0.1114,
      "step": 1095
    },
    {
      "epoch": 5.53,
      "learning_rate": 2.1048850574712645e-05,
      "loss": 0.0262,
      "step": 1100
    },
    {
      "epoch": 5.55,
      "learning_rate": 2.0689655172413793e-05,
      "loss": 0.0745,
      "step": 1105
    },
    {
      "epoch": 5.58,
      "learning_rate": 2.033045977011494e-05,
      "loss": 0.1139,
      "step": 1110
    },
    {
      "epoch": 5.6,
      "learning_rate": 1.9971264367816093e-05,
      "loss": 0.1325,
      "step": 1115
    },
    {
      "epoch": 5.63,
      "learning_rate": 1.961206896551724e-05,
      "loss": 0.0565,
      "step": 1120
    },
    {
      "epoch": 5.65,
      "learning_rate": 1.925287356321839e-05,
      "loss": 0.0274,
      "step": 1125
    },
    {
      "epoch": 5.68,
      "learning_rate": 1.889367816091954e-05,
      "loss": 0.1295,
      "step": 1130
    },
    {
      "epoch": 5.7,
      "learning_rate": 1.8534482758620693e-05,
      "loss": 0.0501,
      "step": 1135
    },
    {
      "epoch": 5.73,
      "learning_rate": 1.817528735632184e-05,
      "loss": 0.0586,
      "step": 1140
    },
    {
      "epoch": 5.75,
      "learning_rate": 1.781609195402299e-05,
      "loss": 0.045,
      "step": 1145
    },
    {
      "epoch": 5.78,
      "learning_rate": 1.7456896551724137e-05,
      "loss": 0.0226,
      "step": 1150
    },
    {
      "epoch": 5.8,
      "learning_rate": 1.709770114942529e-05,
      "loss": 0.0166,
      "step": 1155
    },
    {
      "epoch": 5.83,
      "learning_rate": 1.6738505747126437e-05,
      "loss": 0.1241,
      "step": 1160
    },
    {
      "epoch": 5.85,
      "learning_rate": 1.6379310344827585e-05,
      "loss": 0.0243,
      "step": 1165
    },
    {
      "epoch": 5.88,
      "learning_rate": 1.6020114942528737e-05,
      "loss": 0.0681,
      "step": 1170
    },
    {
      "epoch": 5.9,
      "learning_rate": 1.5660919540229885e-05,
      "loss": 0.0099,
      "step": 1175
    },
    {
      "epoch": 5.93,
      "learning_rate": 1.5301724137931033e-05,
      "loss": 0.0196,
      "step": 1180
    },
    {
      "epoch": 5.95,
      "learning_rate": 1.4942528735632185e-05,
      "loss": 0.1818,
      "step": 1185
    },
    {
      "epoch": 5.98,
      "learning_rate": 1.4583333333333335e-05,
      "loss": 0.0057,
      "step": 1190
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.9943422913719944,
      "eval_loss": 0.022870095446705818,
      "eval_runtime": 102.5148,
      "eval_samples_per_second": 20.69,
      "eval_steps_per_second": 2.595,
      "step": 1194
    },
    {
      "epoch": 6.01,
      "learning_rate": 1.4224137931034485e-05,
      "loss": 0.0652,
      "step": 1195
    },
    {
      "epoch": 6.03,
      "learning_rate": 1.3864942528735633e-05,
      "loss": 0.0672,
      "step": 1200
    },
    {
      "epoch": 6.06,
      "learning_rate": 1.3505747126436783e-05,
      "loss": 0.0322,
      "step": 1205
    },
    {
      "epoch": 6.08,
      "learning_rate": 1.3146551724137931e-05,
      "loss": 0.0374,
      "step": 1210
    },
    {
      "epoch": 6.11,
      "learning_rate": 1.2787356321839081e-05,
      "loss": 0.0112,
      "step": 1215
    },
    {
      "epoch": 6.13,
      "learning_rate": 1.242816091954023e-05,
      "loss": 0.0284,
      "step": 1220
    },
    {
      "epoch": 6.16,
      "learning_rate": 1.206896551724138e-05,
      "loss": 0.0688,
      "step": 1225
    },
    {
      "epoch": 6.18,
      "learning_rate": 1.1709770114942529e-05,
      "loss": 0.0271,
      "step": 1230
    },
    {
      "epoch": 6.21,
      "learning_rate": 1.1350574712643679e-05,
      "loss": 0.0048,
      "step": 1235
    },
    {
      "epoch": 6.23,
      "learning_rate": 1.0991379310344827e-05,
      "loss": 0.0699,
      "step": 1240
    },
    {
      "epoch": 6.26,
      "learning_rate": 1.0632183908045977e-05,
      "loss": 0.0951,
      "step": 1245
    },
    {
      "epoch": 6.28,
      "learning_rate": 1.0272988505747127e-05,
      "loss": 0.0334,
      "step": 1250
    },
    {
      "epoch": 6.31,
      "learning_rate": 9.913793103448277e-06,
      "loss": 0.061,
      "step": 1255
    },
    {
      "epoch": 6.33,
      "learning_rate": 9.554597701149425e-06,
      "loss": 0.0025,
      "step": 1260
    },
    {
      "epoch": 6.36,
      "learning_rate": 9.195402298850575e-06,
      "loss": 0.0532,
      "step": 1265
    },
    {
      "epoch": 6.38,
      "learning_rate": 8.836206896551725e-06,
      "loss": 0.0359,
      "step": 1270
    },
    {
      "epoch": 6.41,
      "learning_rate": 8.477011494252873e-06,
      "loss": 0.0034,
      "step": 1275
    },
    {
      "epoch": 6.43,
      "learning_rate": 8.117816091954025e-06,
      "loss": 0.0309,
      "step": 1280
    },
    {
      "epoch": 6.46,
      "learning_rate": 7.758620689655173e-06,
      "loss": 0.012,
      "step": 1285
    },
    {
      "epoch": 6.48,
      "learning_rate": 7.399425287356322e-06,
      "loss": 0.0651,
      "step": 1290
    },
    {
      "epoch": 6.51,
      "learning_rate": 7.040229885057471e-06,
      "loss": 0.0261,
      "step": 1295
    },
    {
      "epoch": 6.53,
      "learning_rate": 6.68103448275862e-06,
      "loss": 0.0366,
      "step": 1300
    },
    {
      "epoch": 6.56,
      "learning_rate": 6.321839080459771e-06,
      "loss": 0.0095,
      "step": 1305
    },
    {
      "epoch": 6.58,
      "learning_rate": 5.96264367816092e-06,
      "loss": 0.0244,
      "step": 1310
    },
    {
      "epoch": 6.61,
      "learning_rate": 5.603448275862069e-06,
      "loss": 0.0133,
      "step": 1315
    },
    {
      "epoch": 6.63,
      "learning_rate": 5.244252873563219e-06,
      "loss": 0.0611,
      "step": 1320
    },
    {
      "epoch": 6.66,
      "learning_rate": 4.885057471264369e-06,
      "loss": 0.0558,
      "step": 1325
    },
    {
      "epoch": 6.68,
      "learning_rate": 4.525862068965518e-06,
      "loss": 0.0074,
      "step": 1330
    },
    {
      "epoch": 6.71,
      "learning_rate": 4.166666666666667e-06,
      "loss": 0.0219,
      "step": 1335
    },
    {
      "epoch": 6.73,
      "learning_rate": 3.8074712643678163e-06,
      "loss": 0.0106,
      "step": 1340
    },
    {
      "epoch": 6.76,
      "learning_rate": 3.448275862068966e-06,
      "loss": 0.0009,
      "step": 1345
    },
    {
      "epoch": 6.78,
      "learning_rate": 3.0890804597701153e-06,
      "loss": 0.0028,
      "step": 1350
    },
    {
      "epoch": 6.81,
      "learning_rate": 2.729885057471265e-06,
      "loss": 0.018,
      "step": 1355
    },
    {
      "epoch": 6.83,
      "learning_rate": 2.370689655172414e-06,
      "loss": 0.0383,
      "step": 1360
    },
    {
      "epoch": 6.86,
      "learning_rate": 2.0114942528735633e-06,
      "loss": 0.0463,
      "step": 1365
    },
    {
      "epoch": 6.88,
      "learning_rate": 1.6522988505747128e-06,
      "loss": 0.0113,
      "step": 1370
    },
    {
      "epoch": 6.91,
      "learning_rate": 1.293103448275862e-06,
      "loss": 0.0684,
      "step": 1375
    },
    {
      "epoch": 6.93,
      "learning_rate": 9.339080459770116e-07,
      "loss": 0.0319,
      "step": 1380
    },
    {
      "epoch": 6.96,
      "learning_rate": 5.747126436781609e-07,
      "loss": 0.0689,
      "step": 1385
    },
    {
      "epoch": 6.98,
      "learning_rate": 2.1551724137931036e-07,
      "loss": 0.0017,
      "step": 1390
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.9938708156529938,
      "eval_loss": 0.026258554309606552,
      "eval_runtime": 102.4925,
      "eval_samples_per_second": 20.694,
      "eval_steps_per_second": 2.595,
      "step": 1393
    },
    {
      "epoch": 7.0,
      "step": 1393,
      "total_flos": 1.0256650917539217e+19,
      "train_loss": 0.15325962240578392,
      "train_runtime": 8636.3959,
      "train_samples_per_second": 5.157,
      "train_steps_per_second": 0.161
    }
  ],
  "max_steps": 1393,
  "num_train_epochs": 7,
  "total_flos": 1.0256650917539217e+19,
  "trial_name": null,
  "trial_params": null
}