{
  "best_metric": 0.9938708156529938,
  "best_model_checkpoint": "beit-large-patch16-224-finetuned-BreastCancer-Classification-BreakHis-AH-60-20-20/checkpoint-2388",
  "epoch": 12.0,
  "global_step": 2388,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.03,
      "learning_rate": 1.1627906976744186e-08,
      "loss": 0.7335,
      "step": 5
    },
    {
      "epoch": 0.05,
      "learning_rate": 2.3255813953488372e-08,
      "loss": 0.7745,
      "step": 10
    },
    {
      "epoch": 0.08,
      "learning_rate": 3.488372093023256e-08,
      "loss": 0.7823,
      "step": 15
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.6511627906976744e-08,
      "loss": 0.7606,
      "step": 20
    },
    {
      "epoch": 0.13,
      "learning_rate": 5.8139534883720935e-08,
      "loss": 0.7698,
      "step": 25
    },
    {
      "epoch": 0.15,
      "learning_rate": 6.976744186046512e-08,
      "loss": 0.7576,
      "step": 30
    },
    {
      "epoch": 0.18,
      "learning_rate": 8.139534883720931e-08,
      "loss": 0.7102,
      "step": 35
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.302325581395349e-08,
      "loss": 0.7435,
      "step": 40
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.0465116279069768e-07,
      "loss": 0.7648,
      "step": 45
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.1627906976744187e-07,
      "loss": 0.7704,
      "step": 50
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.2790697674418605e-07,
      "loss": 0.688,
      "step": 55
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.3953488372093024e-07,
      "loss": 0.7343,
      "step": 60
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.5116279069767443e-07,
      "loss": 0.7055,
      "step": 65
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.6279069767441862e-07,
      "loss": 0.7237,
      "step": 70
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.744186046511628e-07,
      "loss": 0.7374,
      "step": 75
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.8604651162790698e-07,
      "loss": 0.653,
      "step": 80
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.9767441860465117e-07,
      "loss": 0.6931,
      "step": 85
    },
    {
      "epoch": 0.45,
      "learning_rate": 2.0930232558139536e-07,
      "loss": 0.638,
      "step": 90
    },
    {
      "epoch": 0.48,
      "learning_rate": 2.2093023255813955e-07,
      "loss": 0.6868,
      "step": 95
    },
    {
      "epoch": 0.5,
      "learning_rate": 2.3255813953488374e-07,
      "loss": 0.716,
      "step": 100
    },
    {
      "epoch": 0.53,
      "learning_rate": 2.4418604651162793e-07,
      "loss": 0.725,
      "step": 105
    },
    {
      "epoch": 0.55,
      "learning_rate": 2.558139534883721e-07,
      "loss": 0.6195,
      "step": 110
    },
    {
      "epoch": 0.58,
      "learning_rate": 2.674418604651163e-07,
      "loss": 0.5791,
      "step": 115
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.790697674418605e-07,
      "loss": 0.6498,
      "step": 120
    },
    {
      "epoch": 0.63,
      "learning_rate": 2.906976744186047e-07,
      "loss": 0.6431,
      "step": 125
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.0232558139534886e-07,
      "loss": 0.5811,
      "step": 130
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.139534883720931e-07,
      "loss": 0.5803,
      "step": 135
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.2558139534883724e-07,
      "loss": 0.5635,
      "step": 140
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.3720930232558146e-07,
      "loss": 0.5369,
      "step": 145
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.488372093023256e-07,
      "loss": 0.5395,
      "step": 150
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.6046511627906984e-07,
      "loss": 0.5645,
      "step": 155
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.7209302325581396e-07,
      "loss": 0.5327,
      "step": 160
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.837209302325582e-07,
      "loss": 0.5035,
      "step": 165
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.9534883720930234e-07,
      "loss": 0.5256,
      "step": 170
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.0697674418604655e-07,
      "loss": 0.5225,
      "step": 175
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.186046511627907e-07,
      "loss": 0.5197,
      "step": 180
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.3023255813953494e-07,
      "loss": 0.5052,
      "step": 185
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.418604651162791e-07,
      "loss": 0.4875,
      "step": 190
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.534883720930233e-07,
      "loss": 0.46,
      "step": 195
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8481848184818482,
      "eval_loss": 0.39499393105506897,
      "eval_runtime": 90.9996,
      "eval_samples_per_second": 23.308,
      "eval_steps_per_second": 1.462,
      "step": 199
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.651162790697675e-07,
      "loss": 0.5344,
      "step": 200
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.767441860465117e-07,
      "loss": 0.4448,
      "step": 205
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.883720930232559e-07,
      "loss": 0.5011,
      "step": 210
    },
    {
      "epoch": 1.08,
      "learning_rate": 5.000000000000001e-07,
      "loss": 0.4055,
      "step": 215
    },
    {
      "epoch": 1.11,
      "learning_rate": 5.116279069767442e-07,
      "loss": 0.4714,
      "step": 220
    },
    {
      "epoch": 1.13,
      "learning_rate": 5.232558139534884e-07,
      "loss": 0.4132,
      "step": 225
    },
    {
      "epoch": 1.16,
      "learning_rate": 5.348837209302326e-07,
      "loss": 0.4394,
      "step": 230
    },
    {
      "epoch": 1.18,
      "learning_rate": 5.465116279069768e-07,
      "loss": 0.3649,
      "step": 235
    },
    {
      "epoch": 1.21,
      "learning_rate": 5.58139534883721e-07,
      "loss": 0.3948,
      "step": 240
    },
    {
      "epoch": 1.23,
      "learning_rate": 5.697674418604652e-07,
      "loss": 0.4236,
      "step": 245
    },
    {
      "epoch": 1.26,
      "learning_rate": 5.813953488372094e-07,
      "loss": 0.4115,
      "step": 250
    },
    {
      "epoch": 1.28,
      "learning_rate": 5.930232558139536e-07,
      "loss": 0.465,
      "step": 255
    },
    {
      "epoch": 1.31,
      "learning_rate": 6.046511627906977e-07,
      "loss": 0.3335,
      "step": 260
    },
    {
      "epoch": 1.33,
      "learning_rate": 6.162790697674419e-07,
      "loss": 0.3746,
      "step": 265
    },
    {
      "epoch": 1.36,
      "learning_rate": 6.279069767441862e-07,
      "loss": 0.3266,
      "step": 270
    },
    {
      "epoch": 1.38,
      "learning_rate": 6.395348837209303e-07,
      "loss": 0.3936,
      "step": 275
    },
    {
      "epoch": 1.41,
      "learning_rate": 6.511627906976745e-07,
      "loss": 0.3193,
      "step": 280
    },
    {
      "epoch": 1.43,
      "learning_rate": 6.627906976744186e-07,
      "loss": 0.3624,
      "step": 285
    },
    {
      "epoch": 1.46,
      "learning_rate": 6.744186046511629e-07,
      "loss": 0.303,
      "step": 290
    },
    {
      "epoch": 1.48,
      "learning_rate": 6.86046511627907e-07,
      "loss": 0.3427,
      "step": 295
    },
    {
      "epoch": 1.51,
      "learning_rate": 6.976744186046513e-07,
      "loss": 0.2622,
      "step": 300
    },
    {
      "epoch": 1.53,
      "learning_rate": 7.093023255813954e-07,
      "loss": 0.2665,
      "step": 305
    },
    {
      "epoch": 1.56,
      "learning_rate": 7.209302325581397e-07,
      "loss": 0.3244,
      "step": 310
    },
    {
      "epoch": 1.58,
      "learning_rate": 7.325581395348838e-07,
      "loss": 0.2232,
      "step": 315
    },
    {
      "epoch": 1.61,
      "learning_rate": 7.441860465116279e-07,
      "loss": 0.2798,
      "step": 320
    },
    {
      "epoch": 1.63,
      "learning_rate": 7.558139534883721e-07,
      "loss": 0.261,
      "step": 325
    },
    {
      "epoch": 1.66,
      "learning_rate": 7.674418604651165e-07,
      "loss": 0.2758,
      "step": 330
    },
    {
      "epoch": 1.68,
      "learning_rate": 7.790697674418606e-07,
      "loss": 0.3212,
      "step": 335
    },
    {
      "epoch": 1.71,
      "learning_rate": 7.906976744186047e-07,
      "loss": 0.2536,
      "step": 340
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.023255813953489e-07,
      "loss": 0.2719,
      "step": 345
    },
    {
      "epoch": 1.76,
      "learning_rate": 8.139534883720931e-07,
      "loss": 0.2754,
      "step": 350
    },
    {
      "epoch": 1.78,
      "learning_rate": 8.255813953488373e-07,
      "loss": 0.187,
      "step": 355
    },
    {
      "epoch": 1.81,
      "learning_rate": 8.372093023255814e-07,
      "loss": 0.2383,
      "step": 360
    },
    {
      "epoch": 1.83,
      "learning_rate": 8.488372093023257e-07,
      "loss": 0.3075,
      "step": 365
    },
    {
      "epoch": 1.86,
      "learning_rate": 8.604651162790699e-07,
      "loss": 0.2802,
      "step": 370
    },
    {
      "epoch": 1.88,
      "learning_rate": 8.720930232558141e-07,
      "loss": 0.2815,
      "step": 375
    },
    {
      "epoch": 1.91,
      "learning_rate": 8.837209302325582e-07,
      "loss": 0.2916,
      "step": 380
    },
    {
      "epoch": 1.93,
      "learning_rate": 8.953488372093023e-07,
      "loss": 0.2769,
      "step": 385
    },
    {
      "epoch": 1.96,
      "learning_rate": 9.069767441860466e-07,
      "loss": 0.2174,
      "step": 390
    },
    {
      "epoch": 1.98,
      "learning_rate": 9.186046511627907e-07,
      "loss": 0.2048,
      "step": 395
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.918906176331919,
      "eval_loss": 0.18860779702663422,
      "eval_runtime": 91.0197,
      "eval_samples_per_second": 23.303,
      "eval_steps_per_second": 1.461,
      "step": 398
    },
    {
      "epoch": 2.01,
      "learning_rate": 9.30232558139535e-07,
      "loss": 0.2442,
      "step": 400
    },
    {
      "epoch": 2.04,
      "learning_rate": 9.418604651162791e-07,
      "loss": 0.2001,
      "step": 405
    },
    {
      "epoch": 2.06,
      "learning_rate": 9.534883720930234e-07,
      "loss": 0.2522,
      "step": 410
    },
    {
      "epoch": 2.09,
      "learning_rate": 9.651162790697675e-07,
      "loss": 0.1624,
      "step": 415
    },
    {
      "epoch": 2.11,
      "learning_rate": 9.767441860465117e-07,
      "loss": 0.2131,
      "step": 420
    },
    {
      "epoch": 2.14,
      "learning_rate": 9.88372093023256e-07,
      "loss": 0.1761,
      "step": 425
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.2182,
      "step": 430
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.0116279069767444e-06,
      "loss": 0.2105,
      "step": 435
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.0232558139534884e-06,
      "loss": 0.2178,
      "step": 440
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.0348837209302326e-06,
      "loss": 0.1795,
      "step": 445
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.0465116279069768e-06,
      "loss": 0.1685,
      "step": 450
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.058139534883721e-06,
      "loss": 0.2405,
      "step": 455
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.0697674418604653e-06,
      "loss": 0.1603,
      "step": 460
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.0813953488372093e-06,
      "loss": 0.2294,
      "step": 465
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.0930232558139537e-06,
      "loss": 0.2107,
      "step": 470
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.1046511627906977e-06,
      "loss": 0.1622,
      "step": 475
    },
    {
      "epoch": 2.41,
      "learning_rate": 1.116279069767442e-06,
      "loss": 0.1739,
      "step": 480
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.1279069767441861e-06,
      "loss": 0.2156,
      "step": 485
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.1395348837209304e-06,
      "loss": 0.2137,
      "step": 490
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.1511627906976746e-06,
      "loss": 0.2177,
      "step": 495
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.1627906976744188e-06,
      "loss": 0.2202,
      "step": 500
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.1744186046511628e-06,
      "loss": 0.1772,
      "step": 505
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.1860465116279072e-06,
      "loss": 0.1435,
      "step": 510
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.1976744186046512e-06,
      "loss": 0.1728,
      "step": 515
    },
    {
      "epoch": 2.61,
      "learning_rate": 1.2093023255813954e-06,
      "loss": 0.1686,
      "step": 520
    },
    {
      "epoch": 2.64,
      "learning_rate": 1.2209302325581397e-06,
      "loss": 0.1815,
      "step": 525
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.2325581395348839e-06,
      "loss": 0.2048,
      "step": 530
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.244186046511628e-06,
      "loss": 0.1522,
      "step": 535
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.2558139534883723e-06,
      "loss": 0.1424,
      "step": 540
    },
    {
      "epoch": 2.74,
      "learning_rate": 1.2674418604651165e-06,
      "loss": 0.1503,
      "step": 545
    },
    {
      "epoch": 2.76,
      "learning_rate": 1.2790697674418605e-06,
      "loss": 0.2181,
      "step": 550
    },
    {
      "epoch": 2.79,
      "learning_rate": 1.2906976744186048e-06,
      "loss": 0.1607,
      "step": 555
    },
    {
      "epoch": 2.81,
      "learning_rate": 1.302325581395349e-06,
      "loss": 0.1129,
      "step": 560
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.313953488372093e-06,
      "loss": 0.1496,
      "step": 565
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.3255813953488372e-06,
      "loss": 0.1261,
      "step": 570
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.3372093023255814e-06,
      "loss": 0.1951,
      "step": 575
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.3488372093023258e-06,
      "loss": 0.2033,
      "step": 580
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.36046511627907e-06,
      "loss": 0.2468,
      "step": 585
    },
    {
      "epoch": 2.96,
      "learning_rate": 1.372093023255814e-06,
      "loss": 0.1552,
      "step": 590
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.3837209302325583e-06,
      "loss": 0.182,
      "step": 595
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.9481376709099482,
      "eval_loss": 0.1382409781217575,
      "eval_runtime": 91.1965,
      "eval_samples_per_second": 23.257,
      "eval_steps_per_second": 1.458,
      "step": 597
    },
    {
      "epoch": 3.02,
      "learning_rate": 1.3953488372093025e-06,
      "loss": 0.1781,
      "step": 600
    },
    {
      "epoch": 3.04,
      "learning_rate": 1.4069767441860465e-06,
      "loss": 0.2034,
      "step": 605
    },
    {
      "epoch": 3.07,
      "learning_rate": 1.4186046511627907e-06,
      "loss": 0.1188,
      "step": 610
    },
    {
      "epoch": 3.09,
      "learning_rate": 1.430232558139535e-06,
      "loss": 0.1816,
      "step": 615
    },
    {
      "epoch": 3.12,
      "learning_rate": 1.4418604651162794e-06,
      "loss": 0.1081,
      "step": 620
    },
    {
      "epoch": 3.14,
      "learning_rate": 1.4534883720930234e-06,
      "loss": 0.1409,
      "step": 625
    },
    {
      "epoch": 3.17,
      "learning_rate": 1.4651162790697676e-06,
      "loss": 0.1834,
      "step": 630
    },
    {
      "epoch": 3.19,
      "learning_rate": 1.4767441860465118e-06,
      "loss": 0.2097,
      "step": 635
    },
    {
      "epoch": 3.22,
      "learning_rate": 1.4883720930232558e-06,
      "loss": 0.145,
      "step": 640
    },
    {
      "epoch": 3.24,
      "learning_rate": 1.5e-06,
      "loss": 0.156,
      "step": 645
    },
    {
      "epoch": 3.27,
      "learning_rate": 1.5116279069767443e-06,
      "loss": 0.1742,
      "step": 650
    },
    {
      "epoch": 3.29,
      "learning_rate": 1.5232558139534885e-06,
      "loss": 0.1477,
      "step": 655
    },
    {
      "epoch": 3.32,
      "learning_rate": 1.534883720930233e-06,
      "loss": 0.0985,
      "step": 660
    },
    {
      "epoch": 3.34,
      "learning_rate": 1.546511627906977e-06,
      "loss": 0.1889,
      "step": 665
    },
    {
      "epoch": 3.37,
      "learning_rate": 1.5581395348837211e-06,
      "loss": 0.1832,
      "step": 670
    },
    {
      "epoch": 3.39,
      "learning_rate": 1.5697674418604653e-06,
      "loss": 0.1582,
      "step": 675
    },
    {
      "epoch": 3.42,
      "learning_rate": 1.5813953488372093e-06,
      "loss": 0.1872,
      "step": 680
    },
    {
      "epoch": 3.44,
      "learning_rate": 1.5930232558139536e-06,
      "loss": 0.1571,
      "step": 685
    },
    {
      "epoch": 3.47,
      "learning_rate": 1.6046511627906978e-06,
      "loss": 0.1407,
      "step": 690
    },
    {
      "epoch": 3.49,
      "learning_rate": 1.6162790697674418e-06,
      "loss": 0.0965,
      "step": 695
    },
    {
      "epoch": 3.52,
      "learning_rate": 1.6279069767441862e-06,
      "loss": 0.0725,
      "step": 700
    },
    {
      "epoch": 3.54,
      "learning_rate": 1.6395348837209304e-06,
      "loss": 0.1101,
      "step": 705
    },
    {
      "epoch": 3.57,
      "learning_rate": 1.6511627906976747e-06,
      "loss": 0.1048,
      "step": 710
    },
    {
      "epoch": 3.59,
      "learning_rate": 1.6627906976744187e-06,
      "loss": 0.1131,
      "step": 715
    },
    {
      "epoch": 3.62,
      "learning_rate": 1.6744186046511629e-06,
      "loss": 0.1339,
      "step": 720
    },
    {
      "epoch": 3.64,
      "learning_rate": 1.686046511627907e-06,
      "loss": 0.1134,
      "step": 725
    },
    {
      "epoch": 3.67,
      "learning_rate": 1.6976744186046513e-06,
      "loss": 0.1316,
      "step": 730
    },
    {
      "epoch": 3.69,
      "learning_rate": 1.7093023255813953e-06,
      "loss": 0.1277,
      "step": 735
    },
    {
      "epoch": 3.72,
      "learning_rate": 1.7209302325581397e-06,
      "loss": 0.1207,
      "step": 740
    },
    {
      "epoch": 3.74,
      "learning_rate": 1.732558139534884e-06,
      "loss": 0.1694,
      "step": 745
    },
    {
      "epoch": 3.77,
      "learning_rate": 1.7441860465116282e-06,
      "loss": 0.1486,
      "step": 750
    },
    {
      "epoch": 3.79,
      "learning_rate": 1.7558139534883722e-06,
      "loss": 0.0856,
      "step": 755
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.7674418604651164e-06,
      "loss": 0.1068,
      "step": 760
    },
    {
      "epoch": 3.84,
      "learning_rate": 1.7790697674418606e-06,
      "loss": 0.0985,
      "step": 765
    },
    {
      "epoch": 3.87,
      "learning_rate": 1.7906976744186046e-06,
      "loss": 0.1445,
      "step": 770
    },
    {
      "epoch": 3.89,
      "learning_rate": 1.8023255813953488e-06,
      "loss": 0.0848,
      "step": 775
    },
    {
      "epoch": 3.92,
      "learning_rate": 1.8139534883720933e-06,
      "loss": 0.1124,
      "step": 780
    },
    {
      "epoch": 3.94,
      "learning_rate": 1.8255813953488375e-06,
      "loss": 0.0806,
      "step": 785
    },
    {
      "epoch": 3.97,
      "learning_rate": 1.8372093023255815e-06,
      "loss": 0.078,
      "step": 790
    },
    {
      "epoch": 3.99,
      "learning_rate": 1.8488372093023257e-06,
      "loss": 0.0826,
      "step": 795
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.9693540782649693,
      "eval_loss": 0.0760357454419136,
      "eval_runtime": 90.9354,
      "eval_samples_per_second": 23.324,
      "eval_steps_per_second": 1.463,
      "step": 796
    },
    {
      "epoch": 4.02,
      "learning_rate": 1.86046511627907e-06,
      "loss": 0.0578,
      "step": 800
    },
    {
      "epoch": 4.05,
      "learning_rate": 1.872093023255814e-06,
      "loss": 0.2264,
      "step": 805
    },
    {
      "epoch": 4.07,
      "learning_rate": 1.8837209302325582e-06,
      "loss": 0.0851,
      "step": 810
    },
    {
      "epoch": 4.1,
      "learning_rate": 1.8953488372093024e-06,
      "loss": 0.0803,
      "step": 815
    },
    {
      "epoch": 4.12,
      "learning_rate": 1.9069767441860468e-06,
      "loss": 0.104,
      "step": 820
    },
    {
      "epoch": 4.15,
      "learning_rate": 1.918604651162791e-06,
      "loss": 0.1087,
      "step": 825
    },
    {
      "epoch": 4.17,
      "learning_rate": 1.930232558139535e-06,
      "loss": 0.1494,
      "step": 830
    },
    {
      "epoch": 4.2,
      "learning_rate": 1.941860465116279e-06,
      "loss": 0.1459,
      "step": 835
    },
    {
      "epoch": 4.22,
      "learning_rate": 1.9534883720930235e-06,
      "loss": 0.0806,
      "step": 840
    },
    {
      "epoch": 4.25,
      "learning_rate": 1.9651162790697675e-06,
      "loss": 0.0517,
      "step": 845
    },
    {
      "epoch": 4.27,
      "learning_rate": 1.976744186046512e-06,
      "loss": 0.1205,
      "step": 850
    },
    {
      "epoch": 4.3,
      "learning_rate": 1.988372093023256e-06,
      "loss": 0.1168,
      "step": 855
    },
    {
      "epoch": 4.32,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.1178,
      "step": 860
    },
    {
      "epoch": 4.35,
      "learning_rate": 2.0116279069767443e-06,
      "loss": 0.0972,
      "step": 865
    },
    {
      "epoch": 4.37,
      "learning_rate": 2.0232558139534888e-06,
      "loss": 0.1218,
      "step": 870
    },
    {
      "epoch": 4.4,
      "learning_rate": 2.0348837209302328e-06,
      "loss": 0.0609,
      "step": 875
    },
    {
      "epoch": 4.42,
      "learning_rate": 2.0465116279069768e-06,
      "loss": 0.0926,
      "step": 880
    },
    {
      "epoch": 4.45,
      "learning_rate": 2.058139534883721e-06,
      "loss": 0.0923,
      "step": 885
    },
    {
      "epoch": 4.47,
      "learning_rate": 2.0697674418604652e-06,
      "loss": 0.0673,
      "step": 890
    },
    {
      "epoch": 4.5,
      "learning_rate": 2.0813953488372092e-06,
      "loss": 0.0915,
      "step": 895
    },
    {
      "epoch": 4.52,
      "learning_rate": 2.0930232558139536e-06,
      "loss": 0.0987,
      "step": 900
    },
    {
      "epoch": 4.55,
      "learning_rate": 2.104651162790698e-06,
      "loss": 0.1189,
      "step": 905
    },
    {
      "epoch": 4.57,
      "learning_rate": 2.116279069767442e-06,
      "loss": 0.0615,
      "step": 910
    },
    {
      "epoch": 4.6,
      "learning_rate": 2.127906976744186e-06,
      "loss": 0.1131,
      "step": 915
    },
    {
      "epoch": 4.62,
      "learning_rate": 2.1395348837209305e-06,
      "loss": 0.1153,
      "step": 920
    },
    {
      "epoch": 4.65,
      "learning_rate": 2.1511627906976745e-06,
      "loss": 0.0741,
      "step": 925
    },
    {
      "epoch": 4.67,
      "learning_rate": 2.1627906976744185e-06,
      "loss": 0.1233,
      "step": 930
    },
    {
      "epoch": 4.7,
      "learning_rate": 2.174418604651163e-06,
      "loss": 0.1245,
      "step": 935
    },
    {
      "epoch": 4.72,
      "learning_rate": 2.1860465116279074e-06,
      "loss": 0.058,
      "step": 940
    },
    {
      "epoch": 4.75,
      "learning_rate": 2.1976744186046514e-06,
      "loss": 0.0409,
      "step": 945
    },
    {
      "epoch": 4.77,
      "learning_rate": 2.2093023255813954e-06,
      "loss": 0.0828,
      "step": 950
    },
    {
      "epoch": 4.8,
      "learning_rate": 2.22093023255814e-06,
      "loss": 0.1098,
      "step": 955
    },
    {
      "epoch": 4.82,
      "learning_rate": 2.232558139534884e-06,
      "loss": 0.0561,
      "step": 960
    },
    {
      "epoch": 4.85,
      "learning_rate": 2.244186046511628e-06,
      "loss": 0.1561,
      "step": 965
    },
    {
      "epoch": 4.87,
      "learning_rate": 2.2558139534883723e-06,
      "loss": 0.1465,
      "step": 970
    },
    {
      "epoch": 4.9,
      "learning_rate": 2.2674418604651163e-06,
      "loss": 0.068,
      "step": 975
    },
    {
      "epoch": 4.92,
      "learning_rate": 2.2790697674418607e-06,
      "loss": 0.1774,
      "step": 980
    },
    {
      "epoch": 4.95,
      "learning_rate": 2.2906976744186047e-06,
      "loss": 0.1152,
      "step": 985
    },
    {
      "epoch": 4.97,
      "learning_rate": 2.302325581395349e-06,
      "loss": 0.0964,
      "step": 990
    },
    {
      "epoch": 5.0,
      "learning_rate": 2.313953488372093e-06,
      "loss": 0.0886,
      "step": 995
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.9787835926449788,
      "eval_loss": 0.059965480118989944,
      "eval_runtime": 90.9509,
      "eval_samples_per_second": 23.32,
      "eval_steps_per_second": 1.462,
      "step": 995
    },
    {
      "epoch": 5.03,
      "learning_rate": 2.3255813953488376e-06,
      "loss": 0.0869,
      "step": 1000
    },
    {
      "epoch": 5.05,
      "learning_rate": 2.3372093023255816e-06,
      "loss": 0.1005,
      "step": 1005
    },
    {
      "epoch": 5.08,
      "learning_rate": 2.3488372093023256e-06,
      "loss": 0.0859,
      "step": 1010
    },
    {
      "epoch": 5.1,
      "learning_rate": 2.36046511627907e-06,
      "loss": 0.0807,
      "step": 1015
    },
    {
      "epoch": 5.13,
      "learning_rate": 2.3720930232558144e-06,
      "loss": 0.0645,
      "step": 1020
    },
    {
      "epoch": 5.15,
      "learning_rate": 2.3837209302325585e-06,
      "loss": 0.0598,
      "step": 1025
    },
    {
      "epoch": 5.18,
      "learning_rate": 2.3953488372093025e-06,
      "loss": 0.1259,
      "step": 1030
    },
    {
      "epoch": 5.2,
      "learning_rate": 2.406976744186047e-06,
      "loss": 0.0914,
      "step": 1035
    },
    {
      "epoch": 5.23,
      "learning_rate": 2.418604651162791e-06,
      "loss": 0.1101,
      "step": 1040
    },
    {
      "epoch": 5.25,
      "learning_rate": 2.430232558139535e-06,
      "loss": 0.1553,
      "step": 1045
    },
    {
      "epoch": 5.28,
      "learning_rate": 2.4418604651162793e-06,
      "loss": 0.0465,
      "step": 1050
    },
    {
      "epoch": 5.3,
      "learning_rate": 2.4534883720930233e-06,
      "loss": 0.0226,
      "step": 1055
    },
    {
      "epoch": 5.33,
      "learning_rate": 2.4651162790697678e-06,
      "loss": 0.089,
      "step": 1060
    },
    {
      "epoch": 5.35,
      "learning_rate": 2.4767441860465118e-06,
      "loss": 0.0578,
      "step": 1065
    },
    {
      "epoch": 5.38,
      "learning_rate": 2.488372093023256e-06,
      "loss": 0.0772,
      "step": 1070
    },
    {
      "epoch": 5.4,
      "learning_rate": 2.5e-06,
      "loss": 0.066,
      "step": 1075
    },
    {
      "epoch": 5.43,
      "learning_rate": 2.5116279069767446e-06,
      "loss": 0.0877,
      "step": 1080
    },
    {
      "epoch": 5.45,
      "learning_rate": 2.5232558139534886e-06,
      "loss": 0.107,
      "step": 1085
    },
    {
      "epoch": 5.48,
      "learning_rate": 2.534883720930233e-06,
      "loss": 0.1334,
      "step": 1090
    },
    {
      "epoch": 5.5,
      "learning_rate": 2.5465116279069767e-06,
      "loss": 0.0577,
      "step": 1095
    },
    {
      "epoch": 5.53,
      "learning_rate": 2.558139534883721e-06,
      "loss": 0.1463,
      "step": 1100
    },
    {
      "epoch": 5.55,
      "learning_rate": 2.569767441860465e-06,
      "loss": 0.097,
      "step": 1105
    },
    {
      "epoch": 5.58,
      "learning_rate": 2.5813953488372095e-06,
      "loss": 0.0964,
      "step": 1110
    },
    {
      "epoch": 5.6,
      "learning_rate": 2.593023255813954e-06,
      "loss": 0.0306,
      "step": 1115
    },
    {
      "epoch": 5.63,
      "learning_rate": 2.604651162790698e-06,
      "loss": 0.1012,
      "step": 1120
    },
    {
      "epoch": 5.65,
      "learning_rate": 2.6162790697674424e-06,
      "loss": 0.1271,
      "step": 1125
    },
    {
      "epoch": 5.68,
      "learning_rate": 2.627906976744186e-06,
      "loss": 0.0381,
      "step": 1130
    },
    {
      "epoch": 5.7,
      "learning_rate": 2.6395348837209304e-06,
      "loss": 0.1062,
      "step": 1135
    },
    {
      "epoch": 5.73,
      "learning_rate": 2.6511627906976744e-06,
      "loss": 0.0798,
      "step": 1140
    },
    {
      "epoch": 5.75,
      "learning_rate": 2.662790697674419e-06,
      "loss": 0.0583,
      "step": 1145
    },
    {
      "epoch": 5.78,
      "learning_rate": 2.674418604651163e-06,
      "loss": 0.1107,
      "step": 1150
    },
    {
      "epoch": 5.8,
      "learning_rate": 2.6860465116279073e-06,
      "loss": 0.0535,
      "step": 1155
    },
    {
      "epoch": 5.83,
      "learning_rate": 2.6976744186046517e-06,
      "loss": 0.1115,
      "step": 1160
    },
    {
      "epoch": 5.85,
      "learning_rate": 2.7093023255813957e-06,
      "loss": 0.0569,
      "step": 1165
    },
    {
      "epoch": 5.88,
      "learning_rate": 2.72093023255814e-06,
      "loss": 0.0927,
      "step": 1170
    },
    {
      "epoch": 5.9,
      "learning_rate": 2.7325581395348837e-06,
      "loss": 0.0326,
      "step": 1175
    },
    {
      "epoch": 5.93,
      "learning_rate": 2.744186046511628e-06,
      "loss": 0.1111,
      "step": 1180
    },
    {
      "epoch": 5.95,
      "learning_rate": 2.755813953488372e-06,
      "loss": 0.0345,
      "step": 1185
    },
    {
      "epoch": 5.98,
      "learning_rate": 2.7674418604651166e-06,
      "loss": 0.0896,
      "step": 1190
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.9801980198019802,
      "eval_loss": 0.05234839394688606,
      "eval_runtime": 91.0139,
      "eval_samples_per_second": 23.304,
      "eval_steps_per_second": 1.461,
      "step": 1194
    },
    {
      "epoch": 6.01,
      "learning_rate": 2.779069767441861e-06,
      "loss": 0.0247,
      "step": 1195
    },
    {
      "epoch": 6.03,
      "learning_rate": 2.790697674418605e-06,
      "loss": 0.06,
      "step": 1200
    },
    {
      "epoch": 6.06,
      "learning_rate": 2.8023255813953494e-06,
      "loss": 0.0695,
      "step": 1205
    },
    {
      "epoch": 6.08,
      "learning_rate": 2.813953488372093e-06,
      "loss": 0.0554,
      "step": 1210
    },
    {
      "epoch": 6.11,
      "learning_rate": 2.8255813953488374e-06,
      "loss": 0.1587,
      "step": 1215
    },
    {
      "epoch": 6.13,
      "learning_rate": 2.8372093023255815e-06,
      "loss": 0.0628,
      "step": 1220
    },
    {
      "epoch": 6.16,
      "learning_rate": 2.848837209302326e-06,
      "loss": 0.073,
      "step": 1225
    },
    {
      "epoch": 6.18,
      "learning_rate": 2.86046511627907e-06,
      "loss": 0.0835,
      "step": 1230
    },
    {
      "epoch": 6.21,
      "learning_rate": 2.8720930232558143e-06,
      "loss": 0.1044,
      "step": 1235
    },
    {
      "epoch": 6.23,
      "learning_rate": 2.8837209302325587e-06,
      "loss": 0.0537,
      "step": 1240
    },
    {
      "epoch": 6.26,
      "learning_rate": 2.8953488372093023e-06,
      "loss": 0.1153,
      "step": 1245
    },
    {
      "epoch": 6.28,
      "learning_rate": 2.9069767441860468e-06,
      "loss": 0.0685,
      "step": 1250
    },
    {
      "epoch": 6.31,
      "learning_rate": 2.9186046511627908e-06,
      "loss": 0.0602,
      "step": 1255
    },
    {
      "epoch": 6.33,
      "learning_rate": 2.930232558139535e-06,
      "loss": 0.0731,
      "step": 1260
    },
    {
      "epoch": 6.36,
      "learning_rate": 2.941860465116279e-06,
      "loss": 0.0509,
      "step": 1265
    },
    {
      "epoch": 6.38,
      "learning_rate": 2.9534883720930236e-06,
      "loss": 0.0426,
      "step": 1270
    },
    {
      "epoch": 6.41,
      "learning_rate": 2.965116279069768e-06,
      "loss": 0.0578,
      "step": 1275
    },
    {
      "epoch": 6.43,
      "learning_rate": 2.9767441860465116e-06,
      "loss": 0.0487,
      "step": 1280
    },
    {
      "epoch": 6.46,
      "learning_rate": 2.988372093023256e-06,
      "loss": 0.0266,
      "step": 1285
    },
    {
      "epoch": 6.48,
      "learning_rate": 3e-06,
      "loss": 0.0566,
      "step": 1290
    },
    {
      "epoch": 6.51,
      "learning_rate": 3.0116279069767445e-06,
      "loss": 0.0443,
      "step": 1295
    },
    {
      "epoch": 6.53,
      "learning_rate": 3.0232558139534885e-06,
      "loss": 0.0457,
      "step": 1300
    },
    {
      "epoch": 6.56,
      "learning_rate": 3.034883720930233e-06,
      "loss": 0.122,
      "step": 1305
    },
    {
      "epoch": 6.58,
      "learning_rate": 3.046511627906977e-06,
      "loss": 0.1308,
      "step": 1310
    },
    {
      "epoch": 6.61,
      "learning_rate": 3.0581395348837214e-06,
      "loss": 0.0771,
      "step": 1315
    },
    {
      "epoch": 6.63,
      "learning_rate": 3.069767441860466e-06,
      "loss": 0.0594,
      "step": 1320
    },
    {
      "epoch": 6.66,
      "learning_rate": 3.0813953488372094e-06,
      "loss": 0.0398,
      "step": 1325
    },
    {
      "epoch": 6.68,
      "learning_rate": 3.093023255813954e-06,
      "loss": 0.0973,
      "step": 1330
    },
    {
      "epoch": 6.71,
      "learning_rate": 3.104651162790698e-06,
      "loss": 0.174,
      "step": 1335
    },
    {
      "epoch": 6.73,
      "learning_rate": 3.1162790697674423e-06,
      "loss": 0.0669,
      "step": 1340
    },
    {
      "epoch": 6.76,
      "learning_rate": 3.1279069767441863e-06,
      "loss": 0.0592,
      "step": 1345
    },
    {
      "epoch": 6.78,
      "learning_rate": 3.1395348837209307e-06,
      "loss": 0.0337,
      "step": 1350
    },
    {
      "epoch": 6.81,
      "learning_rate": 3.151162790697675e-06,
      "loss": 0.0338,
      "step": 1355
    },
    {
      "epoch": 6.83,
      "learning_rate": 3.1627906976744187e-06,
      "loss": 0.0956,
      "step": 1360
    },
    {
      "epoch": 6.86,
      "learning_rate": 3.174418604651163e-06,
      "loss": 0.0236,
      "step": 1365
    },
    {
      "epoch": 6.88,
      "learning_rate": 3.186046511627907e-06,
      "loss": 0.0381,
      "step": 1370
    },
    {
      "epoch": 6.91,
      "learning_rate": 3.1976744186046516e-06,
      "loss": 0.0674,
      "step": 1375
    },
    {
      "epoch": 6.93,
      "learning_rate": 3.2093023255813956e-06,
      "loss": 0.0448,
      "step": 1380
    },
    {
      "epoch": 6.96,
      "learning_rate": 3.22093023255814e-06,
      "loss": 0.0564,
      "step": 1385
    },
    {
      "epoch": 6.98,
      "learning_rate": 3.2325581395348836e-06,
      "loss": 0.0774,
      "step": 1390
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.9825553983969826,
      "eval_loss": 0.04816773906350136,
      "eval_runtime": 90.6447,
      "eval_samples_per_second": 23.399,
      "eval_steps_per_second": 1.467,
      "step": 1393
    },
    {
      "epoch": 7.01,
      "learning_rate": 3.244186046511628e-06,
      "loss": 0.1127,
      "step": 1395
    },
    {
      "epoch": 7.04,
      "learning_rate": 3.2558139534883724e-06,
      "loss": 0.0509,
      "step": 1400
    },
    {
      "epoch": 7.06,
      "learning_rate": 3.2674418604651164e-06,
      "loss": 0.1105,
      "step": 1405
    },
    {
      "epoch": 7.09,
      "learning_rate": 3.279069767441861e-06,
      "loss": 0.0226,
      "step": 1410
    },
    {
      "epoch": 7.11,
      "learning_rate": 3.290697674418605e-06,
      "loss": 0.0518,
      "step": 1415
    },
    {
      "epoch": 7.14,
      "learning_rate": 3.3023255813953493e-06,
      "loss": 0.155,
      "step": 1420
    },
    {
      "epoch": 7.16,
      "learning_rate": 3.313953488372093e-06,
      "loss": 0.0815,
      "step": 1425
    },
    {
      "epoch": 7.19,
      "learning_rate": 3.3255813953488373e-06,
      "loss": 0.0785,
      "step": 1430
    },
    {
      "epoch": 7.21,
      "learning_rate": 3.3372093023255817e-06,
      "loss": 0.0339,
      "step": 1435
    },
    {
      "epoch": 7.24,
      "learning_rate": 3.3488372093023258e-06,
      "loss": 0.0358,
      "step": 1440
    },
    {
      "epoch": 7.26,
      "learning_rate": 3.36046511627907e-06,
      "loss": 0.0259,
      "step": 1445
    },
    {
      "epoch": 7.29,
      "learning_rate": 3.372093023255814e-06,
      "loss": 0.0391,
      "step": 1450
    },
    {
      "epoch": 7.31,
      "learning_rate": 3.3837209302325586e-06,
      "loss": 0.0233,
      "step": 1455
    },
    {
      "epoch": 7.34,
      "learning_rate": 3.3953488372093026e-06,
      "loss": 0.024,
      "step": 1460
    },
    {
      "epoch": 7.36,
      "learning_rate": 3.406976744186047e-06,
      "loss": 0.0789,
      "step": 1465
    },
    {
      "epoch": 7.39,
      "learning_rate": 3.4186046511627906e-06,
      "loss": 0.0406,
      "step": 1470
    },
    {
      "epoch": 7.41,
      "learning_rate": 3.430232558139535e-06,
      "loss": 0.0524,
      "step": 1475
    },
    {
      "epoch": 7.44,
      "learning_rate": 3.4418604651162795e-06,
      "loss": 0.1186,
      "step": 1480
    },
    {
      "epoch": 7.46,
      "learning_rate": 3.4534883720930235e-06,
      "loss": 0.0434,
      "step": 1485
    },
    {
      "epoch": 7.49,
      "learning_rate": 3.465116279069768e-06,
      "loss": 0.084,
      "step": 1490
    },
    {
      "epoch": 7.51,
      "learning_rate": 3.476744186046512e-06,
      "loss": 0.0475,
      "step": 1495
    },
    {
      "epoch": 7.54,
      "learning_rate": 3.4883720930232564e-06,
      "loss": 0.1015,
      "step": 1500
    },
    {
      "epoch": 7.56,
      "learning_rate": 3.5e-06,
      "loss": 0.042,
      "step": 1505
    },
    {
      "epoch": 7.59,
      "learning_rate": 3.5116279069767444e-06,
      "loss": 0.0287,
      "step": 1510
    },
    {
      "epoch": 7.61,
      "learning_rate": 3.523255813953489e-06,
      "loss": 0.064,
      "step": 1515
    },
    {
      "epoch": 7.64,
      "learning_rate": 3.534883720930233e-06,
      "loss": 0.095,
      "step": 1520
    },
    {
      "epoch": 7.66,
      "learning_rate": 3.5465116279069772e-06,
      "loss": 0.0677,
      "step": 1525
    },
    {
      "epoch": 7.69,
      "learning_rate": 3.5581395348837212e-06,
      "loss": 0.05,
      "step": 1530
    },
    {
      "epoch": 7.71,
      "learning_rate": 3.5697674418604657e-06,
      "loss": 0.0435,
      "step": 1535
    },
    {
      "epoch": 7.74,
      "learning_rate": 3.5813953488372093e-06,
      "loss": 0.1378,
      "step": 1540
    },
    {
      "epoch": 7.76,
      "learning_rate": 3.5930232558139537e-06,
      "loss": 0.1013,
      "step": 1545
    },
    {
      "epoch": 7.79,
      "learning_rate": 3.6046511627906977e-06,
      "loss": 0.0508,
      "step": 1550
    },
    {
      "epoch": 7.81,
      "learning_rate": 3.616279069767442e-06,
      "loss": 0.0078,
      "step": 1555
    },
    {
      "epoch": 7.84,
      "learning_rate": 3.6279069767441866e-06,
      "loss": 0.0256,
      "step": 1560
    },
    {
      "epoch": 7.86,
      "learning_rate": 3.6395348837209306e-06,
      "loss": 0.0381,
      "step": 1565
    },
    {
      "epoch": 7.89,
      "learning_rate": 3.651162790697675e-06,
      "loss": 0.0318,
      "step": 1570
    },
    {
      "epoch": 7.91,
      "learning_rate": 3.6627906976744186e-06,
      "loss": 0.1263,
      "step": 1575
    },
    {
      "epoch": 7.94,
      "learning_rate": 3.674418604651163e-06,
      "loss": 0.0309,
      "step": 1580
    },
    {
      "epoch": 7.96,
      "learning_rate": 3.686046511627907e-06,
      "loss": 0.0699,
      "step": 1585
    },
    {
      "epoch": 7.99,
      "learning_rate": 3.6976744186046514e-06,
      "loss": 0.0876,
      "step": 1590
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.9877416313059877,
      "eval_loss": 0.0289468877017498,
      "eval_runtime": 90.5383,
      "eval_samples_per_second": 23.427,
      "eval_steps_per_second": 1.469,
      "step": 1592
    },
    {
      "epoch": 8.02,
      "learning_rate": 3.709302325581396e-06,
      "loss": 0.0537,
      "step": 1595
    },
    {
      "epoch": 8.04,
      "learning_rate": 3.72093023255814e-06,
      "loss": 0.0633,
      "step": 1600
    },
    {
      "epoch": 8.07,
      "learning_rate": 3.7325581395348843e-06,
      "loss": 0.0803,
      "step": 1605
    },
    {
      "epoch": 8.09,
      "learning_rate": 3.744186046511628e-06,
      "loss": 0.0594,
      "step": 1610
    },
    {
      "epoch": 8.12,
      "learning_rate": 3.7558139534883727e-06,
      "loss": 0.0894,
      "step": 1615
    },
    {
      "epoch": 8.14,
      "learning_rate": 3.7674418604651163e-06,
      "loss": 0.0336,
      "step": 1620
    },
    {
      "epoch": 8.17,
      "learning_rate": 3.7790697674418607e-06,
      "loss": 0.0151,
      "step": 1625
    },
    {
      "epoch": 8.19,
      "learning_rate": 3.7906976744186048e-06,
      "loss": 0.0575,
      "step": 1630
    },
    {
      "epoch": 8.22,
      "learning_rate": 3.802325581395349e-06,
      "loss": 0.0224,
      "step": 1635
    },
    {
      "epoch": 8.24,
      "learning_rate": 3.8139534883720936e-06,
      "loss": 0.0561,
      "step": 1640
    },
    {
      "epoch": 8.27,
      "learning_rate": 3.825581395348838e-06,
      "loss": 0.0333,
      "step": 1645
    },
    {
      "epoch": 8.29,
      "learning_rate": 3.837209302325582e-06,
      "loss": 0.0881,
      "step": 1650
    },
    {
      "epoch": 8.32,
      "learning_rate": 3.848837209302326e-06,
      "loss": 0.0562,
      "step": 1655
    },
    {
      "epoch": 8.34,
      "learning_rate": 3.86046511627907e-06,
      "loss": 0.0381,
      "step": 1660
    },
    {
      "epoch": 8.37,
      "learning_rate": 3.872093023255814e-06,
      "loss": 0.0172,
      "step": 1665
    },
    {
      "epoch": 8.39,
      "learning_rate": 3.883720930232558e-06,
      "loss": 0.0262,
      "step": 1670
    },
    {
      "epoch": 8.42,
      "learning_rate": 3.8953488372093025e-06,
      "loss": 0.1236,
      "step": 1675
    },
    {
      "epoch": 8.44,
      "learning_rate": 3.906976744186047e-06,
      "loss": 0.0794,
      "step": 1680
    },
    {
      "epoch": 8.47,
      "learning_rate": 3.918604651162791e-06,
      "loss": 0.0319,
      "step": 1685
    },
    {
      "epoch": 8.49,
      "learning_rate": 3.930232558139535e-06,
      "loss": 0.0114,
      "step": 1690
    },
    {
      "epoch": 8.52,
      "learning_rate": 3.941860465116279e-06,
      "loss": 0.0872,
      "step": 1695
    },
    {
      "epoch": 8.54,
      "learning_rate": 3.953488372093024e-06,
      "loss": 0.0592,
      "step": 1700
    },
    {
      "epoch": 8.57,
      "learning_rate": 3.965116279069768e-06,
      "loss": 0.0436,
      "step": 1705
    },
    {
      "epoch": 8.59,
      "learning_rate": 3.976744186046512e-06,
      "loss": 0.0328,
      "step": 1710
    },
    {
      "epoch": 8.62,
      "learning_rate": 3.988372093023256e-06,
      "loss": 0.0602,
      "step": 1715
    },
    {
      "epoch": 8.64,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.0482,
      "step": 1720
    },
    {
      "epoch": 8.67,
      "learning_rate": 4.011627906976744e-06,
      "loss": 0.012,
      "step": 1725
    },
    {
      "epoch": 8.69,
      "learning_rate": 4.023255813953489e-06,
      "loss": 0.0723,
      "step": 1730
    },
    {
      "epoch": 8.72,
      "learning_rate": 4.034883720930233e-06,
      "loss": 0.0854,
      "step": 1735
    },
    {
      "epoch": 8.74,
      "learning_rate": 4.0465116279069775e-06,
      "loss": 0.0185,
      "step": 1740
    },
    {
      "epoch": 8.77,
      "learning_rate": 4.058139534883721e-06,
      "loss": 0.0684,
      "step": 1745
    },
    {
      "epoch": 8.79,
      "learning_rate": 4.0697674418604655e-06,
      "loss": 0.0528,
      "step": 1750
    },
    {
      "epoch": 8.82,
      "learning_rate": 4.08139534883721e-06,
      "loss": 0.0708,
      "step": 1755
    },
    {
      "epoch": 8.84,
      "learning_rate": 4.0930232558139536e-06,
      "loss": 0.0283,
      "step": 1760
    },
    {
      "epoch": 8.87,
      "learning_rate": 4.104651162790698e-06,
      "loss": 0.0126,
      "step": 1765
    },
    {
      "epoch": 8.89,
      "learning_rate": 4.116279069767442e-06,
      "loss": 0.1266,
      "step": 1770
    },
    {
      "epoch": 8.92,
      "learning_rate": 4.127906976744187e-06,
      "loss": 0.138,
      "step": 1775
    },
    {
      "epoch": 8.94,
      "learning_rate": 4.1395348837209304e-06,
      "loss": 0.1034,
      "step": 1780
    },
    {
      "epoch": 8.97,
      "learning_rate": 4.151162790697675e-06,
      "loss": 0.0554,
      "step": 1785
    },
    {
      "epoch": 8.99,
      "learning_rate": 4.1627906976744184e-06,
      "loss": 0.1105,
      "step": 1790
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.9820839226779821,
      "eval_loss": 0.05802316591143608,
      "eval_runtime": 90.801,
      "eval_samples_per_second": 23.359,
      "eval_steps_per_second": 1.465,
      "step": 1791
    },
    {
      "epoch": 9.02,
      "learning_rate": 4.174418604651163e-06,
      "loss": 0.0654,
      "step": 1795
    },
    {
      "epoch": 9.05,
      "learning_rate": 4.186046511627907e-06,
      "loss": 0.0111,
      "step": 1800
    },
    {
      "epoch": 9.07,
      "learning_rate": 4.197674418604652e-06,
      "loss": 0.0222,
      "step": 1805
    },
    {
      "epoch": 9.1,
      "learning_rate": 4.209302325581396e-06,
      "loss": 0.0517,
      "step": 1810
    },
    {
      "epoch": 9.12,
      "learning_rate": 4.22093023255814e-06,
      "loss": 0.0376,
      "step": 1815
    },
    {
      "epoch": 9.15,
      "learning_rate": 4.232558139534884e-06,
      "loss": 0.0881,
      "step": 1820
    },
    {
      "epoch": 9.17,
      "learning_rate": 4.244186046511628e-06,
      "loss": 0.0317,
      "step": 1825
    },
    {
      "epoch": 9.2,
      "learning_rate": 4.255813953488372e-06,
      "loss": 0.0075,
      "step": 1830
    },
    {
      "epoch": 9.22,
      "learning_rate": 4.267441860465117e-06,
      "loss": 0.0807,
      "step": 1835
    },
    {
      "epoch": 9.25,
      "learning_rate": 4.279069767441861e-06,
      "loss": 0.0307,
      "step": 1840
    },
    {
      "epoch": 9.27,
      "learning_rate": 4.2906976744186055e-06,
      "loss": 0.0826,
      "step": 1845
    },
    {
      "epoch": 9.3,
      "learning_rate": 4.302325581395349e-06,
      "loss": 0.0526,
      "step": 1850
    },
    {
      "epoch": 9.32,
      "learning_rate": 4.3139534883720935e-06,
      "loss": 0.0748,
      "step": 1855
    },
    {
      "epoch": 9.35,
      "learning_rate": 4.325581395348837e-06,
      "loss": 0.02,
      "step": 1860
    },
    {
      "epoch": 9.37,
      "learning_rate": 4.3372093023255815e-06,
      "loss": 0.0865,
      "step": 1865
    },
    {
      "epoch": 9.4,
      "learning_rate": 4.348837209302326e-06,
      "loss": 0.0962,
      "step": 1870
    },
    {
      "epoch": 9.42,
      "learning_rate": 4.36046511627907e-06,
      "loss": 0.0448,
      "step": 1875
    },
    {
      "epoch": 9.45,
      "learning_rate": 4.372093023255815e-06,
      "loss": 0.0414,
      "step": 1880
    },
    {
      "epoch": 9.47,
      "learning_rate": 4.383720930232558e-06,
      "loss": 0.0367,
      "step": 1885
    },
    {
      "epoch": 9.5,
      "learning_rate": 4.395348837209303e-06,
      "loss": 0.018,
      "step": 1890
    },
    {
      "epoch": 9.52,
      "learning_rate": 4.406976744186046e-06,
      "loss": 0.0238,
      "step": 1895
    },
    {
      "epoch": 9.55,
      "learning_rate": 4.418604651162791e-06,
      "loss": 0.0449,
      "step": 1900
    },
    {
      "epoch": 9.57,
      "learning_rate": 4.430232558139535e-06,
      "loss": 0.0461,
      "step": 1905
    },
    {
      "epoch": 9.6,
      "learning_rate": 4.44186046511628e-06,
      "loss": 0.0269,
      "step": 1910
    },
    {
      "epoch": 9.62,
      "learning_rate": 4.453488372093024e-06,
      "loss": 0.1015,
      "step": 1915
    },
    {
      "epoch": 9.65,
      "learning_rate": 4.465116279069768e-06,
      "loss": 0.0235,
      "step": 1920
    },
    {
      "epoch": 9.67,
      "learning_rate": 4.476744186046512e-06,
      "loss": 0.0445,
      "step": 1925
    },
    {
      "epoch": 9.7,
      "learning_rate": 4.488372093023256e-06,
      "loss": 0.0707,
      "step": 1930
    },
    {
      "epoch": 9.72,
      "learning_rate": 4.5e-06,
      "loss": 0.0562,
      "step": 1935
    },
    {
      "epoch": 9.75,
      "learning_rate": 4.5116279069767445e-06,
      "loss": 0.0444,
      "step": 1940
    },
    {
      "epoch": 9.77,
      "learning_rate": 4.523255813953489e-06,
      "loss": 0.0733,
      "step": 1945
    },
    {
      "epoch": 9.8,
      "learning_rate": 4.5348837209302326e-06,
      "loss": 0.041,
      "step": 1950
    },
    {
      "epoch": 9.82,
      "learning_rate": 4.546511627906977e-06,
      "loss": 0.0205,
      "step": 1955
    },
    {
      "epoch": 9.85,
      "learning_rate": 4.558139534883721e-06,
      "loss": 0.067,
      "step": 1960
    },
    {
      "epoch": 9.87,
      "learning_rate": 4.569767441860465e-06,
      "loss": 0.0548,
      "step": 1965
    },
    {
      "epoch": 9.9,
      "learning_rate": 4.5813953488372094e-06,
      "loss": 0.0704,
      "step": 1970
    },
    {
      "epoch": 9.92,
      "learning_rate": 4.593023255813954e-06,
      "loss": 0.0203,
      "step": 1975
    },
    {
      "epoch": 9.95,
      "learning_rate": 4.604651162790698e-06,
      "loss": 0.0189,
      "step": 1980
    },
    {
      "epoch": 9.97,
      "learning_rate": 4.616279069767442e-06,
      "loss": 0.0342,
      "step": 1985
    },
    {
      "epoch": 10.0,
      "learning_rate": 4.627906976744186e-06,
      "loss": 0.0289,
      "step": 1990
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.9924563884959925,
      "eval_loss": 0.029388291761279106,
      "eval_runtime": 90.9871,
      "eval_samples_per_second": 23.311,
      "eval_steps_per_second": 1.462,
      "step": 1990
    },
    {
      "epoch": 10.03,
      "learning_rate": 4.639534883720931e-06,
      "loss": 0.0206,
      "step": 1995
    },
    {
      "epoch": 10.05,
      "learning_rate": 4.651162790697675e-06,
      "loss": 0.0817,
      "step": 2000
    },
    {
      "epoch": 10.08,
      "learning_rate": 4.66279069767442e-06,
      "loss": 0.0143,
      "step": 2005
    },
    {
      "epoch": 10.1,
      "learning_rate": 4.674418604651163e-06,
      "loss": 0.0036,
      "step": 2010
    },
    {
      "epoch": 10.13,
      "learning_rate": 4.686046511627908e-06,
      "loss": 0.0335,
      "step": 2015
    },
    {
      "epoch": 10.15,
      "learning_rate": 4.697674418604651e-06,
      "loss": 0.0304,
      "step": 2020
    },
    {
      "epoch": 10.18,
      "learning_rate": 4.709302325581396e-06,
      "loss": 0.099,
      "step": 2025
    },
    {
      "epoch": 10.2,
      "learning_rate": 4.72093023255814e-06,
      "loss": 0.037,
      "step": 2030
    },
    {
      "epoch": 10.23,
      "learning_rate": 4.7325581395348845e-06,
      "loss": 0.0534,
      "step": 2035
    },
    {
      "epoch": 10.25,
      "learning_rate": 4.744186046511629e-06,
      "loss": 0.0079,
      "step": 2040
    },
    {
      "epoch": 10.28,
      "learning_rate": 4.7558139534883725e-06,
      "loss": 0.0551,
      "step": 2045
    },
    {
      "epoch": 10.3,
      "learning_rate": 4.767441860465117e-06,
      "loss": 0.0175,
      "step": 2050
    },
    {
      "epoch": 10.33,
      "learning_rate": 4.7790697674418605e-06,
      "loss": 0.0048,
      "step": 2055
    },
    {
      "epoch": 10.35,
      "learning_rate": 4.790697674418605e-06,
      "loss": 0.0216,
      "step": 2060
    },
    {
      "epoch": 10.38,
      "learning_rate": 4.802325581395349e-06,
      "loss": 0.0821,
      "step": 2065
    },
    {
      "epoch": 10.4,
      "learning_rate": 4.813953488372094e-06,
      "loss": 0.0045,
      "step": 2070
    },
    {
      "epoch": 10.43,
      "learning_rate": 4.825581395348838e-06,
      "loss": 0.0971,
      "step": 2075
    },
    {
      "epoch": 10.45,
      "learning_rate": 4.837209302325582e-06,
      "loss": 0.0088,
      "step": 2080
    },
    {
      "epoch": 10.48,
      "learning_rate": 4.848837209302326e-06,
      "loss": 0.0364,
      "step": 2085
    },
    {
      "epoch": 10.5,
      "learning_rate": 4.86046511627907e-06,
      "loss": 0.0391,
      "step": 2090
    },
    {
      "epoch": 10.53,
      "learning_rate": 4.872093023255814e-06,
      "loss": 0.0233,
      "step": 2095
    },
    {
      "epoch": 10.55,
      "learning_rate": 4.883720930232559e-06,
      "loss": 0.0171,
      "step": 2100
    },
    {
      "epoch": 10.58,
      "learning_rate": 4.895348837209303e-06,
      "loss": 0.0406,
      "step": 2105
    },
    {
      "epoch": 10.6,
      "learning_rate": 4.906976744186047e-06,
      "loss": 0.0124,
      "step": 2110
    },
    {
      "epoch": 10.63,
      "learning_rate": 4.918604651162791e-06,
      "loss": 0.042,
      "step": 2115
    },
    {
      "epoch": 10.65,
      "learning_rate": 4.9302325581395355e-06,
      "loss": 0.0625,
      "step": 2120
    },
    {
      "epoch": 10.68,
      "learning_rate": 4.941860465116279e-06,
      "loss": 0.0485,
      "step": 2125
    },
    {
      "epoch": 10.7,
      "learning_rate": 4.9534883720930235e-06,
      "loss": 0.0038,
      "step": 2130
    },
    {
      "epoch": 10.73,
      "learning_rate": 4.965116279069768e-06,
      "loss": 0.0635,
      "step": 2135
    },
    {
      "epoch": 10.75,
      "learning_rate": 4.976744186046512e-06,
      "loss": 0.0032,
      "step": 2140
    },
    {
      "epoch": 10.78,
      "learning_rate": 4.988372093023256e-06,
      "loss": 0.0495,
      "step": 2145
    },
    {
      "epoch": 10.8,
      "learning_rate": 5e-06,
      "loss": 0.08,
      "step": 2150
    },
    {
      "epoch": 10.83,
      "learning_rate": 4.894957983193277e-06,
      "loss": 0.0515,
      "step": 2155
    },
    {
      "epoch": 10.85,
      "learning_rate": 4.7899159663865555e-06,
      "loss": 0.0506,
      "step": 2160
    },
    {
      "epoch": 10.88,
      "learning_rate": 4.684873949579832e-06,
      "loss": 0.0511,
      "step": 2165
    },
    {
      "epoch": 10.9,
      "learning_rate": 4.57983193277311e-06,
      "loss": 0.0343,
      "step": 2170
    },
    {
      "epoch": 10.93,
      "learning_rate": 4.474789915966387e-06,
      "loss": 0.129,
      "step": 2175
    },
    {
      "epoch": 10.95,
      "learning_rate": 4.369747899159664e-06,
      "loss": 0.0301,
      "step": 2180
    },
    {
      "epoch": 10.98,
      "learning_rate": 4.264705882352942e-06,
      "loss": 0.0594,
      "step": 2185
    },
    {
      "epoch": 11.0,
      "eval_accuracy": 0.9905704856199906,
      "eval_loss": 0.03312527760863304,
      "eval_runtime": 90.7854,
      "eval_samples_per_second": 23.363,
      "eval_steps_per_second": 1.465,
      "step": 2189
    },
    {
      "epoch": 11.01,
      "learning_rate": 4.159663865546219e-06,
      "loss": 0.0181,
      "step": 2190
    },
    {
      "epoch": 11.03,
      "learning_rate": 4.054621848739496e-06,
      "loss": 0.0074,
      "step": 2195
    },
    {
      "epoch": 11.06,
      "learning_rate": 3.9495798319327735e-06,
      "loss": 0.0418,
      "step": 2200
    },
    {
      "epoch": 11.08,
      "learning_rate": 3.844537815126051e-06,
      "loss": 0.0897,
      "step": 2205
    },
    {
      "epoch": 11.11,
      "learning_rate": 3.739495798319328e-06,
      "loss": 0.0387,
      "step": 2210
    },
    {
      "epoch": 11.13,
      "learning_rate": 3.6344537815126053e-06,
      "loss": 0.0034,
      "step": 2215
    },
    {
      "epoch": 11.16,
      "learning_rate": 3.529411764705883e-06,
      "loss": 0.0177,
      "step": 2220
    },
    {
      "epoch": 11.18,
      "learning_rate": 3.42436974789916e-06,
      "loss": 0.0186,
      "step": 2225
    },
    {
      "epoch": 11.21,
      "learning_rate": 3.319327731092437e-06,
      "loss": 0.0361,
      "step": 2230
    },
    {
      "epoch": 11.23,
      "learning_rate": 3.2142857142857147e-06,
      "loss": 0.0547,
      "step": 2235
    },
    {
      "epoch": 11.26,
      "learning_rate": 3.109243697478992e-06,
      "loss": 0.1177,
      "step": 2240
    },
    {
      "epoch": 11.28,
      "learning_rate": 3.004201680672269e-06,
      "loss": 0.0343,
      "step": 2245
    },
    {
      "epoch": 11.31,
      "learning_rate": 2.8991596638655466e-06,
      "loss": 0.0807,
      "step": 2250
    },
    {
      "epoch": 11.33,
      "learning_rate": 2.7941176470588237e-06,
      "loss": 0.0048,
      "step": 2255
    },
    {
      "epoch": 11.36,
      "learning_rate": 2.689075630252101e-06,
      "loss": 0.0128,
      "step": 2260
    },
    {
      "epoch": 11.38,
      "learning_rate": 2.5840336134453784e-06,
      "loss": 0.0142,
      "step": 2265
    },
    {
      "epoch": 11.41,
      "learning_rate": 2.4789915966386555e-06,
      "loss": 0.0401,
      "step": 2270
    },
    {
      "epoch": 11.43,
      "learning_rate": 2.3739495798319327e-06,
      "loss": 0.002,
      "step": 2275
    },
    {
      "epoch": 11.46,
      "learning_rate": 2.2689075630252102e-06,
      "loss": 0.0072,
      "step": 2280
    },
    {
      "epoch": 11.48,
      "learning_rate": 2.163865546218488e-06,
      "loss": 0.0341,
      "step": 2285
    },
    {
      "epoch": 11.51,
      "learning_rate": 2.058823529411765e-06,
      "loss": 0.0059,
      "step": 2290
    },
    {
      "epoch": 11.53,
      "learning_rate": 1.953781512605042e-06,
      "loss": 0.0588,
      "step": 2295
    },
    {
      "epoch": 11.56,
      "learning_rate": 1.8487394957983196e-06,
      "loss": 0.0601,
      "step": 2300
    },
    {
      "epoch": 11.58,
      "learning_rate": 1.7436974789915968e-06,
      "loss": 0.0146,
      "step": 2305
    },
    {
      "epoch": 11.61,
      "learning_rate": 1.6386554621848741e-06,
      "loss": 0.0052,
      "step": 2310
    },
    {
      "epoch": 11.63,
      "learning_rate": 1.5336134453781515e-06,
      "loss": 0.0437,
      "step": 2315
    },
    {
      "epoch": 11.66,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 0.0075,
      "step": 2320
    },
    {
      "epoch": 11.68,
      "learning_rate": 1.323529411764706e-06,
      "loss": 0.0561,
      "step": 2325
    },
    {
      "epoch": 11.71,
      "learning_rate": 1.2184873949579833e-06,
      "loss": 0.0718,
      "step": 2330
    },
    {
      "epoch": 11.73,
      "learning_rate": 1.1134453781512607e-06,
      "loss": 0.0006,
      "step": 2335
    },
    {
      "epoch": 11.76,
      "learning_rate": 1.0084033613445378e-06,
      "loss": 0.0184,
      "step": 2340
    },
    {
      "epoch": 11.78,
      "learning_rate": 9.033613445378152e-07,
      "loss": 0.0341,
      "step": 2345
    },
    {
      "epoch": 11.81,
      "learning_rate": 7.983193277310924e-07,
      "loss": 0.0058,
      "step": 2350
    },
    {
      "epoch": 11.83,
      "learning_rate": 6.932773109243699e-07,
      "loss": 0.0086,
      "step": 2355
    },
    {
      "epoch": 11.86,
      "learning_rate": 5.882352941176471e-07,
      "loss": 0.0563,
      "step": 2360
    },
    {
      "epoch": 11.88,
      "learning_rate": 4.831932773109245e-07,
      "loss": 0.0036,
      "step": 2365
    },
    {
      "epoch": 11.91,
      "learning_rate": 3.781512605042017e-07,
      "loss": 0.018,
      "step": 2370
    },
    {
      "epoch": 11.93,
      "learning_rate": 2.73109243697479e-07,
      "loss": 0.0058,
      "step": 2375
    },
    {
      "epoch": 11.96,
      "learning_rate": 1.680672268907563e-07,
      "loss": 0.0008,
      "step": 2380
    },
    {
      "epoch": 11.98,
      "learning_rate": 6.302521008403361e-08,
      "loss": 0.0011,
      "step": 2385
    },
    {
      "epoch": 12.0,
      "eval_accuracy": 0.9938708156529938,
      "eval_loss": 0.0275154709815979,
      "eval_runtime": 90.8878,
      "eval_samples_per_second": 23.336,
      "eval_steps_per_second": 1.463,
      "step": 2388
    },
    {
      "epoch": 12.0,
      "step": 2388,
      "total_flos": 2.0923685902916665e+19,
      "train_loss": 0.1486589150943784,
      "train_runtime": 11149.9067,
      "train_samples_per_second": 6.848,
      "train_steps_per_second": 0.214
    }
  ],
  "max_steps": 2388,
  "num_train_epochs": 12,
  "total_flos": 2.0923685902916665e+19,
  "trial_name": null,
  "trial_params": null
}