{
  "best_metric": 0.78,
  "best_model_checkpoint": "videomae-base-finetuned-ucf101-subset/checkpoint-3184",
  "epoch": 10.002506265664161,
  "global_step": 3990,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 1.2531328320802005e-06,
      "loss": 2.6652,
      "step": 10
    },
    {
      "epoch": 0.01,
      "learning_rate": 2.506265664160401e-06,
      "loss": 2.6755,
      "step": 20
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.7593984962406014e-06,
      "loss": 2.5836,
      "step": 30
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.012531328320802e-06,
      "loss": 2.661,
      "step": 40
    },
    {
      "epoch": 0.01,
      "learning_rate": 6.265664160401003e-06,
      "loss": 2.3916,
      "step": 50
    },
    {
      "epoch": 0.02,
      "learning_rate": 7.518796992481203e-06,
      "loss": 2.5649,
      "step": 60
    },
    {
      "epoch": 0.02,
      "learning_rate": 8.771929824561403e-06,
      "loss": 2.6852,
      "step": 70
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.0025062656641604e-05,
      "loss": 2.6172,
      "step": 80
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.1278195488721805e-05,
      "loss": 2.5301,
      "step": 90
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.2531328320802006e-05,
      "loss": 2.6607,
      "step": 100
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.3784461152882205e-05,
      "loss": 2.4796,
      "step": 110
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.5037593984962406e-05,
      "loss": 2.4986,
      "step": 120
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.6290726817042605e-05,
      "loss": 2.5192,
      "step": 130
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.7543859649122806e-05,
      "loss": 2.615,
      "step": 140
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.8796992481203007e-05,
      "loss": 2.5708,
      "step": 150
    },
    {
      "epoch": 0.04,
      "learning_rate": 2.0050125313283208e-05,
      "loss": 2.3817,
      "step": 160
    },
    {
      "epoch": 0.04,
      "learning_rate": 2.130325814536341e-05,
      "loss": 2.4658,
      "step": 170
    },
    {
      "epoch": 0.05,
      "learning_rate": 2.255639097744361e-05,
      "loss": 2.5414,
      "step": 180
    },
    {
      "epoch": 0.05,
      "learning_rate": 2.380952380952381e-05,
      "loss": 2.6049,
      "step": 190
    },
    {
      "epoch": 0.05,
      "learning_rate": 2.506265664160401e-05,
      "loss": 2.5113,
      "step": 200
    },
    {
      "epoch": 0.05,
      "learning_rate": 2.6315789473684212e-05,
      "loss": 2.6226,
      "step": 210
    },
    {
      "epoch": 0.06,
      "learning_rate": 2.756892230576441e-05,
      "loss": 2.4038,
      "step": 220
    },
    {
      "epoch": 0.06,
      "learning_rate": 2.882205513784461e-05,
      "loss": 2.2845,
      "step": 230
    },
    {
      "epoch": 0.06,
      "learning_rate": 3.007518796992481e-05,
      "loss": 2.3279,
      "step": 240
    },
    {
      "epoch": 0.06,
      "learning_rate": 3.132832080200501e-05,
      "loss": 2.3244,
      "step": 250
    },
    {
      "epoch": 0.07,
      "learning_rate": 3.258145363408521e-05,
      "loss": 2.2399,
      "step": 260
    },
    {
      "epoch": 0.07,
      "learning_rate": 3.3834586466165414e-05,
      "loss": 2.5944,
      "step": 270
    },
    {
      "epoch": 0.07,
      "learning_rate": 3.508771929824561e-05,
      "loss": 2.3696,
      "step": 280
    },
    {
      "epoch": 0.07,
      "learning_rate": 3.6340852130325816e-05,
      "loss": 2.3556,
      "step": 290
    },
    {
      "epoch": 0.08,
      "learning_rate": 3.759398496240601e-05,
      "loss": 2.1208,
      "step": 300
    },
    {
      "epoch": 0.08,
      "learning_rate": 3.884711779448622e-05,
      "loss": 1.9662,
      "step": 310
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.0100250626566415e-05,
      "loss": 2.3956,
      "step": 320
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.135338345864662e-05,
      "loss": 2.3429,
      "step": 330
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.260651629072682e-05,
      "loss": 2.3078,
      "step": 340
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.3859649122807014e-05,
      "loss": 2.1788,
      "step": 350
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.511278195488722e-05,
      "loss": 2.2873,
      "step": 360
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.6365914786967416e-05,
      "loss": 2.4288,
      "step": 370
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.761904761904762e-05,
      "loss": 1.7902,
      "step": 380
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.887218045112782e-05,
      "loss": 2.1012,
      "step": 390
    },
    {
      "epoch": 0.1,
      "eval_accuracy": 0.38,
      "eval_loss": 1.9809015989303589,
      "eval_runtime": 27.6095,
      "eval_samples_per_second": 3.622,
      "eval_steps_per_second": 3.622,
      "step": 398
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.998607630186578e-05,
      "loss": 1.6446,
      "step": 400
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.9846839320523534e-05,
      "loss": 2.495,
      "step": 410
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.970760233918128e-05,
      "loss": 2.1706,
      "step": 420
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.956836535783904e-05,
      "loss": 1.6511,
      "step": 430
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.9429128376496795e-05,
      "loss": 2.0323,
      "step": 440
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.928989139515455e-05,
      "loss": 1.8594,
      "step": 450
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.9150654413812306e-05,
      "loss": 1.6434,
      "step": 460
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.901141743247006e-05,
      "loss": 1.5981,
      "step": 470
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.887218045112782e-05,
      "loss": 2.3124,
      "step": 480
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.8732943469785574e-05,
      "loss": 1.3398,
      "step": 490
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.859370648844333e-05,
      "loss": 2.075,
      "step": 500
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.8454469507101085e-05,
      "loss": 2.1273,
      "step": 510
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.831523252575884e-05,
      "loss": 1.4408,
      "step": 520
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.81759955444166e-05,
      "loss": 1.8349,
      "step": 530
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.803675856307435e-05,
      "loss": 1.4592,
      "step": 540
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.789752158173211e-05,
      "loss": 2.0761,
      "step": 550
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.7758284600389865e-05,
      "loss": 1.6495,
      "step": 560
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.761904761904762e-05,
      "loss": 1.0583,
      "step": 570
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.7479810637705376e-05,
      "loss": 1.4962,
      "step": 580
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.734057365636313e-05,
      "loss": 1.065,
      "step": 590
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.720133667502089e-05,
      "loss": 2.0059,
      "step": 600
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.7062099693678644e-05,
      "loss": 1.8712,
      "step": 610
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.69228627123364e-05,
      "loss": 1.9663,
      "step": 620
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.678362573099415e-05,
      "loss": 2.1448,
      "step": 630
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.6644388749651904e-05,
      "loss": 1.5725,
      "step": 640
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.650515176830966e-05,
      "loss": 1.6991,
      "step": 650
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.6365914786967416e-05,
      "loss": 1.291,
      "step": 660
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.622667780562517e-05,
      "loss": 1.8409,
      "step": 670
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.608744082428293e-05,
      "loss": 2.2183,
      "step": 680
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.5948203842940684e-05,
      "loss": 1.6034,
      "step": 690
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.580896686159844e-05,
      "loss": 1.4481,
      "step": 700
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.5669729880256195e-05,
      "loss": 0.9722,
      "step": 710
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.553049289891395e-05,
      "loss": 1.447,
      "step": 720
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.539125591757171e-05,
      "loss": 1.6108,
      "step": 730
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.525201893622946e-05,
      "loss": 2.3506,
      "step": 740
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.511278195488722e-05,
      "loss": 0.9662,
      "step": 750
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.4973544973544974e-05,
      "loss": 1.3062,
      "step": 760
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.483430799220273e-05,
      "loss": 1.3071,
      "step": 770
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.4695071010860486e-05,
      "loss": 0.9615,
      "step": 780
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.455583402951824e-05,
      "loss": 1.0416,
      "step": 790
    },
    {
      "epoch": 1.1,
      "eval_accuracy": 0.56,
      "eval_loss": 1.6140196323394775,
      "eval_runtime": 27.3792,
      "eval_samples_per_second": 3.652,
      "eval_steps_per_second": 3.652,
      "step": 796
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.4416597048176e-05,
      "loss": 0.567,
      "step": 800
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.4277360066833754e-05,
      "loss": 1.2888,
      "step": 810
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.413812308549151e-05,
      "loss": 1.2286,
      "step": 820
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.399888610414926e-05,
      "loss": 0.7935,
      "step": 830
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.3859649122807014e-05,
      "loss": 0.9961,
      "step": 840
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.372041214146477e-05,
      "loss": 1.3997,
      "step": 850
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.3581175160122526e-05,
      "loss": 1.112,
      "step": 860
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.344193817878028e-05,
      "loss": 0.3837,
      "step": 870
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.330270119743804e-05,
      "loss": 2.3791,
      "step": 880
    },
    {
      "epoch": 2.02,
      "learning_rate": 4.3163464216095793e-05,
      "loss": 0.7274,
      "step": 890
    },
    {
      "epoch": 2.03,
      "learning_rate": 4.302422723475355e-05,
      "loss": 1.2903,
      "step": 900
    },
    {
      "epoch": 2.03,
      "learning_rate": 4.2884990253411305e-05,
      "loss": 0.3864,
      "step": 910
    },
    {
      "epoch": 2.03,
      "learning_rate": 4.274575327206906e-05,
      "loss": 0.8519,
      "step": 920
    },
    {
      "epoch": 2.03,
      "learning_rate": 4.260651629072682e-05,
      "loss": 0.2129,
      "step": 930
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.246727930938457e-05,
      "loss": 0.9185,
      "step": 940
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.232804232804233e-05,
      "loss": 1.5575,
      "step": 950
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.2188805346700084e-05,
      "loss": 0.3068,
      "step": 960
    },
    {
      "epoch": 2.04,
      "learning_rate": 4.204956836535784e-05,
      "loss": 0.8577,
      "step": 970
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.1910331384015596e-05,
      "loss": 1.2818,
      "step": 980
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.177109440267335e-05,
      "loss": 0.2316,
      "step": 990
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.163185742133111e-05,
      "loss": 0.9911,
      "step": 1000
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.1492620439988863e-05,
      "loss": 1.0393,
      "step": 1010
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.135338345864662e-05,
      "loss": 1.7585,
      "step": 1020
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.121414647730437e-05,
      "loss": 0.6806,
      "step": 1030
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.1074909495962124e-05,
      "loss": 0.3257,
      "step": 1040
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.093567251461988e-05,
      "loss": 1.249,
      "step": 1050
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.0796435533277636e-05,
      "loss": 0.5324,
      "step": 1060
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.065719855193539e-05,
      "loss": 1.3123,
      "step": 1070
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.051796157059315e-05,
      "loss": 0.285,
      "step": 1080
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.03787245892509e-05,
      "loss": 0.5411,
      "step": 1090
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.023948760790866e-05,
      "loss": 1.7209,
      "step": 1100
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.0100250626566415e-05,
      "loss": 0.5554,
      "step": 1110
    },
    {
      "epoch": 2.08,
      "learning_rate": 3.996101364522417e-05,
      "loss": 1.8234,
      "step": 1120
    },
    {
      "epoch": 2.08,
      "learning_rate": 3.982177666388193e-05,
      "loss": 0.9143,
      "step": 1130
    },
    {
      "epoch": 2.09,
      "learning_rate": 3.968253968253968e-05,
      "loss": 1.6595,
      "step": 1140
    },
    {
      "epoch": 2.09,
      "learning_rate": 3.954330270119744e-05,
      "loss": 1.0291,
      "step": 1150
    },
    {
      "epoch": 2.09,
      "learning_rate": 3.9404065719855194e-05,
      "loss": 1.24,
      "step": 1160
    },
    {
      "epoch": 2.09,
      "learning_rate": 3.926482873851295e-05,
      "loss": 0.849,
      "step": 1170
    },
    {
      "epoch": 2.1,
      "learning_rate": 3.9125591757170706e-05,
      "loss": 1.5756,
      "step": 1180
    },
    {
      "epoch": 2.1,
      "learning_rate": 3.898635477582846e-05,
      "loss": 0.2096,
      "step": 1190
    },
    {
      "epoch": 2.1,
      "eval_accuracy": 0.66,
      "eval_loss": 1.5775662660598755,
      "eval_runtime": 28.2107,
      "eval_samples_per_second": 3.545,
      "eval_steps_per_second": 3.545,
      "step": 1194
    },
    {
      "epoch": 3.0,
      "learning_rate": 3.884711779448622e-05,
      "loss": 1.3493,
      "step": 1200
    },
    {
      "epoch": 3.0,
      "learning_rate": 3.870788081314397e-05,
      "loss": 1.3687,
      "step": 1210
    },
    {
      "epoch": 3.01,
      "learning_rate": 3.856864383180173e-05,
      "loss": 0.8319,
      "step": 1220
    },
    {
      "epoch": 3.01,
      "learning_rate": 3.8429406850459485e-05,
      "loss": 0.9219,
      "step": 1230
    },
    {
      "epoch": 3.01,
      "learning_rate": 3.8290169869117234e-05,
      "loss": 0.3956,
      "step": 1240
    },
    {
      "epoch": 3.01,
      "learning_rate": 3.815093288777499e-05,
      "loss": 0.6158,
      "step": 1250
    },
    {
      "epoch": 3.02,
      "learning_rate": 3.8011695906432746e-05,
      "loss": 0.8011,
      "step": 1260
    },
    {
      "epoch": 3.02,
      "learning_rate": 3.78724589250905e-05,
      "loss": 1.3677,
      "step": 1270
    },
    {
      "epoch": 3.02,
      "learning_rate": 3.773322194374826e-05,
      "loss": 1.5107,
      "step": 1280
    },
    {
      "epoch": 3.02,
      "learning_rate": 3.759398496240601e-05,
      "loss": 0.1896,
      "step": 1290
    },
    {
      "epoch": 3.03,
      "learning_rate": 3.745474798106377e-05,
      "loss": 0.6759,
      "step": 1300
    },
    {
      "epoch": 3.03,
      "learning_rate": 3.7315510999721525e-05,
      "loss": 0.0941,
      "step": 1310
    },
    {
      "epoch": 3.03,
      "learning_rate": 3.717627401837928e-05,
      "loss": 0.7546,
      "step": 1320
    },
    {
      "epoch": 3.03,
      "learning_rate": 3.7037037037037037e-05,
      "loss": 0.6543,
      "step": 1330
    },
    {
      "epoch": 3.04,
      "learning_rate": 3.689780005569479e-05,
      "loss": 0.5026,
      "step": 1340
    },
    {
      "epoch": 3.04,
      "learning_rate": 3.675856307435255e-05,
      "loss": 0.376,
      "step": 1350
    },
    {
      "epoch": 3.04,
      "learning_rate": 3.6619326093010304e-05,
      "loss": 0.9097,
      "step": 1360
    },
    {
      "epoch": 3.04,
      "learning_rate": 3.648008911166806e-05,
      "loss": 0.5431,
      "step": 1370
    },
    {
      "epoch": 3.05,
      "learning_rate": 3.6340852130325816e-05,
      "loss": 0.6503,
      "step": 1380
    },
    {
      "epoch": 3.05,
      "learning_rate": 3.620161514898357e-05,
      "loss": 0.6169,
      "step": 1390
    },
    {
      "epoch": 3.05,
      "learning_rate": 3.606237816764133e-05,
      "loss": 0.3653,
      "step": 1400
    },
    {
      "epoch": 3.05,
      "learning_rate": 3.592314118629908e-05,
      "loss": 0.1474,
      "step": 1410
    },
    {
      "epoch": 3.06,
      "learning_rate": 3.578390420495684e-05,
      "loss": 0.9832,
      "step": 1420
    },
    {
      "epoch": 3.06,
      "learning_rate": 3.5644667223614595e-05,
      "loss": 0.3245,
      "step": 1430
    },
    {
      "epoch": 3.06,
      "learning_rate": 3.5505430242272344e-05,
      "loss": 1.3136,
      "step": 1440
    },
    {
      "epoch": 3.06,
      "learning_rate": 3.53661932609301e-05,
      "loss": 0.8991,
      "step": 1450
    },
    {
      "epoch": 3.07,
      "learning_rate": 3.5226956279587856e-05,
      "loss": 1.8971,
      "step": 1460
    },
    {
      "epoch": 3.07,
      "learning_rate": 3.508771929824561e-05,
      "loss": 1.1886,
      "step": 1470
    },
    {
      "epoch": 3.07,
      "learning_rate": 3.494848231690337e-05,
      "loss": 0.6224,
      "step": 1480
    },
    {
      "epoch": 3.07,
      "learning_rate": 3.480924533556112e-05,
      "loss": 0.1107,
      "step": 1490
    },
    {
      "epoch": 3.08,
      "learning_rate": 3.467000835421888e-05,
      "loss": 2.0252,
      "step": 1500
    },
    {
      "epoch": 3.08,
      "learning_rate": 3.4530771372876635e-05,
      "loss": 0.5866,
      "step": 1510
    },
    {
      "epoch": 3.08,
      "learning_rate": 3.439153439153439e-05,
      "loss": 0.5512,
      "step": 1520
    },
    {
      "epoch": 3.08,
      "learning_rate": 3.4252297410192146e-05,
      "loss": 0.5085,
      "step": 1530
    },
    {
      "epoch": 3.09,
      "learning_rate": 3.41130604288499e-05,
      "loss": 0.0272,
      "step": 1540
    },
    {
      "epoch": 3.09,
      "learning_rate": 3.397382344750766e-05,
      "loss": 0.0063,
      "step": 1550
    },
    {
      "epoch": 3.09,
      "learning_rate": 3.3834586466165414e-05,
      "loss": 0.1569,
      "step": 1560
    },
    {
      "epoch": 3.09,
      "learning_rate": 3.369534948482317e-05,
      "loss": 1.3151,
      "step": 1570
    },
    {
      "epoch": 3.1,
      "learning_rate": 3.3556112503480926e-05,
      "loss": 0.0256,
      "step": 1580
    },
    {
      "epoch": 3.1,
      "learning_rate": 3.341687552213868e-05,
      "loss": 0.7101,
      "step": 1590
    },
    {
      "epoch": 3.1,
      "eval_accuracy": 0.74,
      "eval_loss": 1.2003577947616577,
      "eval_runtime": 28.5623,
      "eval_samples_per_second": 3.501,
      "eval_steps_per_second": 3.501,
      "step": 1592
    },
    {
      "epoch": 4.0,
      "learning_rate": 3.327763854079644e-05,
      "loss": 0.0158,
      "step": 1600
    },
    {
      "epoch": 4.0,
      "learning_rate": 3.313840155945419e-05,
      "loss": 1.1735,
      "step": 1610
    },
    {
      "epoch": 4.01,
      "learning_rate": 3.299916457811195e-05,
      "loss": 0.0063,
      "step": 1620
    },
    {
      "epoch": 4.01,
      "learning_rate": 3.2859927596769705e-05,
      "loss": 0.2802,
      "step": 1630
    },
    {
      "epoch": 4.01,
      "learning_rate": 3.272069061542746e-05,
      "loss": 1.0417,
      "step": 1640
    },
    {
      "epoch": 4.01,
      "learning_rate": 3.258145363408521e-05,
      "loss": 0.2255,
      "step": 1650
    },
    {
      "epoch": 4.02,
      "learning_rate": 3.2442216652742965e-05,
      "loss": 0.2863,
      "step": 1660
    },
    {
      "epoch": 4.02,
      "learning_rate": 3.230297967140072e-05,
      "loss": 0.0177,
      "step": 1670
    },
    {
      "epoch": 4.02,
      "learning_rate": 3.216374269005848e-05,
      "loss": 0.0132,
      "step": 1680
    },
    {
      "epoch": 4.02,
      "learning_rate": 3.202450570871623e-05,
      "loss": 1.1547,
      "step": 1690
    },
    {
      "epoch": 4.03,
      "learning_rate": 3.188526872737399e-05,
      "loss": 0.4432,
      "step": 1700
    },
    {
      "epoch": 4.03,
      "learning_rate": 3.1746031746031745e-05,
      "loss": 0.4861,
      "step": 1710
    },
    {
      "epoch": 4.03,
      "learning_rate": 3.16067947646895e-05,
      "loss": 0.5689,
      "step": 1720
    },
    {
      "epoch": 4.03,
      "learning_rate": 3.1467557783347256e-05,
      "loss": 0.0335,
      "step": 1730
    },
    {
      "epoch": 4.04,
      "learning_rate": 3.132832080200501e-05,
      "loss": 0.6061,
      "step": 1740
    },
    {
      "epoch": 4.04,
      "learning_rate": 3.118908382066277e-05,
      "loss": 0.0297,
      "step": 1750
    },
    {
      "epoch": 4.04,
      "learning_rate": 3.1049846839320524e-05,
      "loss": 0.6656,
      "step": 1760
    },
    {
      "epoch": 4.04,
      "learning_rate": 3.091060985797828e-05,
      "loss": 0.0602,
      "step": 1770
    },
    {
      "epoch": 4.05,
      "learning_rate": 3.0771372876636035e-05,
      "loss": 0.7032,
      "step": 1780
    },
    {
      "epoch": 4.05,
      "learning_rate": 3.063213589529379e-05,
      "loss": 0.4036,
      "step": 1790
    },
    {
      "epoch": 4.05,
      "learning_rate": 3.0492898913951544e-05,
      "loss": 0.0042,
      "step": 1800
    },
    {
      "epoch": 4.05,
      "learning_rate": 3.03536619326093e-05,
      "loss": 0.3816,
      "step": 1810
    },
    {
      "epoch": 4.06,
      "learning_rate": 3.0214424951267055e-05,
      "loss": 0.1152,
      "step": 1820
    },
    {
      "epoch": 4.06,
      "learning_rate": 3.007518796992481e-05,
      "loss": 0.0035,
      "step": 1830
    },
    {
      "epoch": 4.06,
      "learning_rate": 2.9935950988582567e-05,
      "loss": 0.0045,
      "step": 1840
    },
    {
      "epoch": 4.06,
      "learning_rate": 2.9796714007240323e-05,
      "loss": 0.5133,
      "step": 1850
    },
    {
      "epoch": 4.07,
      "learning_rate": 2.965747702589808e-05,
      "loss": 0.0034,
      "step": 1860
    },
    {
      "epoch": 4.07,
      "learning_rate": 2.9518240044555835e-05,
      "loss": 0.3994,
      "step": 1870
    },
    {
      "epoch": 4.07,
      "learning_rate": 2.937900306321359e-05,
      "loss": 0.727,
      "step": 1880
    },
    {
      "epoch": 4.07,
      "learning_rate": 2.9239766081871346e-05,
      "loss": 0.5794,
      "step": 1890
    },
    {
      "epoch": 4.08,
      "learning_rate": 2.91005291005291e-05,
      "loss": 1.3733,
      "step": 1900
    },
    {
      "epoch": 4.08,
      "learning_rate": 2.8961292119186855e-05,
      "loss": 0.1292,
      "step": 1910
    },
    {
      "epoch": 4.08,
      "learning_rate": 2.882205513784461e-05,
      "loss": 0.6721,
      "step": 1920
    },
    {
      "epoch": 4.08,
      "learning_rate": 2.8682818156502366e-05,
      "loss": 0.6031,
      "step": 1930
    },
    {
      "epoch": 4.09,
      "learning_rate": 2.8543581175160122e-05,
      "loss": 0.0148,
      "step": 1940
    },
    {
      "epoch": 4.09,
      "learning_rate": 2.8404344193817878e-05,
      "loss": 0.7074,
      "step": 1950
    },
    {
      "epoch": 4.09,
      "learning_rate": 2.8265107212475634e-05,
      "loss": 0.004,
      "step": 1960
    },
    {
      "epoch": 4.09,
      "learning_rate": 2.812587023113339e-05,
      "loss": 1.1926,
      "step": 1970
    },
    {
      "epoch": 4.1,
      "learning_rate": 2.7986633249791145e-05,
      "loss": 0.7448,
      "step": 1980
    },
    {
      "epoch": 4.1,
      "learning_rate": 2.78473962684489e-05,
      "loss": 1.2344,
      "step": 1990
    },
    {
      "epoch": 4.1,
      "eval_accuracy": 0.58,
      "eval_loss": 1.962124228477478,
      "eval_runtime": 28.4167,
      "eval_samples_per_second": 3.519,
      "eval_steps_per_second": 3.519,
      "step": 1990
    },
    {
      "epoch": 5.0,
      "learning_rate": 2.7708159287106654e-05,
      "loss": 0.8195,
      "step": 2000
    },
    {
      "epoch": 5.01,
      "learning_rate": 2.756892230576441e-05,
      "loss": 0.4829,
      "step": 2010
    },
    {
      "epoch": 5.01,
      "learning_rate": 2.7429685324422165e-05,
      "loss": 0.0041,
      "step": 2020
    },
    {
      "epoch": 5.01,
      "learning_rate": 2.729044834307992e-05,
      "loss": 0.2123,
      "step": 2030
    },
    {
      "epoch": 5.01,
      "learning_rate": 2.7151211361737677e-05,
      "loss": 0.5932,
      "step": 2040
    },
    {
      "epoch": 5.02,
      "learning_rate": 2.7011974380395433e-05,
      "loss": 0.0063,
      "step": 2050
    },
    {
      "epoch": 5.02,
      "learning_rate": 2.687273739905319e-05,
      "loss": 0.3498,
      "step": 2060
    },
    {
      "epoch": 5.02,
      "learning_rate": 2.6733500417710944e-05,
      "loss": 0.9616,
      "step": 2070
    },
    {
      "epoch": 5.02,
      "learning_rate": 2.65942634363687e-05,
      "loss": 0.0021,
      "step": 2080
    },
    {
      "epoch": 5.03,
      "learning_rate": 2.6455026455026456e-05,
      "loss": 0.4937,
      "step": 2090
    },
    {
      "epoch": 5.03,
      "learning_rate": 2.6315789473684212e-05,
      "loss": 0.2354,
      "step": 2100
    },
    {
      "epoch": 5.03,
      "learning_rate": 2.6176552492341964e-05,
      "loss": 0.3209,
      "step": 2110
    },
    {
      "epoch": 5.03,
      "learning_rate": 2.603731551099972e-05,
      "loss": 0.0217,
      "step": 2120
    },
    {
      "epoch": 5.04,
      "learning_rate": 2.5898078529657476e-05,
      "loss": 0.5236,
      "step": 2130
    },
    {
      "epoch": 5.04,
      "learning_rate": 2.5758841548315232e-05,
      "loss": 0.0044,
      "step": 2140
    },
    {
      "epoch": 5.04,
      "learning_rate": 2.5619604566972988e-05,
      "loss": 0.0032,
      "step": 2150
    },
    {
      "epoch": 5.04,
      "learning_rate": 2.5480367585630744e-05,
      "loss": 0.0117,
      "step": 2160
    },
    {
      "epoch": 5.05,
      "learning_rate": 2.53411306042885e-05,
      "loss": 0.0071,
      "step": 2170
    },
    {
      "epoch": 5.05,
      "learning_rate": 2.5201893622946255e-05,
      "loss": 0.0029,
      "step": 2180
    },
    {
      "epoch": 5.05,
      "learning_rate": 2.506265664160401e-05,
      "loss": 0.5591,
      "step": 2190
    },
    {
      "epoch": 5.05,
      "learning_rate": 2.4923419660261767e-05,
      "loss": 0.0042,
      "step": 2200
    },
    {
      "epoch": 5.06,
      "learning_rate": 2.478418267891952e-05,
      "loss": 0.0027,
      "step": 2210
    },
    {
      "epoch": 5.06,
      "learning_rate": 2.4644945697577275e-05,
      "loss": 0.2002,
      "step": 2220
    },
    {
      "epoch": 5.06,
      "learning_rate": 2.450570871623503e-05,
      "loss": 0.4917,
      "step": 2230
    },
    {
      "epoch": 5.06,
      "learning_rate": 2.4366471734892787e-05,
      "loss": 0.5791,
      "step": 2240
    },
    {
      "epoch": 5.07,
      "learning_rate": 2.4227234753550543e-05,
      "loss": 0.6229,
      "step": 2250
    },
    {
      "epoch": 5.07,
      "learning_rate": 2.40879977722083e-05,
      "loss": 0.4335,
      "step": 2260
    },
    {
      "epoch": 5.07,
      "learning_rate": 2.3948760790866054e-05,
      "loss": 0.7612,
      "step": 2270
    },
    {
      "epoch": 5.07,
      "learning_rate": 2.380952380952381e-05,
      "loss": 0.8699,
      "step": 2280
    },
    {
      "epoch": 5.08,
      "learning_rate": 2.3670286828181566e-05,
      "loss": 0.7213,
      "step": 2290
    },
    {
      "epoch": 5.08,
      "learning_rate": 2.3531049846839322e-05,
      "loss": 0.0036,
      "step": 2300
    },
    {
      "epoch": 5.08,
      "learning_rate": 2.3391812865497074e-05,
      "loss": 0.5293,
      "step": 2310
    },
    {
      "epoch": 5.08,
      "learning_rate": 2.325257588415483e-05,
      "loss": 0.002,
      "step": 2320
    },
    {
      "epoch": 5.09,
      "learning_rate": 2.3113338902812586e-05,
      "loss": 0.0098,
      "step": 2330
    },
    {
      "epoch": 5.09,
      "learning_rate": 2.2974101921470342e-05,
      "loss": 0.0034,
      "step": 2340
    },
    {
      "epoch": 5.09,
      "learning_rate": 2.2834864940128098e-05,
      "loss": 0.0018,
      "step": 2350
    },
    {
      "epoch": 5.09,
      "learning_rate": 2.2695627958785853e-05,
      "loss": 0.0022,
      "step": 2360
    },
    {
      "epoch": 5.1,
      "learning_rate": 2.255639097744361e-05,
      "loss": 0.9986,
      "step": 2370
    },
    {
      "epoch": 5.1,
      "learning_rate": 2.2417153996101365e-05,
      "loss": 0.1809,
      "step": 2380
    },
    {
      "epoch": 5.1,
      "eval_accuracy": 0.71,
      "eval_loss": 1.6321651935577393,
      "eval_runtime": 26.5939,
      "eval_samples_per_second": 3.76,
      "eval_steps_per_second": 3.76,
      "step": 2388
    },
    {
      "epoch": 6.0,
      "learning_rate": 2.227791701475912e-05,
      "loss": 0.0023,
      "step": 2390
    },
    {
      "epoch": 6.0,
      "learning_rate": 2.2138680033416877e-05,
      "loss": 0.2565,
      "step": 2400
    },
    {
      "epoch": 6.01,
      "learning_rate": 2.199944305207463e-05,
      "loss": 0.2508,
      "step": 2410
    },
    {
      "epoch": 6.01,
      "learning_rate": 2.1860206070732385e-05,
      "loss": 0.0077,
      "step": 2420
    },
    {
      "epoch": 6.01,
      "learning_rate": 2.172096908939014e-05,
      "loss": 0.0019,
      "step": 2430
    },
    {
      "epoch": 6.01,
      "learning_rate": 2.1581732108047897e-05,
      "loss": 0.1219,
      "step": 2440
    },
    {
      "epoch": 6.02,
      "learning_rate": 2.1442495126705653e-05,
      "loss": 0.004,
      "step": 2450
    },
    {
      "epoch": 6.02,
      "learning_rate": 2.130325814536341e-05,
      "loss": 0.002,
      "step": 2460
    },
    {
      "epoch": 6.02,
      "learning_rate": 2.1164021164021164e-05,
      "loss": 0.0189,
      "step": 2470
    },
    {
      "epoch": 6.02,
      "learning_rate": 2.102478418267892e-05,
      "loss": 0.0526,
      "step": 2480
    },
    {
      "epoch": 6.03,
      "learning_rate": 2.0885547201336676e-05,
      "loss": 0.0016,
      "step": 2490
    },
    {
      "epoch": 6.03,
      "learning_rate": 2.0746310219994432e-05,
      "loss": 0.0014,
      "step": 2500
    },
    {
      "epoch": 6.03,
      "learning_rate": 2.0607073238652184e-05,
      "loss": 0.002,
      "step": 2510
    },
    {
      "epoch": 6.03,
      "learning_rate": 2.046783625730994e-05,
      "loss": 0.0288,
      "step": 2520
    },
    {
      "epoch": 6.04,
      "learning_rate": 2.0328599275967696e-05,
      "loss": 0.0026,
      "step": 2530
    },
    {
      "epoch": 6.04,
      "learning_rate": 2.018936229462545e-05,
      "loss": 0.0537,
      "step": 2540
    },
    {
      "epoch": 6.04,
      "learning_rate": 2.0050125313283208e-05,
      "loss": 0.0024,
      "step": 2550
    },
    {
      "epoch": 6.04,
      "learning_rate": 1.9910888331940963e-05,
      "loss": 0.0017,
      "step": 2560
    },
    {
      "epoch": 6.05,
      "learning_rate": 1.977165135059872e-05,
      "loss": 0.005,
      "step": 2570
    },
    {
      "epoch": 6.05,
      "learning_rate": 1.9632414369256475e-05,
      "loss": 0.0019,
      "step": 2580
    },
    {
      "epoch": 6.05,
      "learning_rate": 1.949317738791423e-05,
      "loss": 0.4555,
      "step": 2590
    },
    {
      "epoch": 6.05,
      "learning_rate": 1.9353940406571987e-05,
      "loss": 0.5544,
      "step": 2600
    },
    {
      "epoch": 6.06,
      "learning_rate": 1.9214703425229743e-05,
      "loss": 0.0016,
      "step": 2610
    },
    {
      "epoch": 6.06,
      "learning_rate": 1.9075466443887495e-05,
      "loss": 0.0032,
      "step": 2620
    },
    {
      "epoch": 6.06,
      "learning_rate": 1.893622946254525e-05,
      "loss": 0.3974,
      "step": 2630
    },
    {
      "epoch": 6.06,
      "learning_rate": 1.8796992481203007e-05,
      "loss": 0.1622,
      "step": 2640
    },
    {
      "epoch": 6.07,
      "learning_rate": 1.8657755499860762e-05,
      "loss": 0.5769,
      "step": 2650
    },
    {
      "epoch": 6.07,
      "learning_rate": 1.8518518518518518e-05,
      "loss": 0.0036,
      "step": 2660
    },
    {
      "epoch": 6.07,
      "learning_rate": 1.8379281537176274e-05,
      "loss": 0.0026,
      "step": 2670
    },
    {
      "epoch": 6.07,
      "learning_rate": 1.824004455583403e-05,
      "loss": 0.0015,
      "step": 2680
    },
    {
      "epoch": 6.08,
      "learning_rate": 1.8100807574491786e-05,
      "loss": 0.2167,
      "step": 2690
    },
    {
      "epoch": 6.08,
      "learning_rate": 1.796157059314954e-05,
      "loss": 0.5677,
      "step": 2700
    },
    {
      "epoch": 6.08,
      "learning_rate": 1.7822333611807297e-05,
      "loss": 0.0023,
      "step": 2710
    },
    {
      "epoch": 6.08,
      "learning_rate": 1.768309663046505e-05,
      "loss": 0.0018,
      "step": 2720
    },
    {
      "epoch": 6.09,
      "learning_rate": 1.7543859649122806e-05,
      "loss": 0.0217,
      "step": 2730
    },
    {
      "epoch": 6.09,
      "learning_rate": 1.740462266778056e-05,
      "loss": 0.0017,
      "step": 2740
    },
    {
      "epoch": 6.09,
      "learning_rate": 1.7265385686438317e-05,
      "loss": 0.0015,
      "step": 2750
    },
    {
      "epoch": 6.09,
      "learning_rate": 1.7126148705096073e-05,
      "loss": 0.265,
      "step": 2760
    },
    {
      "epoch": 6.1,
      "learning_rate": 1.698691172375383e-05,
      "loss": 0.0016,
      "step": 2770
    },
    {
      "epoch": 6.1,
      "learning_rate": 1.6847674742411585e-05,
      "loss": 0.0011,
      "step": 2780
    },
    {
      "epoch": 6.1,
      "eval_accuracy": 0.71,
      "eval_loss": 1.8265665769577026,
      "eval_runtime": 26.2083,
      "eval_samples_per_second": 3.816,
      "eval_steps_per_second": 3.816,
      "step": 2786
    },
    {
      "epoch": 7.0,
      "learning_rate": 1.670843776106934e-05,
      "loss": 0.0023,
      "step": 2790
    },
    {
      "epoch": 7.0,
      "learning_rate": 1.6569200779727097e-05,
      "loss": 0.0036,
      "step": 2800
    },
    {
      "epoch": 7.01,
      "learning_rate": 1.6429963798384852e-05,
      "loss": 0.0022,
      "step": 2810
    },
    {
      "epoch": 7.01,
      "learning_rate": 1.6290726817042605e-05,
      "loss": 0.0016,
      "step": 2820
    },
    {
      "epoch": 7.01,
      "learning_rate": 1.615148983570036e-05,
      "loss": 0.0029,
      "step": 2830
    },
    {
      "epoch": 7.01,
      "learning_rate": 1.6012252854358116e-05,
      "loss": 0.7778,
      "step": 2840
    },
    {
      "epoch": 7.02,
      "learning_rate": 1.5873015873015872e-05,
      "loss": 0.0014,
      "step": 2850
    },
    {
      "epoch": 7.02,
      "learning_rate": 1.5733778891673628e-05,
      "loss": 0.0317,
      "step": 2860
    },
    {
      "epoch": 7.02,
      "learning_rate": 1.5594541910331384e-05,
      "loss": 0.0015,
      "step": 2870
    },
    {
      "epoch": 7.02,
      "learning_rate": 1.545530492898914e-05,
      "loss": 0.3381,
      "step": 2880
    },
    {
      "epoch": 7.03,
      "learning_rate": 1.5316067947646896e-05,
      "loss": 0.0014,
      "step": 2890
    },
    {
      "epoch": 7.03,
      "learning_rate": 1.517683096630465e-05,
      "loss": 0.0014,
      "step": 2900
    },
    {
      "epoch": 7.03,
      "learning_rate": 1.5037593984962406e-05,
      "loss": 0.0017,
      "step": 2910
    },
    {
      "epoch": 7.03,
      "learning_rate": 1.4898357003620161e-05,
      "loss": 0.7069,
      "step": 2920
    },
    {
      "epoch": 7.04,
      "learning_rate": 1.4759120022277917e-05,
      "loss": 0.0016,
      "step": 2930
    },
    {
      "epoch": 7.04,
      "learning_rate": 1.4619883040935673e-05,
      "loss": 0.0069,
      "step": 2940
    },
    {
      "epoch": 7.04,
      "learning_rate": 1.4480646059593427e-05,
      "loss": 0.0411,
      "step": 2950
    },
    {
      "epoch": 7.04,
      "learning_rate": 1.4341409078251183e-05,
      "loss": 0.0076,
      "step": 2960
    },
    {
      "epoch": 7.05,
      "learning_rate": 1.4202172096908939e-05,
      "loss": 0.0174,
      "step": 2970
    },
    {
      "epoch": 7.05,
      "learning_rate": 1.4062935115566695e-05,
      "loss": 0.007,
      "step": 2980
    },
    {
      "epoch": 7.05,
      "learning_rate": 1.392369813422445e-05,
      "loss": 0.7089,
      "step": 2990
    },
    {
      "epoch": 7.05,
      "learning_rate": 1.3784461152882205e-05,
      "loss": 0.2224,
      "step": 3000
    },
    {
      "epoch": 7.06,
      "learning_rate": 1.364522417153996e-05,
      "loss": 0.0039,
      "step": 3010
    },
    {
      "epoch": 7.06,
      "learning_rate": 1.3505987190197716e-05,
      "loss": 0.0011,
      "step": 3020
    },
    {
      "epoch": 7.06,
      "learning_rate": 1.3366750208855472e-05,
      "loss": 0.0014,
      "step": 3030
    },
    {
      "epoch": 7.06,
      "learning_rate": 1.3227513227513228e-05,
      "loss": 0.0749,
      "step": 3040
    },
    {
      "epoch": 7.07,
      "learning_rate": 1.3088276246170982e-05,
      "loss": 0.7371,
      "step": 3050
    },
    {
      "epoch": 7.07,
      "learning_rate": 1.2949039264828738e-05,
      "loss": 0.0012,
      "step": 3060
    },
    {
      "epoch": 7.07,
      "learning_rate": 1.2809802283486494e-05,
      "loss": 0.6728,
      "step": 3070
    },
    {
      "epoch": 7.07,
      "learning_rate": 1.267056530214425e-05,
      "loss": 0.0017,
      "step": 3080
    },
    {
      "epoch": 7.08,
      "learning_rate": 1.2531328320802006e-05,
      "loss": 0.0018,
      "step": 3090
    },
    {
      "epoch": 7.08,
      "learning_rate": 1.239209133945976e-05,
      "loss": 0.0038,
      "step": 3100
    },
    {
      "epoch": 7.08,
      "learning_rate": 1.2252854358117516e-05,
      "loss": 0.0014,
      "step": 3110
    },
    {
      "epoch": 7.08,
      "learning_rate": 1.2113617376775271e-05,
      "loss": 0.0026,
      "step": 3120
    },
    {
      "epoch": 7.09,
      "learning_rate": 1.1974380395433027e-05,
      "loss": 0.0014,
      "step": 3130
    },
    {
      "epoch": 7.09,
      "learning_rate": 1.1835143414090783e-05,
      "loss": 0.0013,
      "step": 3140
    },
    {
      "epoch": 7.09,
      "learning_rate": 1.1695906432748537e-05,
      "loss": 0.003,
      "step": 3150
    },
    {
      "epoch": 7.09,
      "learning_rate": 1.1556669451406293e-05,
      "loss": 0.0014,
      "step": 3160
    },
    {
      "epoch": 7.1,
      "learning_rate": 1.1417432470064049e-05,
      "loss": 0.0018,
      "step": 3170
    },
    {
      "epoch": 7.1,
      "learning_rate": 1.1278195488721805e-05,
      "loss": 0.0951,
      "step": 3180
    },
    {
      "epoch": 7.1,
      "eval_accuracy": 0.78,
      "eval_loss": 1.5909953117370605,
      "eval_runtime": 27.0785,
      "eval_samples_per_second": 3.693,
      "eval_steps_per_second": 3.693,
      "step": 3184
    },
    {
      "epoch": 8.0,
      "learning_rate": 1.113895850737956e-05,
      "loss": 0.0011,
      "step": 3190
    },
    {
      "epoch": 8.0,
      "learning_rate": 1.0999721526037315e-05,
      "loss": 0.0011,
      "step": 3200
    },
    {
      "epoch": 8.01,
      "learning_rate": 1.086048454469507e-05,
      "loss": 0.0016,
      "step": 3210
    },
    {
      "epoch": 8.01,
      "learning_rate": 1.0721247563352826e-05,
      "loss": 0.0014,
      "step": 3220
    },
    {
      "epoch": 8.01,
      "learning_rate": 1.0582010582010582e-05,
      "loss": 0.0011,
      "step": 3230
    },
    {
      "epoch": 8.01,
      "learning_rate": 1.0442773600668338e-05,
      "loss": 0.0015,
      "step": 3240
    },
    {
      "epoch": 8.02,
      "learning_rate": 1.0303536619326092e-05,
      "loss": 0.0023,
      "step": 3250
    },
    {
      "epoch": 8.02,
      "learning_rate": 1.0164299637983848e-05,
      "loss": 0.0021,
      "step": 3260
    },
    {
      "epoch": 8.02,
      "learning_rate": 1.0025062656641604e-05,
      "loss": 0.0013,
      "step": 3270
    },
    {
      "epoch": 8.02,
      "learning_rate": 9.88582567529936e-06,
      "loss": 0.0011,
      "step": 3280
    },
    {
      "epoch": 8.03,
      "learning_rate": 9.746588693957115e-06,
      "loss": 0.0011,
      "step": 3290
    },
    {
      "epoch": 8.03,
      "learning_rate": 9.607351712614871e-06,
      "loss": 0.0013,
      "step": 3300
    },
    {
      "epoch": 8.03,
      "learning_rate": 9.468114731272625e-06,
      "loss": 0.0011,
      "step": 3310
    },
    {
      "epoch": 8.03,
      "learning_rate": 9.328877749930381e-06,
      "loss": 0.0563,
      "step": 3320
    },
    {
      "epoch": 8.04,
      "learning_rate": 9.189640768588137e-06,
      "loss": 0.0018,
      "step": 3330
    },
    {
      "epoch": 8.04,
      "learning_rate": 9.050403787245893e-06,
      "loss": 0.002,
      "step": 3340
    },
    {
      "epoch": 8.04,
      "learning_rate": 8.911166805903649e-06,
      "loss": 0.0016,
      "step": 3350
    },
    {
      "epoch": 8.04,
      "learning_rate": 8.771929824561403e-06,
      "loss": 0.0017,
      "step": 3360
    },
    {
      "epoch": 8.05,
      "learning_rate": 8.632692843219159e-06,
      "loss": 0.0013,
      "step": 3370
    },
    {
      "epoch": 8.05,
      "learning_rate": 8.493455861876915e-06,
      "loss": 0.0015,
      "step": 3380
    },
    {
      "epoch": 8.05,
      "learning_rate": 8.35421888053467e-06,
      "loss": 0.0013,
      "step": 3390
    },
    {
      "epoch": 8.05,
      "learning_rate": 8.214981899192426e-06,
      "loss": 0.0011,
      "step": 3400
    },
    {
      "epoch": 8.06,
      "learning_rate": 8.07574491785018e-06,
      "loss": 0.0012,
      "step": 3410
    },
    {
      "epoch": 8.06,
      "learning_rate": 7.936507936507936e-06,
      "loss": 0.0015,
      "step": 3420
    },
    {
      "epoch": 8.06,
      "learning_rate": 7.797270955165692e-06,
      "loss": 0.0015,
      "step": 3430
    },
    {
      "epoch": 8.06,
      "learning_rate": 7.658033973823448e-06,
      "loss": 0.0015,
      "step": 3440
    },
    {
      "epoch": 8.07,
      "learning_rate": 7.518796992481203e-06,
      "loss": 1.1364,
      "step": 3450
    },
    {
      "epoch": 8.07,
      "learning_rate": 7.379560011138959e-06,
      "loss": 0.0014,
      "step": 3460
    },
    {
      "epoch": 8.07,
      "learning_rate": 7.240323029796714e-06,
      "loss": 0.0016,
      "step": 3470
    },
    {
      "epoch": 8.07,
      "learning_rate": 7.1010860484544695e-06,
      "loss": 0.0013,
      "step": 3480
    },
    {
      "epoch": 8.08,
      "learning_rate": 6.961849067112225e-06,
      "loss": 0.0015,
      "step": 3490
    },
    {
      "epoch": 8.08,
      "learning_rate": 6.82261208576998e-06,
      "loss": 0.0012,
      "step": 3500
    },
    {
      "epoch": 8.08,
      "learning_rate": 6.683375104427736e-06,
      "loss": 0.0014,
      "step": 3510
    },
    {
      "epoch": 8.08,
      "learning_rate": 6.544138123085491e-06,
      "loss": 0.001,
      "step": 3520
    },
    {
      "epoch": 8.09,
      "learning_rate": 6.404901141743247e-06,
      "loss": 0.0009,
      "step": 3530
    },
    {
      "epoch": 8.09,
      "learning_rate": 6.265664160401003e-06,
      "loss": 0.0014,
      "step": 3540
    },
    {
      "epoch": 8.09,
      "learning_rate": 6.126427179058758e-06,
      "loss": 0.0014,
      "step": 3550
    },
    {
      "epoch": 8.09,
      "learning_rate": 5.987190197716514e-06,
      "loss": 0.0016,
      "step": 3560
    },
    {
      "epoch": 8.1,
      "learning_rate": 5.8479532163742686e-06,
      "loss": 0.0014,
      "step": 3570
    },
    {
      "epoch": 8.1,
      "learning_rate": 5.708716235032024e-06,
      "loss": 0.4047,
      "step": 3580
    },
    {
      "epoch": 8.1,
      "eval_accuracy": 0.7,
      "eval_loss": 1.9999462366104126,
      "eval_runtime": 28.4623,
      "eval_samples_per_second": 3.513,
      "eval_steps_per_second": 3.513,
      "step": 3582
    },
    {
      "epoch": 9.0,
      "learning_rate": 5.56947925368978e-06,
      "loss": 0.0014,
      "step": 3590
    },
    {
      "epoch": 9.0,
      "learning_rate": 5.430242272347535e-06,
      "loss": 0.0013,
      "step": 3600
    },
    {
      "epoch": 9.01,
      "learning_rate": 5.291005291005291e-06,
      "loss": 0.0354,
      "step": 3610
    },
    {
      "epoch": 9.01,
      "learning_rate": 5.151768309663046e-06,
      "loss": 0.0011,
      "step": 3620
    },
    {
      "epoch": 9.01,
      "learning_rate": 5.012531328320802e-06,
      "loss": 0.576,
      "step": 3630
    },
    {
      "epoch": 9.01,
      "learning_rate": 4.873294346978558e-06,
      "loss": 0.002,
      "step": 3640
    },
    {
      "epoch": 9.02,
      "learning_rate": 4.734057365636313e-06,
      "loss": 0.0009,
      "step": 3650
    },
    {
      "epoch": 9.02,
      "learning_rate": 4.5948203842940685e-06,
      "loss": 0.048,
      "step": 3660
    },
    {
      "epoch": 9.02,
      "learning_rate": 4.455583402951824e-06,
      "loss": 0.0017,
      "step": 3670
    },
    {
      "epoch": 9.02,
      "learning_rate": 4.316346421609579e-06,
      "loss": 0.0011,
      "step": 3680
    },
    {
      "epoch": 9.03,
      "learning_rate": 4.177109440267335e-06,
      "loss": 0.7831,
      "step": 3690
    },
    {
      "epoch": 9.03,
      "learning_rate": 4.03787245892509e-06,
      "loss": 0.001,
      "step": 3700
    },
    {
      "epoch": 9.03,
      "learning_rate": 3.898635477582846e-06,
      "loss": 0.0011,
      "step": 3710
    },
    {
      "epoch": 9.03,
      "learning_rate": 3.7593984962406014e-06,
      "loss": 0.4288,
      "step": 3720
    },
    {
      "epoch": 9.04,
      "learning_rate": 3.620161514898357e-06,
      "loss": 0.001,
      "step": 3730
    },
    {
      "epoch": 9.04,
      "learning_rate": 3.4809245335561126e-06,
      "loss": 0.001,
      "step": 3740
    },
    {
      "epoch": 9.04,
      "learning_rate": 3.341687552213868e-06,
      "loss": 0.0012,
      "step": 3750
    },
    {
      "epoch": 9.04,
      "learning_rate": 3.2024505708716235e-06,
      "loss": 0.0012,
      "step": 3760
    },
    {
      "epoch": 9.05,
      "learning_rate": 3.063213589529379e-06,
      "loss": 0.0227,
      "step": 3770
    },
    {
      "epoch": 9.05,
      "learning_rate": 2.9239766081871343e-06,
      "loss": 0.0012,
      "step": 3780
    },
    {
      "epoch": 9.05,
      "learning_rate": 2.78473962684489e-06,
      "loss": 0.001,
      "step": 3790
    },
    {
      "epoch": 9.05,
      "learning_rate": 2.6455026455026455e-06,
      "loss": 0.001,
      "step": 3800
    },
    {
      "epoch": 9.06,
      "learning_rate": 2.506265664160401e-06,
      "loss": 0.0014,
      "step": 3810
    },
    {
      "epoch": 9.06,
      "learning_rate": 2.3670286828181563e-06,
      "loss": 0.0012,
      "step": 3820
    },
    {
      "epoch": 9.06,
      "learning_rate": 2.227791701475912e-06,
      "loss": 0.0016,
      "step": 3830
    },
    {
      "epoch": 9.06,
      "learning_rate": 2.0885547201336676e-06,
      "loss": 0.0011,
      "step": 3840
    },
    {
      "epoch": 9.07,
      "learning_rate": 1.949317738791423e-06,
      "loss": 0.2977,
      "step": 3850
    },
    {
      "epoch": 9.07,
      "learning_rate": 1.8100807574491784e-06,
      "loss": 0.0011,
      "step": 3860
    },
    {
      "epoch": 9.07,
      "learning_rate": 1.670843776106934e-06,
      "loss": 0.0012,
      "step": 3870
    },
    {
      "epoch": 9.07,
      "learning_rate": 1.5316067947646894e-06,
      "loss": 0.0011,
      "step": 3880
    },
    {
      "epoch": 9.08,
      "learning_rate": 1.392369813422445e-06,
      "loss": 0.8924,
      "step": 3890
    },
    {
      "epoch": 9.08,
      "learning_rate": 1.2531328320802005e-06,
      "loss": 0.0011,
      "step": 3900
    },
    {
      "epoch": 9.08,
      "learning_rate": 1.113895850737956e-06,
      "loss": 0.0269,
      "step": 3910
    },
    {
      "epoch": 9.08,
      "learning_rate": 9.746588693957115e-07,
      "loss": 0.0018,
      "step": 3920
    },
    {
      "epoch": 9.09,
      "learning_rate": 8.35421888053467e-07,
      "loss": 0.001,
      "step": 3930
    },
    {
      "epoch": 9.09,
      "learning_rate": 6.961849067112225e-07,
      "loss": 0.0015,
      "step": 3940
    },
    {
      "epoch": 9.09,
      "learning_rate": 5.56947925368978e-07,
      "loss": 0.0011,
      "step": 3950
    },
    {
      "epoch": 9.09,
      "learning_rate": 4.177109440267335e-07,
      "loss": 0.0012,
      "step": 3960
    },
    {
      "epoch": 9.1,
      "learning_rate": 2.78473962684489e-07,
      "loss": 0.0013,
      "step": 3970
    },
    {
      "epoch": 9.1,
      "learning_rate": 1.392369813422445e-07,
      "loss": 0.0011,
      "step": 3980
    },
    {
      "epoch": 9.1,
      "eval_accuracy": 0.78,
      "eval_loss": 1.5902831554412842,
      "eval_runtime": 28.5194,
      "eval_samples_per_second": 3.506,
      "eval_steps_per_second": 3.506,
      "step": 3980
    },
    {
      "epoch": 10.0,
      "learning_rate": 0.0,
      "loss": 0.001,
      "step": 3990
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.78,
      "eval_loss": 1.5902682542800903,
      "eval_runtime": 30.2846,
      "eval_samples_per_second": 3.302,
      "eval_steps_per_second": 3.302,
      "step": 3990
    },
    {
      "epoch": 10.0,
      "step": 3990,
      "total_flos": 4.972285443087729e+18,
      "train_loss": 0.6824237227138776,
      "train_runtime": 3124.1401,
      "train_samples_per_second": 1.277,
      "train_steps_per_second": 1.277
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.8324022346368715,
      "eval_loss": 0.8124022483825684,
      "eval_runtime": 57.24,
      "eval_samples_per_second": 3.127,
      "eval_steps_per_second": 3.127,
      "step": 3990
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.8324022346368715,
      "eval_loss": 0.8124023079872131,
      "eval_runtime": 56.7655,
      "eval_samples_per_second": 3.153,
      "eval_steps_per_second": 3.153,
      "step": 3990
    }
  ],
  "max_steps": 3990,
  "num_train_epochs": 9223372036854775807,
  "total_flos": 4.972285443087729e+18,
  "trial_name": null,
  "trial_params": null
}