{
  "best_metric": 0.3000366985797882,
  "best_model_checkpoint": "./new_exper3/checkpoint-4200",
  "epoch": 8.0,
  "global_step": 5112,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 9.980438184663537e-05,
      "loss": 4.9498,
      "step": 10
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.960876369327074e-05,
      "loss": 4.8556,
      "step": 20
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.94131455399061e-05,
      "loss": 4.7389,
      "step": 30
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.921752738654147e-05,
      "loss": 4.6112,
      "step": 40
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.902190923317684e-05,
      "loss": 4.54,
      "step": 50
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.882629107981222e-05,
      "loss": 4.448,
      "step": 60
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.863067292644758e-05,
      "loss": 4.3441,
      "step": 70
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.843505477308295e-05,
      "loss": 4.2899,
      "step": 80
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.823943661971832e-05,
      "loss": 4.2878,
      "step": 90
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.804381846635369e-05,
      "loss": 4.093,
      "step": 100
    },
    {
      "epoch": 0.16,
      "eval_accuracy": 0.18851195065535853,
      "eval_loss": 4.104459285736084,
      "eval_runtime": 43.0936,
      "eval_samples_per_second": 60.195,
      "eval_steps_per_second": 7.542,
      "step": 100
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.784820031298904e-05,
      "loss": 4.1228,
      "step": 110
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.765258215962441e-05,
      "loss": 4.0244,
      "step": 120
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.745696400625978e-05,
      "loss": 3.9376,
      "step": 130
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.726134585289515e-05,
      "loss": 3.7948,
      "step": 140
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.706572769953052e-05,
      "loss": 3.8001,
      "step": 150
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.687010954616589e-05,
      "loss": 3.7539,
      "step": 160
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.667449139280126e-05,
      "loss": 3.6066,
      "step": 170
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.647887323943663e-05,
      "loss": 3.5595,
      "step": 180
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.628325508607199e-05,
      "loss": 3.4972,
      "step": 190
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.608763693270736e-05,
      "loss": 3.5057,
      "step": 200
    },
    {
      "epoch": 0.31,
      "eval_accuracy": 0.323053199691596,
      "eval_loss": 3.444770336151123,
      "eval_runtime": 43.0641,
      "eval_samples_per_second": 60.236,
      "eval_steps_per_second": 7.547,
      "step": 200
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.589201877934273e-05,
      "loss": 3.3105,
      "step": 210
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.56964006259781e-05,
      "loss": 3.4108,
      "step": 220
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.550078247261345e-05,
      "loss": 3.4285,
      "step": 230
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.530516431924882e-05,
      "loss": 3.213,
      "step": 240
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.510954616588421e-05,
      "loss": 3.1473,
      "step": 250
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.491392801251958e-05,
      "loss": 3.1898,
      "step": 260
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.471830985915493e-05,
      "loss": 3.1523,
      "step": 270
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.45226917057903e-05,
      "loss": 3.186,
      "step": 280
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.432707355242567e-05,
      "loss": 2.9603,
      "step": 290
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.413145539906104e-05,
      "loss": 2.9116,
      "step": 300
    },
    {
      "epoch": 0.47,
      "eval_accuracy": 0.45373939861218193,
      "eval_loss": 2.9483001232147217,
      "eval_runtime": 42.4599,
      "eval_samples_per_second": 61.093,
      "eval_steps_per_second": 7.654,
      "step": 300
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.39358372456964e-05,
      "loss": 3.0387,
      "step": 310
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.374021909233177e-05,
      "loss": 2.8818,
      "step": 320
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.354460093896714e-05,
      "loss": 2.9362,
      "step": 330
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.334898278560251e-05,
      "loss": 2.9021,
      "step": 340
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.315336463223788e-05,
      "loss": 2.7243,
      "step": 350
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.295774647887325e-05,
      "loss": 2.6124,
      "step": 360
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.276212832550862e-05,
      "loss": 2.6288,
      "step": 370
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.256651017214399e-05,
      "loss": 2.6455,
      "step": 380
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.237089201877934e-05,
      "loss": 2.6136,
      "step": 390
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.217527386541471e-05,
      "loss": 2.561,
      "step": 400
    },
    {
      "epoch": 0.63,
      "eval_accuracy": 0.525828835774865,
      "eval_loss": 2.5700132846832275,
      "eval_runtime": 42.7904,
      "eval_samples_per_second": 60.621,
      "eval_steps_per_second": 7.595,
      "step": 400
    },
    {
      "epoch": 0.64,
      "learning_rate": 9.197965571205008e-05,
      "loss": 2.5745,
      "step": 410
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.178403755868545e-05,
      "loss": 2.5422,
      "step": 420
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.158841940532081e-05,
      "loss": 2.4631,
      "step": 430
    },
    {
      "epoch": 0.69,
      "learning_rate": 9.139280125195618e-05,
      "loss": 2.3095,
      "step": 440
    },
    {
      "epoch": 0.7,
      "learning_rate": 9.119718309859156e-05,
      "loss": 2.3756,
      "step": 450
    },
    {
      "epoch": 0.72,
      "learning_rate": 9.100156494522693e-05,
      "loss": 2.485,
      "step": 460
    },
    {
      "epoch": 0.74,
      "learning_rate": 9.080594679186229e-05,
      "loss": 2.3036,
      "step": 470
    },
    {
      "epoch": 0.75,
      "learning_rate": 9.061032863849766e-05,
      "loss": 2.2608,
      "step": 480
    },
    {
      "epoch": 0.77,
      "learning_rate": 9.041471048513303e-05,
      "loss": 2.296,
      "step": 490
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.02190923317684e-05,
      "loss": 2.1611,
      "step": 500
    },
    {
      "epoch": 0.78,
      "eval_accuracy": 0.6144949884348496,
      "eval_loss": 2.172065258026123,
      "eval_runtime": 42.9818,
      "eval_samples_per_second": 60.351,
      "eval_steps_per_second": 7.561,
      "step": 500
    },
    {
      "epoch": 0.8,
      "learning_rate": 9.002347417840375e-05,
      "loss": 2.0372,
      "step": 510
    },
    {
      "epoch": 0.81,
      "learning_rate": 8.982785602503912e-05,
      "loss": 2.0652,
      "step": 520
    },
    {
      "epoch": 0.83,
      "learning_rate": 8.963223787167449e-05,
      "loss": 2.1529,
      "step": 530
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.943661971830986e-05,
      "loss": 1.9779,
      "step": 540
    },
    {
      "epoch": 0.86,
      "learning_rate": 8.924100156494523e-05,
      "loss": 1.9492,
      "step": 550
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.90453834115806e-05,
      "loss": 1.9624,
      "step": 560
    },
    {
      "epoch": 0.89,
      "learning_rate": 8.884976525821597e-05,
      "loss": 1.8139,
      "step": 570
    },
    {
      "epoch": 0.91,
      "learning_rate": 8.865414710485134e-05,
      "loss": 1.9124,
      "step": 580
    },
    {
      "epoch": 0.92,
      "learning_rate": 8.845852895148671e-05,
      "loss": 1.7514,
      "step": 590
    },
    {
      "epoch": 0.94,
      "learning_rate": 8.826291079812207e-05,
      "loss": 1.715,
      "step": 600
    },
    {
      "epoch": 0.94,
      "eval_accuracy": 0.6407093292212799,
      "eval_loss": 1.8254655599594116,
      "eval_runtime": 42.9275,
      "eval_samples_per_second": 60.428,
      "eval_steps_per_second": 7.571,
      "step": 600
    },
    {
      "epoch": 0.95,
      "learning_rate": 8.806729264475744e-05,
      "loss": 1.864,
      "step": 610
    },
    {
      "epoch": 0.97,
      "learning_rate": 8.78716744913928e-05,
      "loss": 1.8157,
      "step": 620
    },
    {
      "epoch": 0.99,
      "learning_rate": 8.767605633802817e-05,
      "loss": 1.8333,
      "step": 630
    },
    {
      "epoch": 1.0,
      "learning_rate": 8.748043818466354e-05,
      "loss": 1.6364,
      "step": 640
    },
    {
      "epoch": 1.02,
      "learning_rate": 8.728482003129891e-05,
      "loss": 1.5205,
      "step": 650
    },
    {
      "epoch": 1.03,
      "learning_rate": 8.708920187793428e-05,
      "loss": 1.5193,
      "step": 660
    },
    {
      "epoch": 1.05,
      "learning_rate": 8.689358372456965e-05,
      "loss": 1.5602,
      "step": 670
    },
    {
      "epoch": 1.06,
      "learning_rate": 8.669796557120501e-05,
      "loss": 1.3509,
      "step": 680
    },
    {
      "epoch": 1.08,
      "learning_rate": 8.650234741784038e-05,
      "loss": 1.408,
      "step": 690
    },
    {
      "epoch": 1.1,
      "learning_rate": 8.630672926447575e-05,
      "loss": 1.2752,
      "step": 700
    },
    {
      "epoch": 1.1,
      "eval_accuracy": 0.70508866615266,
      "eval_loss": 1.5340100526809692,
      "eval_runtime": 42.6015,
      "eval_samples_per_second": 60.89,
      "eval_steps_per_second": 7.629,
      "step": 700
    },
    {
      "epoch": 1.11,
      "learning_rate": 8.611111111111112e-05,
      "loss": 1.184,
      "step": 710
    },
    {
      "epoch": 1.13,
      "learning_rate": 8.591549295774647e-05,
      "loss": 1.284,
      "step": 720
    },
    {
      "epoch": 1.14,
      "learning_rate": 8.571987480438184e-05,
      "loss": 1.2331,
      "step": 730
    },
    {
      "epoch": 1.16,
      "learning_rate": 8.552425665101721e-05,
      "loss": 1.2059,
      "step": 740
    },
    {
      "epoch": 1.17,
      "learning_rate": 8.53286384976526e-05,
      "loss": 1.3036,
      "step": 750
    },
    {
      "epoch": 1.19,
      "learning_rate": 8.513302034428795e-05,
      "loss": 1.2848,
      "step": 760
    },
    {
      "epoch": 1.21,
      "learning_rate": 8.493740219092332e-05,
      "loss": 1.073,
      "step": 770
    },
    {
      "epoch": 1.22,
      "learning_rate": 8.474178403755869e-05,
      "loss": 1.4211,
      "step": 780
    },
    {
      "epoch": 1.24,
      "learning_rate": 8.454616588419406e-05,
      "loss": 1.067,
      "step": 790
    },
    {
      "epoch": 1.25,
      "learning_rate": 8.435054773082942e-05,
      "loss": 1.2487,
      "step": 800
    },
    {
      "epoch": 1.25,
      "eval_accuracy": 0.7201233616037008,
      "eval_loss": 1.353263258934021,
      "eval_runtime": 42.5111,
      "eval_samples_per_second": 61.019,
      "eval_steps_per_second": 7.645,
      "step": 800
    },
    {
      "epoch": 1.27,
      "learning_rate": 8.415492957746479e-05,
      "loss": 1.2377,
      "step": 810
    },
    {
      "epoch": 1.28,
      "learning_rate": 8.395931142410016e-05,
      "loss": 1.1797,
      "step": 820
    },
    {
      "epoch": 1.3,
      "learning_rate": 8.376369327073553e-05,
      "loss": 1.2482,
      "step": 830
    },
    {
      "epoch": 1.31,
      "learning_rate": 8.35680751173709e-05,
      "loss": 1.0641,
      "step": 840
    },
    {
      "epoch": 1.33,
      "learning_rate": 8.337245696400627e-05,
      "loss": 1.14,
      "step": 850
    },
    {
      "epoch": 1.35,
      "learning_rate": 8.317683881064164e-05,
      "loss": 1.1587,
      "step": 860
    },
    {
      "epoch": 1.36,
      "learning_rate": 8.298122065727701e-05,
      "loss": 1.1029,
      "step": 870
    },
    {
      "epoch": 1.38,
      "learning_rate": 8.278560250391236e-05,
      "loss": 0.9988,
      "step": 880
    },
    {
      "epoch": 1.39,
      "learning_rate": 8.258998435054773e-05,
      "loss": 1.0692,
      "step": 890
    },
    {
      "epoch": 1.41,
      "learning_rate": 8.23943661971831e-05,
      "loss": 1.0333,
      "step": 900
    },
    {
      "epoch": 1.41,
      "eval_accuracy": 0.7825751734772552,
      "eval_loss": 1.1474497318267822,
      "eval_runtime": 43.2872,
      "eval_samples_per_second": 59.925,
      "eval_steps_per_second": 7.508,
      "step": 900
    },
    {
      "epoch": 1.42,
      "learning_rate": 8.219874804381847e-05,
      "loss": 1.0357,
      "step": 910
    },
    {
      "epoch": 1.44,
      "learning_rate": 8.200312989045383e-05,
      "loss": 0.8625,
      "step": 920
    },
    {
      "epoch": 1.46,
      "learning_rate": 8.18075117370892e-05,
      "loss": 1.0712,
      "step": 930
    },
    {
      "epoch": 1.47,
      "learning_rate": 8.161189358372458e-05,
      "loss": 1.1329,
      "step": 940
    },
    {
      "epoch": 1.49,
      "learning_rate": 8.141627543035995e-05,
      "loss": 1.0017,
      "step": 950
    },
    {
      "epoch": 1.5,
      "learning_rate": 8.122065727699531e-05,
      "loss": 1.1086,
      "step": 960
    },
    {
      "epoch": 1.52,
      "learning_rate": 8.102503912363068e-05,
      "loss": 0.8891,
      "step": 970
    },
    {
      "epoch": 1.53,
      "learning_rate": 8.082942097026605e-05,
      "loss": 0.8059,
      "step": 980
    },
    {
      "epoch": 1.55,
      "learning_rate": 8.063380281690142e-05,
      "loss": 1.0192,
      "step": 990
    },
    {
      "epoch": 1.56,
      "learning_rate": 8.043818466353677e-05,
      "loss": 0.8856,
      "step": 1000
    },
    {
      "epoch": 1.56,
      "eval_accuracy": 0.7644564379336931,
      "eval_loss": 1.0914219617843628,
      "eval_runtime": 42.8447,
      "eval_samples_per_second": 60.544,
      "eval_steps_per_second": 7.586,
      "step": 1000
    },
    {
      "epoch": 1.58,
      "learning_rate": 8.024256651017214e-05,
      "loss": 0.911,
      "step": 1010
    },
    {
      "epoch": 1.6,
      "learning_rate": 8.004694835680751e-05,
      "loss": 0.8939,
      "step": 1020
    },
    {
      "epoch": 1.61,
      "learning_rate": 7.985133020344288e-05,
      "loss": 0.7816,
      "step": 1030
    },
    {
      "epoch": 1.63,
      "learning_rate": 7.965571205007825e-05,
      "loss": 0.8397,
      "step": 1040
    },
    {
      "epoch": 1.64,
      "learning_rate": 7.946009389671362e-05,
      "loss": 0.8172,
      "step": 1050
    },
    {
      "epoch": 1.66,
      "learning_rate": 7.926447574334899e-05,
      "loss": 0.7408,
      "step": 1060
    },
    {
      "epoch": 1.67,
      "learning_rate": 7.906885758998436e-05,
      "loss": 0.6926,
      "step": 1070
    },
    {
      "epoch": 1.69,
      "learning_rate": 7.887323943661972e-05,
      "loss": 0.8984,
      "step": 1080
    },
    {
      "epoch": 1.71,
      "learning_rate": 7.867762128325509e-05,
      "loss": 0.7221,
      "step": 1090
    },
    {
      "epoch": 1.72,
      "learning_rate": 7.848200312989046e-05,
      "loss": 0.7512,
      "step": 1100
    },
    {
      "epoch": 1.72,
      "eval_accuracy": 0.8118735543562067,
      "eval_loss": 0.8893365263938904,
      "eval_runtime": 42.2991,
      "eval_samples_per_second": 61.325,
      "eval_steps_per_second": 7.683,
      "step": 1100
    },
    {
      "epoch": 1.74,
      "learning_rate": 7.828638497652583e-05,
      "loss": 0.7297,
      "step": 1110
    },
    {
      "epoch": 1.75,
      "learning_rate": 7.809076682316118e-05,
      "loss": 0.6638,
      "step": 1120
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.789514866979655e-05,
      "loss": 0.7419,
      "step": 1130
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.769953051643193e-05,
      "loss": 0.7263,
      "step": 1140
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.75039123630673e-05,
      "loss": 0.6707,
      "step": 1150
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.730829420970266e-05,
      "loss": 0.7675,
      "step": 1160
    },
    {
      "epoch": 1.83,
      "learning_rate": 7.711267605633803e-05,
      "loss": 0.9189,
      "step": 1170
    },
    {
      "epoch": 1.85,
      "learning_rate": 7.69170579029734e-05,
      "loss": 0.5663,
      "step": 1180
    },
    {
      "epoch": 1.86,
      "learning_rate": 7.672143974960877e-05,
      "loss": 0.679,
      "step": 1190
    },
    {
      "epoch": 1.88,
      "learning_rate": 7.652582159624414e-05,
      "loss": 0.747,
      "step": 1200
    },
    {
      "epoch": 1.88,
      "eval_accuracy": 0.8303777949113339,
      "eval_loss": 0.8370148539543152,
      "eval_runtime": 42.5695,
      "eval_samples_per_second": 60.936,
      "eval_steps_per_second": 7.635,
      "step": 1200
    },
    {
      "epoch": 1.89,
      "learning_rate": 7.63302034428795e-05,
      "loss": 0.8146,
      "step": 1210
    },
    {
      "epoch": 1.91,
      "learning_rate": 7.613458528951487e-05,
      "loss": 0.715,
      "step": 1220
    },
    {
      "epoch": 1.92,
      "learning_rate": 7.593896713615024e-05,
      "loss": 0.8337,
      "step": 1230
    },
    {
      "epoch": 1.94,
      "learning_rate": 7.57433489827856e-05,
      "loss": 0.6553,
      "step": 1240
    },
    {
      "epoch": 1.96,
      "learning_rate": 7.554773082942097e-05,
      "loss": 0.7805,
      "step": 1250
    },
    {
      "epoch": 1.97,
      "learning_rate": 7.535211267605634e-05,
      "loss": 0.6115,
      "step": 1260
    },
    {
      "epoch": 1.99,
      "learning_rate": 7.515649452269171e-05,
      "loss": 0.7046,
      "step": 1270
    },
    {
      "epoch": 2.0,
      "learning_rate": 7.496087636932708e-05,
      "loss": 0.551,
      "step": 1280
    },
    {
      "epoch": 2.02,
      "learning_rate": 7.476525821596244e-05,
      "loss": 0.4497,
      "step": 1290
    },
    {
      "epoch": 2.03,
      "learning_rate": 7.456964006259781e-05,
      "loss": 0.5082,
      "step": 1300
    },
    {
      "epoch": 2.03,
      "eval_accuracy": 0.856592135697764,
      "eval_loss": 0.7130723595619202,
      "eval_runtime": 43.188,
      "eval_samples_per_second": 60.063,
      "eval_steps_per_second": 7.525,
      "step": 1300
    },
    {
      "epoch": 2.05,
      "learning_rate": 7.437402190923318e-05,
      "loss": 0.347,
      "step": 1310
    },
    {
      "epoch": 2.07,
      "learning_rate": 7.417840375586855e-05,
      "loss": 0.463,
      "step": 1320
    },
    {
      "epoch": 2.08,
      "learning_rate": 7.398278560250392e-05,
      "loss": 0.4014,
      "step": 1330
    },
    {
      "epoch": 2.1,
      "learning_rate": 7.378716744913929e-05,
      "loss": 0.488,
      "step": 1340
    },
    {
      "epoch": 2.11,
      "learning_rate": 7.359154929577466e-05,
      "loss": 0.4239,
      "step": 1350
    },
    {
      "epoch": 2.13,
      "learning_rate": 7.339593114241003e-05,
      "loss": 0.4246,
      "step": 1360
    },
    {
      "epoch": 2.14,
      "learning_rate": 7.320031298904538e-05,
      "loss": 0.3698,
      "step": 1370
    },
    {
      "epoch": 2.16,
      "learning_rate": 7.300469483568075e-05,
      "loss": 0.4264,
      "step": 1380
    },
    {
      "epoch": 2.18,
      "learning_rate": 7.280907668231612e-05,
      "loss": 0.3167,
      "step": 1390
    },
    {
      "epoch": 2.19,
      "learning_rate": 7.261345852895149e-05,
      "loss": 0.4449,
      "step": 1400
    },
    {
      "epoch": 2.19,
      "eval_accuracy": 0.8546646106399384,
      "eval_loss": 0.6572707295417786,
      "eval_runtime": 43.2163,
      "eval_samples_per_second": 60.024,
      "eval_steps_per_second": 7.52,
      "step": 1400
    },
    {
      "epoch": 2.21,
      "learning_rate": 7.241784037558685e-05,
      "loss": 0.3504,
      "step": 1410
    },
    {
      "epoch": 2.22,
      "learning_rate": 7.222222222222222e-05,
      "loss": 0.397,
      "step": 1420
    },
    {
      "epoch": 2.24,
      "learning_rate": 7.202660406885759e-05,
      "loss": 0.4599,
      "step": 1430
    },
    {
      "epoch": 2.25,
      "learning_rate": 7.183098591549297e-05,
      "loss": 0.4151,
      "step": 1440
    },
    {
      "epoch": 2.27,
      "learning_rate": 7.163536776212833e-05,
      "loss": 0.3605,
      "step": 1450
    },
    {
      "epoch": 2.28,
      "learning_rate": 7.14397496087637e-05,
      "loss": 0.3248,
      "step": 1460
    },
    {
      "epoch": 2.3,
      "learning_rate": 7.124413145539907e-05,
      "loss": 0.32,
      "step": 1470
    },
    {
      "epoch": 2.32,
      "learning_rate": 7.104851330203444e-05,
      "loss": 0.4012,
      "step": 1480
    },
    {
      "epoch": 2.33,
      "learning_rate": 7.08528951486698e-05,
      "loss": 0.3526,
      "step": 1490
    },
    {
      "epoch": 2.35,
      "learning_rate": 7.065727699530516e-05,
      "loss": 0.2912,
      "step": 1500
    },
    {
      "epoch": 2.35,
      "eval_accuracy": 0.8596761757902853,
      "eval_loss": 0.6183947324752808,
      "eval_runtime": 43.0374,
      "eval_samples_per_second": 60.273,
      "eval_steps_per_second": 7.552,
      "step": 1500
    },
    {
      "epoch": 2.36,
      "learning_rate": 7.046165884194053e-05,
      "loss": 0.427,
      "step": 1510
    },
    {
      "epoch": 2.38,
      "learning_rate": 7.02660406885759e-05,
      "loss": 0.4321,
      "step": 1520
    },
    {
      "epoch": 2.39,
      "learning_rate": 7.007042253521127e-05,
      "loss": 0.363,
      "step": 1530
    },
    {
      "epoch": 2.41,
      "learning_rate": 6.987480438184664e-05,
      "loss": 0.2761,
      "step": 1540
    },
    {
      "epoch": 2.43,
      "learning_rate": 6.967918622848201e-05,
      "loss": 0.3189,
      "step": 1550
    },
    {
      "epoch": 2.44,
      "learning_rate": 6.948356807511738e-05,
      "loss": 0.3227,
      "step": 1560
    },
    {
      "epoch": 2.46,
      "learning_rate": 6.928794992175274e-05,
      "loss": 0.2792,
      "step": 1570
    },
    {
      "epoch": 2.47,
      "learning_rate": 6.909233176838811e-05,
      "loss": 0.1959,
      "step": 1580
    },
    {
      "epoch": 2.49,
      "learning_rate": 6.889671361502348e-05,
      "loss": 0.2785,
      "step": 1590
    },
    {
      "epoch": 2.5,
      "learning_rate": 6.870109546165885e-05,
      "loss": 0.285,
      "step": 1600
    },
    {
      "epoch": 2.5,
      "eval_accuracy": 0.8569776407093292,
      "eval_loss": 0.5973872542381287,
      "eval_runtime": 42.9634,
      "eval_samples_per_second": 60.377,
      "eval_steps_per_second": 7.565,
      "step": 1600
    },
    {
      "epoch": 2.52,
      "learning_rate": 6.85054773082942e-05,
      "loss": 0.2649,
      "step": 1610
    },
    {
      "epoch": 2.54,
      "learning_rate": 6.830985915492957e-05,
      "loss": 0.3259,
      "step": 1620
    },
    {
      "epoch": 2.55,
      "learning_rate": 6.811424100156496e-05,
      "loss": 0.3085,
      "step": 1630
    },
    {
      "epoch": 2.57,
      "learning_rate": 6.791862284820033e-05,
      "loss": 0.3485,
      "step": 1640
    },
    {
      "epoch": 2.58,
      "learning_rate": 6.772300469483568e-05,
      "loss": 0.2735,
      "step": 1650
    },
    {
      "epoch": 2.6,
      "learning_rate": 6.752738654147105e-05,
      "loss": 0.3112,
      "step": 1660
    },
    {
      "epoch": 2.61,
      "learning_rate": 6.733176838810642e-05,
      "loss": 0.3373,
      "step": 1670
    },
    {
      "epoch": 2.63,
      "learning_rate": 6.713615023474179e-05,
      "loss": 0.1682,
      "step": 1680
    },
    {
      "epoch": 2.64,
      "learning_rate": 6.694053208137715e-05,
      "loss": 0.2515,
      "step": 1690
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.674491392801252e-05,
      "loss": 0.2267,
      "step": 1700
    },
    {
      "epoch": 2.66,
      "eval_accuracy": 0.8646877409406323,
      "eval_loss": 0.5621365904808044,
      "eval_runtime": 43.0986,
      "eval_samples_per_second": 60.188,
      "eval_steps_per_second": 7.541,
      "step": 1700
    },
    {
      "epoch": 2.68,
      "learning_rate": 6.654929577464789e-05,
      "loss": 0.1861,
      "step": 1710
    },
    {
      "epoch": 2.69,
      "learning_rate": 6.635367762128326e-05,
      "loss": 0.3325,
      "step": 1720
    },
    {
      "epoch": 2.71,
      "learning_rate": 6.615805946791863e-05,
      "loss": 0.3621,
      "step": 1730
    },
    {
      "epoch": 2.72,
      "learning_rate": 6.5962441314554e-05,
      "loss": 0.3046,
      "step": 1740
    },
    {
      "epoch": 2.74,
      "learning_rate": 6.576682316118937e-05,
      "loss": 0.347,
      "step": 1750
    },
    {
      "epoch": 2.75,
      "learning_rate": 6.557120500782473e-05,
      "loss": 0.3916,
      "step": 1760
    },
    {
      "epoch": 2.77,
      "learning_rate": 6.53755868544601e-05,
      "loss": 0.3221,
      "step": 1770
    },
    {
      "epoch": 2.79,
      "learning_rate": 6.517996870109546e-05,
      "loss": 0.1877,
      "step": 1780
    },
    {
      "epoch": 2.8,
      "learning_rate": 6.498435054773083e-05,
      "loss": 0.3638,
      "step": 1790
    },
    {
      "epoch": 2.82,
      "learning_rate": 6.47887323943662e-05,
      "loss": 0.2553,
      "step": 1800
    },
    {
      "epoch": 2.82,
      "eval_accuracy": 0.8816499614494988,
      "eval_loss": 0.5043683052062988,
      "eval_runtime": 42.7859,
      "eval_samples_per_second": 60.628,
      "eval_steps_per_second": 7.596,
      "step": 1800
    },
    {
      "epoch": 2.83,
      "learning_rate": 6.459311424100157e-05,
      "loss": 0.3439,
      "step": 1810
    },
    {
      "epoch": 2.85,
      "learning_rate": 6.439749608763693e-05,
      "loss": 0.2463,
      "step": 1820
    },
    {
      "epoch": 2.86,
      "learning_rate": 6.420187793427231e-05,
      "loss": 0.2192,
      "step": 1830
    },
    {
      "epoch": 2.88,
      "learning_rate": 6.400625978090768e-05,
      "loss": 0.236,
      "step": 1840
    },
    {
      "epoch": 2.9,
      "learning_rate": 6.381064162754305e-05,
      "loss": 0.2441,
      "step": 1850
    },
    {
      "epoch": 2.91,
      "learning_rate": 6.36150234741784e-05,
      "loss": 0.2125,
      "step": 1860
    },
    {
      "epoch": 2.93,
      "learning_rate": 6.341940532081377e-05,
      "loss": 0.2112,
      "step": 1870
    },
    {
      "epoch": 2.94,
      "learning_rate": 6.322378716744914e-05,
      "loss": 0.2905,
      "step": 1880
    },
    {
      "epoch": 2.96,
      "learning_rate": 6.302816901408451e-05,
      "loss": 0.3244,
      "step": 1890
    },
    {
      "epoch": 2.97,
      "learning_rate": 6.283255086071987e-05,
      "loss": 0.2029,
      "step": 1900
    },
    {
      "epoch": 2.97,
      "eval_accuracy": 0.8955281418658443,
      "eval_loss": 0.43422141671180725,
      "eval_runtime": 42.7572,
      "eval_samples_per_second": 60.668,
      "eval_steps_per_second": 7.601,
      "step": 1900
    },
    {
      "epoch": 2.99,
      "learning_rate": 6.263693270735524e-05,
      "loss": 0.2284,
      "step": 1910
    },
    {
      "epoch": 3.0,
      "learning_rate": 6.244131455399061e-05,
      "loss": 0.2399,
      "step": 1920
    },
    {
      "epoch": 3.02,
      "learning_rate": 6.224569640062598e-05,
      "loss": 0.1875,
      "step": 1930
    },
    {
      "epoch": 3.04,
      "learning_rate": 6.205007824726135e-05,
      "loss": 0.1721,
      "step": 1940
    },
    {
      "epoch": 3.05,
      "learning_rate": 6.185446009389672e-05,
      "loss": 0.2115,
      "step": 1950
    },
    {
      "epoch": 3.07,
      "learning_rate": 6.165884194053209e-05,
      "loss": 0.1698,
      "step": 1960
    },
    {
      "epoch": 3.08,
      "learning_rate": 6.146322378716746e-05,
      "loss": 0.1321,
      "step": 1970
    },
    {
      "epoch": 3.1,
      "learning_rate": 6.126760563380281e-05,
      "loss": 0.1165,
      "step": 1980
    },
    {
      "epoch": 3.11,
      "learning_rate": 6.107198748043818e-05,
      "loss": 0.1121,
      "step": 1990
    },
    {
      "epoch": 3.13,
      "learning_rate": 6.0876369327073554e-05,
      "loss": 0.1763,
      "step": 2000
    },
    {
      "epoch": 3.13,
      "eval_accuracy": 0.8905165767154973,
      "eval_loss": 0.44871243834495544,
      "eval_runtime": 42.6933,
      "eval_samples_per_second": 60.759,
      "eval_steps_per_second": 7.612,
      "step": 2000
    },
    {
      "epoch": 3.15,
      "learning_rate": 6.068075117370893e-05,
      "loss": 0.1473,
      "step": 2010
    },
    {
      "epoch": 3.16,
      "learning_rate": 6.0485133020344286e-05,
      "loss": 0.1697,
      "step": 2020
    },
    {
      "epoch": 3.18,
      "learning_rate": 6.0289514866979656e-05,
      "loss": 0.1712,
      "step": 2030
    },
    {
      "epoch": 3.19,
      "learning_rate": 6.0093896713615026e-05,
      "loss": 0.1468,
      "step": 2040
    },
    {
      "epoch": 3.21,
      "learning_rate": 5.9898278560250395e-05,
      "loss": 0.1395,
      "step": 2050
    },
    {
      "epoch": 3.22,
      "learning_rate": 5.970266040688576e-05,
      "loss": 0.1122,
      "step": 2060
    },
    {
      "epoch": 3.24,
      "learning_rate": 5.950704225352113e-05,
      "loss": 0.1426,
      "step": 2070
    },
    {
      "epoch": 3.26,
      "learning_rate": 5.93114241001565e-05,
      "loss": 0.111,
      "step": 2080
    },
    {
      "epoch": 3.27,
      "learning_rate": 5.911580594679187e-05,
      "loss": 0.141,
      "step": 2090
    },
    {
      "epoch": 3.29,
      "learning_rate": 5.892018779342723e-05,
      "loss": 0.1418,
      "step": 2100
    },
    {
      "epoch": 3.29,
      "eval_accuracy": 0.9005397070161912,
      "eval_loss": 0.41731029748916626,
      "eval_runtime": 42.6624,
      "eval_samples_per_second": 60.803,
      "eval_steps_per_second": 7.618,
      "step": 2100
    },
    {
      "epoch": 3.3,
      "learning_rate": 5.87245696400626e-05,
      "loss": 0.1462,
      "step": 2110
    },
    {
      "epoch": 3.32,
      "learning_rate": 5.852895148669797e-05,
      "loss": 0.0985,
      "step": 2120
    },
    {
      "epoch": 3.33,
      "learning_rate": 5.833333333333334e-05,
      "loss": 0.133,
      "step": 2130
    },
    {
      "epoch": 3.35,
      "learning_rate": 5.81377151799687e-05,
      "loss": 0.1414,
      "step": 2140
    },
    {
      "epoch": 3.36,
      "learning_rate": 5.794209702660407e-05,
      "loss": 0.1297,
      "step": 2150
    },
    {
      "epoch": 3.38,
      "learning_rate": 5.774647887323944e-05,
      "loss": 0.1346,
      "step": 2160
    },
    {
      "epoch": 3.4,
      "learning_rate": 5.755086071987481e-05,
      "loss": 0.1237,
      "step": 2170
    },
    {
      "epoch": 3.41,
      "learning_rate": 5.735524256651017e-05,
      "loss": 0.1664,
      "step": 2180
    },
    {
      "epoch": 3.43,
      "learning_rate": 5.715962441314554e-05,
      "loss": 0.0701,
      "step": 2190
    },
    {
      "epoch": 3.44,
      "learning_rate": 5.6964006259780914e-05,
      "loss": 0.0563,
      "step": 2200
    },
    {
      "epoch": 3.44,
      "eval_accuracy": 0.9047802621434079,
      "eval_loss": 0.387023389339447,
      "eval_runtime": 42.2972,
      "eval_samples_per_second": 61.328,
      "eval_steps_per_second": 7.684,
      "step": 2200
    },
    {
      "epoch": 3.46,
      "learning_rate": 5.6768388106416284e-05,
      "loss": 0.0832,
      "step": 2210
    },
    {
      "epoch": 3.47,
      "learning_rate": 5.657276995305164e-05,
      "loss": 0.0966,
      "step": 2220
    },
    {
      "epoch": 3.49,
      "learning_rate": 5.637715179968701e-05,
      "loss": 0.0926,
      "step": 2230
    },
    {
      "epoch": 3.51,
      "learning_rate": 5.618153364632238e-05,
      "loss": 0.1285,
      "step": 2240
    },
    {
      "epoch": 3.52,
      "learning_rate": 5.598591549295775e-05,
      "loss": 0.0809,
      "step": 2250
    },
    {
      "epoch": 3.54,
      "learning_rate": 5.579029733959311e-05,
      "loss": 0.1119,
      "step": 2260
    },
    {
      "epoch": 3.55,
      "learning_rate": 5.559467918622848e-05,
      "loss": 0.067,
      "step": 2270
    },
    {
      "epoch": 3.57,
      "learning_rate": 5.539906103286385e-05,
      "loss": 0.1022,
      "step": 2280
    },
    {
      "epoch": 3.58,
      "learning_rate": 5.520344287949922e-05,
      "loss": 0.1318,
      "step": 2290
    },
    {
      "epoch": 3.6,
      "learning_rate": 5.5007824726134584e-05,
      "loss": 0.0579,
      "step": 2300
    },
    {
      "epoch": 3.6,
      "eval_accuracy": 0.9036237471087124,
      "eval_loss": 0.38491636514663696,
      "eval_runtime": 42.9889,
      "eval_samples_per_second": 60.341,
      "eval_steps_per_second": 7.56,
      "step": 2300
    },
    {
      "epoch": 3.62,
      "learning_rate": 5.4812206572769954e-05,
      "loss": 0.1294,
      "step": 2310
    },
    {
      "epoch": 3.63,
      "learning_rate": 5.461658841940532e-05,
      "loss": 0.0777,
      "step": 2320
    },
    {
      "epoch": 3.65,
      "learning_rate": 5.442097026604069e-05,
      "loss": 0.0754,
      "step": 2330
    },
    {
      "epoch": 3.66,
      "learning_rate": 5.422535211267606e-05,
      "loss": 0.1463,
      "step": 2340
    },
    {
      "epoch": 3.68,
      "learning_rate": 5.4029733959311426e-05,
      "loss": 0.0578,
      "step": 2350
    },
    {
      "epoch": 3.69,
      "learning_rate": 5.3834115805946795e-05,
      "loss": 0.1084,
      "step": 2360
    },
    {
      "epoch": 3.71,
      "learning_rate": 5.3638497652582165e-05,
      "loss": 0.0534,
      "step": 2370
    },
    {
      "epoch": 3.72,
      "learning_rate": 5.3442879499217535e-05,
      "loss": 0.0598,
      "step": 2380
    },
    {
      "epoch": 3.74,
      "learning_rate": 5.32472613458529e-05,
      "loss": 0.1353,
      "step": 2390
    },
    {
      "epoch": 3.76,
      "learning_rate": 5.305164319248827e-05,
      "loss": 0.166,
      "step": 2400
    },
    {
      "epoch": 3.76,
      "eval_accuracy": 0.9024672320740169,
      "eval_loss": 0.3933294415473938,
      "eval_runtime": 42.6729,
      "eval_samples_per_second": 60.788,
      "eval_steps_per_second": 7.616,
      "step": 2400
    },
    {
      "epoch": 3.77,
      "learning_rate": 5.285602503912364e-05,
      "loss": 0.0961,
      "step": 2410
    },
    {
      "epoch": 3.79,
      "learning_rate": 5.266040688575901e-05,
      "loss": 0.0647,
      "step": 2420
    },
    {
      "epoch": 3.8,
      "learning_rate": 5.246478873239436e-05,
      "loss": 0.0744,
      "step": 2430
    },
    {
      "epoch": 3.82,
      "learning_rate": 5.226917057902973e-05,
      "loss": 0.1046,
      "step": 2440
    },
    {
      "epoch": 3.83,
      "learning_rate": 5.207355242566511e-05,
      "loss": 0.0925,
      "step": 2450
    },
    {
      "epoch": 3.85,
      "learning_rate": 5.187793427230048e-05,
      "loss": 0.1343,
      "step": 2460
    },
    {
      "epoch": 3.87,
      "learning_rate": 5.1682316118935835e-05,
      "loss": 0.0721,
      "step": 2470
    },
    {
      "epoch": 3.88,
      "learning_rate": 5.1486697965571205e-05,
      "loss": 0.1446,
      "step": 2480
    },
    {
      "epoch": 3.9,
      "learning_rate": 5.1291079812206575e-05,
      "loss": 0.0807,
      "step": 2490
    },
    {
      "epoch": 3.91,
      "learning_rate": 5.109546165884195e-05,
      "loss": 0.11,
      "step": 2500
    },
    {
      "epoch": 3.91,
      "eval_accuracy": 0.9055512721665382,
      "eval_loss": 0.39182865619659424,
      "eval_runtime": 42.4672,
      "eval_samples_per_second": 61.082,
      "eval_steps_per_second": 7.653,
      "step": 2500
    },
    {
      "epoch": 3.93,
      "learning_rate": 5.089984350547731e-05,
      "loss": 0.1331,
      "step": 2510
    },
    {
      "epoch": 3.94,
      "learning_rate": 5.070422535211268e-05,
      "loss": 0.0506,
      "step": 2520
    },
    {
      "epoch": 3.96,
      "learning_rate": 5.0508607198748047e-05,
      "loss": 0.1025,
      "step": 2530
    },
    {
      "epoch": 3.97,
      "learning_rate": 5.0312989045383416e-05,
      "loss": 0.0792,
      "step": 2540
    },
    {
      "epoch": 3.99,
      "learning_rate": 5.011737089201878e-05,
      "loss": 0.099,
      "step": 2550
    },
    {
      "epoch": 4.01,
      "learning_rate": 4.992175273865415e-05,
      "loss": 0.0861,
      "step": 2560
    },
    {
      "epoch": 4.02,
      "learning_rate": 4.972613458528952e-05,
      "loss": 0.0467,
      "step": 2570
    },
    {
      "epoch": 4.04,
      "learning_rate": 4.953051643192488e-05,
      "loss": 0.0587,
      "step": 2580
    },
    {
      "epoch": 4.05,
      "learning_rate": 4.933489827856025e-05,
      "loss": 0.064,
      "step": 2590
    },
    {
      "epoch": 4.07,
      "learning_rate": 4.913928012519562e-05,
      "loss": 0.0356,
      "step": 2600
    },
    {
      "epoch": 4.07,
      "eval_accuracy": 0.9202004626060138,
      "eval_loss": 0.3298385739326477,
      "eval_runtime": 42.9604,
      "eval_samples_per_second": 60.381,
      "eval_steps_per_second": 7.565,
      "step": 2600
    },
    {
      "epoch": 4.08,
      "learning_rate": 4.894366197183099e-05,
      "loss": 0.0376,
      "step": 2610
    },
    {
      "epoch": 4.1,
      "learning_rate": 4.8748043818466354e-05,
      "loss": 0.0275,
      "step": 2620
    },
    {
      "epoch": 4.12,
      "learning_rate": 4.855242566510172e-05,
      "loss": 0.0297,
      "step": 2630
    },
    {
      "epoch": 4.13,
      "learning_rate": 4.835680751173709e-05,
      "loss": 0.0323,
      "step": 2640
    },
    {
      "epoch": 4.15,
      "learning_rate": 4.816118935837246e-05,
      "loss": 0.0561,
      "step": 2650
    },
    {
      "epoch": 4.16,
      "learning_rate": 4.7965571205007826e-05,
      "loss": 0.0489,
      "step": 2660
    },
    {
      "epoch": 4.18,
      "learning_rate": 4.7769953051643195e-05,
      "loss": 0.0448,
      "step": 2670
    },
    {
      "epoch": 4.19,
      "learning_rate": 4.757433489827856e-05,
      "loss": 0.0398,
      "step": 2680
    },
    {
      "epoch": 4.21,
      "learning_rate": 4.737871674491393e-05,
      "loss": 0.0588,
      "step": 2690
    },
    {
      "epoch": 4.23,
      "learning_rate": 4.71830985915493e-05,
      "loss": 0.0513,
      "step": 2700
    },
    {
      "epoch": 4.23,
      "eval_accuracy": 0.9209714726291441,
      "eval_loss": 0.337054580450058,
      "eval_runtime": 42.6677,
      "eval_samples_per_second": 60.795,
      "eval_steps_per_second": 7.617,
      "step": 2700
    },
    {
      "epoch": 4.24,
      "learning_rate": 4.698748043818467e-05,
      "loss": 0.0396,
      "step": 2710
    },
    {
      "epoch": 4.26,
      "learning_rate": 4.679186228482003e-05,
      "loss": 0.0352,
      "step": 2720
    },
    {
      "epoch": 4.27,
      "learning_rate": 4.65962441314554e-05,
      "loss": 0.0695,
      "step": 2730
    },
    {
      "epoch": 4.29,
      "learning_rate": 4.640062597809077e-05,
      "loss": 0.0614,
      "step": 2740
    },
    {
      "epoch": 4.3,
      "learning_rate": 4.620500782472614e-05,
      "loss": 0.0702,
      "step": 2750
    },
    {
      "epoch": 4.32,
      "learning_rate": 4.60093896713615e-05,
      "loss": 0.037,
      "step": 2760
    },
    {
      "epoch": 4.33,
      "learning_rate": 4.581377151799687e-05,
      "loss": 0.0567,
      "step": 2770
    },
    {
      "epoch": 4.35,
      "learning_rate": 4.5618153364632235e-05,
      "loss": 0.0327,
      "step": 2780
    },
    {
      "epoch": 4.37,
      "learning_rate": 4.542253521126761e-05,
      "loss": 0.0358,
      "step": 2790
    },
    {
      "epoch": 4.38,
      "learning_rate": 4.5226917057902975e-05,
      "loss": 0.0762,
      "step": 2800
    },
    {
      "epoch": 4.38,
      "eval_accuracy": 0.9225134926754048,
      "eval_loss": 0.32532238960266113,
      "eval_runtime": 42.6755,
      "eval_samples_per_second": 60.784,
      "eval_steps_per_second": 7.616,
      "step": 2800
    },
    {
      "epoch": 4.4,
      "learning_rate": 4.5031298904538344e-05,
      "loss": 0.0223,
      "step": 2810
    },
    {
      "epoch": 4.41,
      "learning_rate": 4.483568075117371e-05,
      "loss": 0.028,
      "step": 2820
    },
    {
      "epoch": 4.43,
      "learning_rate": 4.464006259780908e-05,
      "loss": 0.0572,
      "step": 2830
    },
    {
      "epoch": 4.44,
      "learning_rate": 4.4444444444444447e-05,
      "loss": 0.0487,
      "step": 2840
    },
    {
      "epoch": 4.46,
      "learning_rate": 4.4248826291079816e-05,
      "loss": 0.0711,
      "step": 2850
    },
    {
      "epoch": 4.48,
      "learning_rate": 4.405320813771518e-05,
      "loss": 0.0334,
      "step": 2860
    },
    {
      "epoch": 4.49,
      "learning_rate": 4.385758998435055e-05,
      "loss": 0.0219,
      "step": 2870
    },
    {
      "epoch": 4.51,
      "learning_rate": 4.366197183098591e-05,
      "loss": 0.0405,
      "step": 2880
    },
    {
      "epoch": 4.52,
      "learning_rate": 4.346635367762129e-05,
      "loss": 0.0329,
      "step": 2890
    },
    {
      "epoch": 4.54,
      "learning_rate": 4.327073552425665e-05,
      "loss": 0.018,
      "step": 2900
    },
    {
      "epoch": 4.54,
      "eval_accuracy": 0.9148033924441018,
      "eval_loss": 0.34668266773223877,
      "eval_runtime": 42.9733,
      "eval_samples_per_second": 60.363,
      "eval_steps_per_second": 7.563,
      "step": 2900
    },
    {
      "epoch": 4.55,
      "learning_rate": 4.307511737089202e-05,
      "loss": 0.0413,
      "step": 2910
    },
    {
      "epoch": 4.57,
      "learning_rate": 4.287949921752739e-05,
      "loss": 0.018,
      "step": 2920
    },
    {
      "epoch": 4.59,
      "learning_rate": 4.2683881064162754e-05,
      "loss": 0.0417,
      "step": 2930
    },
    {
      "epoch": 4.6,
      "learning_rate": 4.248826291079812e-05,
      "loss": 0.0254,
      "step": 2940
    },
    {
      "epoch": 4.62,
      "learning_rate": 4.229264475743349e-05,
      "loss": 0.0411,
      "step": 2950
    },
    {
      "epoch": 4.63,
      "learning_rate": 4.209702660406886e-05,
      "loss": 0.0625,
      "step": 2960
    },
    {
      "epoch": 4.65,
      "learning_rate": 4.1901408450704226e-05,
      "loss": 0.0416,
      "step": 2970
    },
    {
      "epoch": 4.66,
      "learning_rate": 4.1705790297339595e-05,
      "loss": 0.0486,
      "step": 2980
    },
    {
      "epoch": 4.68,
      "learning_rate": 4.1510172143974965e-05,
      "loss": 0.0378,
      "step": 2990
    },
    {
      "epoch": 4.69,
      "learning_rate": 4.1314553990610335e-05,
      "loss": 0.0263,
      "step": 3000
    },
    {
      "epoch": 4.69,
      "eval_accuracy": 0.9144178874325366,
      "eval_loss": 0.3544096052646637,
      "eval_runtime": 43.2558,
      "eval_samples_per_second": 59.969,
      "eval_steps_per_second": 7.513,
      "step": 3000
    },
    {
      "epoch": 4.71,
      "learning_rate": 4.11189358372457e-05,
      "loss": 0.0449,
      "step": 3010
    },
    {
      "epoch": 4.73,
      "learning_rate": 4.092331768388107e-05,
      "loss": 0.0758,
      "step": 3020
    },
    {
      "epoch": 4.74,
      "learning_rate": 4.072769953051643e-05,
      "loss": 0.0418,
      "step": 3030
    },
    {
      "epoch": 4.76,
      "learning_rate": 4.053208137715181e-05,
      "loss": 0.0392,
      "step": 3040
    },
    {
      "epoch": 4.77,
      "learning_rate": 4.033646322378717e-05,
      "loss": 0.0195,
      "step": 3050
    },
    {
      "epoch": 4.79,
      "learning_rate": 4.014084507042254e-05,
      "loss": 0.0317,
      "step": 3060
    },
    {
      "epoch": 4.8,
      "learning_rate": 3.99452269170579e-05,
      "loss": 0.0162,
      "step": 3070
    },
    {
      "epoch": 4.82,
      "learning_rate": 3.974960876369327e-05,
      "loss": 0.0256,
      "step": 3080
    },
    {
      "epoch": 4.84,
      "learning_rate": 3.955399061032864e-05,
      "loss": 0.017,
      "step": 3090
    },
    {
      "epoch": 4.85,
      "learning_rate": 3.935837245696401e-05,
      "loss": 0.0205,
      "step": 3100
    },
    {
      "epoch": 4.85,
      "eval_accuracy": 0.9221279876638396,
      "eval_loss": 0.33404412865638733,
      "eval_runtime": 43.1306,
      "eval_samples_per_second": 60.143,
      "eval_steps_per_second": 7.535,
      "step": 3100
    },
    {
      "epoch": 4.87,
      "learning_rate": 3.9162754303599375e-05,
      "loss": 0.0549,
      "step": 3110
    },
    {
      "epoch": 4.88,
      "learning_rate": 3.8967136150234744e-05,
      "loss": 0.066,
      "step": 3120
    },
    {
      "epoch": 4.9,
      "learning_rate": 3.877151799687011e-05,
      "loss": 0.0307,
      "step": 3130
    },
    {
      "epoch": 4.91,
      "learning_rate": 3.8575899843505484e-05,
      "loss": 0.0212,
      "step": 3140
    },
    {
      "epoch": 4.93,
      "learning_rate": 3.8380281690140847e-05,
      "loss": 0.0311,
      "step": 3150
    },
    {
      "epoch": 4.95,
      "learning_rate": 3.8184663536776216e-05,
      "loss": 0.0146,
      "step": 3160
    },
    {
      "epoch": 4.96,
      "learning_rate": 3.798904538341158e-05,
      "loss": 0.0173,
      "step": 3170
    },
    {
      "epoch": 4.98,
      "learning_rate": 3.779342723004695e-05,
      "loss": 0.0336,
      "step": 3180
    },
    {
      "epoch": 4.99,
      "learning_rate": 3.759780907668232e-05,
      "loss": 0.0295,
      "step": 3190
    },
    {
      "epoch": 5.01,
      "learning_rate": 3.740219092331769e-05,
      "loss": 0.0237,
      "step": 3200
    },
    {
      "epoch": 5.01,
      "eval_accuracy": 0.9144178874325366,
      "eval_loss": 0.33526894450187683,
      "eval_runtime": 42.9414,
      "eval_samples_per_second": 60.408,
      "eval_steps_per_second": 7.568,
      "step": 3200
    },
    {
      "epoch": 5.02,
      "learning_rate": 3.720657276995305e-05,
      "loss": 0.0137,
      "step": 3210
    },
    {
      "epoch": 5.04,
      "learning_rate": 3.701095461658842e-05,
      "loss": 0.0161,
      "step": 3220
    },
    {
      "epoch": 5.05,
      "learning_rate": 3.681533646322379e-05,
      "loss": 0.0217,
      "step": 3230
    },
    {
      "epoch": 5.07,
      "learning_rate": 3.661971830985916e-05,
      "loss": 0.0184,
      "step": 3240
    },
    {
      "epoch": 5.09,
      "learning_rate": 3.642410015649452e-05,
      "loss": 0.0177,
      "step": 3250
    },
    {
      "epoch": 5.1,
      "learning_rate": 3.622848200312989e-05,
      "loss": 0.051,
      "step": 3260
    },
    {
      "epoch": 5.12,
      "learning_rate": 3.6032863849765256e-05,
      "loss": 0.012,
      "step": 3270
    },
    {
      "epoch": 5.13,
      "learning_rate": 3.5837245696400626e-05,
      "loss": 0.0156,
      "step": 3280
    },
    {
      "epoch": 5.15,
      "learning_rate": 3.5641627543035995e-05,
      "loss": 0.0129,
      "step": 3290
    },
    {
      "epoch": 5.16,
      "learning_rate": 3.5446009389671365e-05,
      "loss": 0.013,
      "step": 3300
    },
    {
      "epoch": 5.16,
      "eval_accuracy": 0.9228989976869699,
      "eval_loss": 0.3218042850494385,
      "eval_runtime": 42.5041,
      "eval_samples_per_second": 61.029,
      "eval_steps_per_second": 7.646,
      "step": 3300
    },
    {
      "epoch": 5.18,
      "learning_rate": 3.525039123630673e-05,
      "loss": 0.0106,
      "step": 3310
    },
    {
      "epoch": 5.2,
      "learning_rate": 3.50547730829421e-05,
      "loss": 0.0132,
      "step": 3320
    },
    {
      "epoch": 5.21,
      "learning_rate": 3.485915492957747e-05,
      "loss": 0.0212,
      "step": 3330
    },
    {
      "epoch": 5.23,
      "learning_rate": 3.466353677621284e-05,
      "loss": 0.0124,
      "step": 3340
    },
    {
      "epoch": 5.24,
      "learning_rate": 3.44679186228482e-05,
      "loss": 0.0135,
      "step": 3350
    },
    {
      "epoch": 5.26,
      "learning_rate": 3.427230046948357e-05,
      "loss": 0.0112,
      "step": 3360
    },
    {
      "epoch": 5.27,
      "learning_rate": 3.407668231611893e-05,
      "loss": 0.0175,
      "step": 3370
    },
    {
      "epoch": 5.29,
      "learning_rate": 3.38810641627543e-05,
      "loss": 0.0234,
      "step": 3380
    },
    {
      "epoch": 5.31,
      "learning_rate": 3.368544600938967e-05,
      "loss": 0.0133,
      "step": 3390
    },
    {
      "epoch": 5.32,
      "learning_rate": 3.348982785602504e-05,
      "loss": 0.0116,
      "step": 3400
    },
    {
      "epoch": 5.32,
      "eval_accuracy": 0.9290670778720124,
      "eval_loss": 0.308786541223526,
      "eval_runtime": 43.1692,
      "eval_samples_per_second": 60.089,
      "eval_steps_per_second": 7.529,
      "step": 3400
    },
    {
      "epoch": 5.34,
      "learning_rate": 3.3294209702660405e-05,
      "loss": 0.011,
      "step": 3410
    },
    {
      "epoch": 5.35,
      "learning_rate": 3.3098591549295775e-05,
      "loss": 0.0119,
      "step": 3420
    },
    {
      "epoch": 5.37,
      "learning_rate": 3.2902973395931144e-05,
      "loss": 0.0175,
      "step": 3430
    },
    {
      "epoch": 5.38,
      "learning_rate": 3.2707355242566514e-05,
      "loss": 0.0116,
      "step": 3440
    },
    {
      "epoch": 5.4,
      "learning_rate": 3.251173708920188e-05,
      "loss": 0.0113,
      "step": 3450
    },
    {
      "epoch": 5.41,
      "learning_rate": 3.2316118935837247e-05,
      "loss": 0.0152,
      "step": 3460
    },
    {
      "epoch": 5.43,
      "learning_rate": 3.212050078247261e-05,
      "loss": 0.012,
      "step": 3470
    },
    {
      "epoch": 5.45,
      "learning_rate": 3.1924882629107986e-05,
      "loss": 0.0096,
      "step": 3480
    },
    {
      "epoch": 5.46,
      "learning_rate": 3.1729264475743356e-05,
      "loss": 0.0109,
      "step": 3490
    },
    {
      "epoch": 5.48,
      "learning_rate": 3.153364632237872e-05,
      "loss": 0.0119,
      "step": 3500
    },
    {
      "epoch": 5.48,
      "eval_accuracy": 0.9279105628373169,
      "eval_loss": 0.3046722412109375,
      "eval_runtime": 42.5452,
      "eval_samples_per_second": 60.97,
      "eval_steps_per_second": 7.639,
      "step": 3500
    },
    {
      "epoch": 5.49,
      "learning_rate": 3.133802816901409e-05,
      "loss": 0.0091,
      "step": 3510
    },
    {
      "epoch": 5.51,
      "learning_rate": 3.114241001564945e-05,
      "loss": 0.0089,
      "step": 3520
    },
    {
      "epoch": 5.52,
      "learning_rate": 3.094679186228482e-05,
      "loss": 0.0113,
      "step": 3530
    },
    {
      "epoch": 5.54,
      "learning_rate": 3.075117370892019e-05,
      "loss": 0.0162,
      "step": 3540
    },
    {
      "epoch": 5.56,
      "learning_rate": 3.055555555555556e-05,
      "loss": 0.0089,
      "step": 3550
    },
    {
      "epoch": 5.57,
      "learning_rate": 3.0359937402190923e-05,
      "loss": 0.0105,
      "step": 3560
    },
    {
      "epoch": 5.59,
      "learning_rate": 3.0164319248826296e-05,
      "loss": 0.037,
      "step": 3570
    },
    {
      "epoch": 5.6,
      "learning_rate": 2.996870109546166e-05,
      "loss": 0.0097,
      "step": 3580
    },
    {
      "epoch": 5.62,
      "learning_rate": 2.977308294209703e-05,
      "loss": 0.0087,
      "step": 3590
    },
    {
      "epoch": 5.63,
      "learning_rate": 2.9577464788732395e-05,
      "loss": 0.0098,
      "step": 3600
    },
    {
      "epoch": 5.63,
      "eval_accuracy": 0.9282960678488821,
      "eval_loss": 0.30633866786956787,
      "eval_runtime": 42.8022,
      "eval_samples_per_second": 60.604,
      "eval_steps_per_second": 7.593,
      "step": 3600
    },
    {
      "epoch": 5.65,
      "learning_rate": 2.9381846635367765e-05,
      "loss": 0.009,
      "step": 3610
    },
    {
      "epoch": 5.67,
      "learning_rate": 2.918622848200313e-05,
      "loss": 0.0263,
      "step": 3620
    },
    {
      "epoch": 5.68,
      "learning_rate": 2.89906103286385e-05,
      "loss": 0.0112,
      "step": 3630
    },
    {
      "epoch": 5.7,
      "learning_rate": 2.8794992175273867e-05,
      "loss": 0.0104,
      "step": 3640
    },
    {
      "epoch": 5.71,
      "learning_rate": 2.8599374021909237e-05,
      "loss": 0.0082,
      "step": 3650
    },
    {
      "epoch": 5.73,
      "learning_rate": 2.84037558685446e-05,
      "loss": 0.0091,
      "step": 3660
    },
    {
      "epoch": 5.74,
      "learning_rate": 2.8208137715179973e-05,
      "loss": 0.0099,
      "step": 3670
    },
    {
      "epoch": 5.76,
      "learning_rate": 2.8012519561815336e-05,
      "loss": 0.0355,
      "step": 3680
    },
    {
      "epoch": 5.77,
      "learning_rate": 2.7816901408450706e-05,
      "loss": 0.0088,
      "step": 3690
    },
    {
      "epoch": 5.79,
      "learning_rate": 2.7621283255086072e-05,
      "loss": 0.0086,
      "step": 3700
    },
    {
      "epoch": 5.79,
      "eval_accuracy": 0.9267540478026214,
      "eval_loss": 0.30737537145614624,
      "eval_runtime": 42.6258,
      "eval_samples_per_second": 60.855,
      "eval_steps_per_second": 7.624,
      "step": 3700
    },
    {
      "epoch": 5.81,
      "learning_rate": 2.7425665101721442e-05,
      "loss": 0.0097,
      "step": 3710
    },
    {
      "epoch": 5.82,
      "learning_rate": 2.7230046948356808e-05,
      "loss": 0.0092,
      "step": 3720
    },
    {
      "epoch": 5.84,
      "learning_rate": 2.7034428794992178e-05,
      "loss": 0.0085,
      "step": 3730
    },
    {
      "epoch": 5.85,
      "learning_rate": 2.6838810641627544e-05,
      "loss": 0.0089,
      "step": 3740
    },
    {
      "epoch": 5.87,
      "learning_rate": 2.6643192488262914e-05,
      "loss": 0.0341,
      "step": 3750
    },
    {
      "epoch": 5.88,
      "learning_rate": 2.6447574334898277e-05,
      "loss": 0.0357,
      "step": 3760
    },
    {
      "epoch": 5.9,
      "learning_rate": 2.625195618153365e-05,
      "loss": 0.0088,
      "step": 3770
    },
    {
      "epoch": 5.92,
      "learning_rate": 2.6056338028169013e-05,
      "loss": 0.0101,
      "step": 3780
    },
    {
      "epoch": 5.93,
      "learning_rate": 2.5860719874804386e-05,
      "loss": 0.0124,
      "step": 3790
    },
    {
      "epoch": 5.95,
      "learning_rate": 2.566510172143975e-05,
      "loss": 0.0081,
      "step": 3800
    },
    {
      "epoch": 5.95,
      "eval_accuracy": 0.9236700077101002,
      "eval_loss": 0.32199642062187195,
      "eval_runtime": 43.3628,
      "eval_samples_per_second": 59.821,
      "eval_steps_per_second": 7.495,
      "step": 3800
    },
    {
      "epoch": 5.96,
      "learning_rate": 2.546948356807512e-05,
      "loss": 0.0096,
      "step": 3810
    },
    {
      "epoch": 5.98,
      "learning_rate": 2.5273865414710485e-05,
      "loss": 0.0217,
      "step": 3820
    },
    {
      "epoch": 5.99,
      "learning_rate": 2.5078247261345855e-05,
      "loss": 0.0112,
      "step": 3830
    },
    {
      "epoch": 6.01,
      "learning_rate": 2.4882629107981224e-05,
      "loss": 0.0115,
      "step": 3840
    },
    {
      "epoch": 6.03,
      "learning_rate": 2.468701095461659e-05,
      "loss": 0.0081,
      "step": 3850
    },
    {
      "epoch": 6.04,
      "learning_rate": 2.4491392801251957e-05,
      "loss": 0.0076,
      "step": 3860
    },
    {
      "epoch": 6.06,
      "learning_rate": 2.4295774647887327e-05,
      "loss": 0.0078,
      "step": 3870
    },
    {
      "epoch": 6.07,
      "learning_rate": 2.4100156494522693e-05,
      "loss": 0.0082,
      "step": 3880
    },
    {
      "epoch": 6.09,
      "learning_rate": 2.3904538341158063e-05,
      "loss": 0.0079,
      "step": 3890
    },
    {
      "epoch": 6.1,
      "learning_rate": 2.370892018779343e-05,
      "loss": 0.0078,
      "step": 3900
    },
    {
      "epoch": 6.1,
      "eval_accuracy": 0.9267540478026214,
      "eval_loss": 0.30635374784469604,
      "eval_runtime": 43.3159,
      "eval_samples_per_second": 59.886,
      "eval_steps_per_second": 7.503,
      "step": 3900
    },
    {
      "epoch": 6.12,
      "learning_rate": 2.3513302034428795e-05,
      "loss": 0.008,
      "step": 3910
    },
    {
      "epoch": 6.13,
      "learning_rate": 2.3317683881064165e-05,
      "loss": 0.0081,
      "step": 3920
    },
    {
      "epoch": 6.15,
      "learning_rate": 2.312206572769953e-05,
      "loss": 0.0074,
      "step": 3930
    },
    {
      "epoch": 6.17,
      "learning_rate": 2.29264475743349e-05,
      "loss": 0.0075,
      "step": 3940
    },
    {
      "epoch": 6.18,
      "learning_rate": 2.2730829420970267e-05,
      "loss": 0.0082,
      "step": 3950
    },
    {
      "epoch": 6.2,
      "learning_rate": 2.2535211267605634e-05,
      "loss": 0.0079,
      "step": 3960
    },
    {
      "epoch": 6.21,
      "learning_rate": 2.2339593114241003e-05,
      "loss": 0.0075,
      "step": 3970
    },
    {
      "epoch": 6.23,
      "learning_rate": 2.214397496087637e-05,
      "loss": 0.0074,
      "step": 3980
    },
    {
      "epoch": 6.24,
      "learning_rate": 2.194835680751174e-05,
      "loss": 0.0067,
      "step": 3990
    },
    {
      "epoch": 6.26,
      "learning_rate": 2.1752738654147106e-05,
      "loss": 0.0074,
      "step": 4000
    },
    {
      "epoch": 6.26,
      "eval_accuracy": 0.9279105628373169,
      "eval_loss": 0.30622774362564087,
      "eval_runtime": 42.6237,
      "eval_samples_per_second": 60.858,
      "eval_steps_per_second": 7.625,
      "step": 4000
    },
    {
      "epoch": 6.28,
      "learning_rate": 2.1557120500782476e-05,
      "loss": 0.0077,
      "step": 4010
    },
    {
      "epoch": 6.29,
      "learning_rate": 2.1361502347417842e-05,
      "loss": 0.0074,
      "step": 4020
    },
    {
      "epoch": 6.31,
      "learning_rate": 2.1165884194053208e-05,
      "loss": 0.0076,
      "step": 4030
    },
    {
      "epoch": 6.32,
      "learning_rate": 2.0970266040688578e-05,
      "loss": 0.0077,
      "step": 4040
    },
    {
      "epoch": 6.34,
      "learning_rate": 2.0774647887323944e-05,
      "loss": 0.0075,
      "step": 4050
    },
    {
      "epoch": 6.35,
      "learning_rate": 2.0579029733959314e-05,
      "loss": 0.0066,
      "step": 4060
    },
    {
      "epoch": 6.37,
      "learning_rate": 2.038341158059468e-05,
      "loss": 0.0077,
      "step": 4070
    },
    {
      "epoch": 6.38,
      "learning_rate": 2.0187793427230047e-05,
      "loss": 0.0076,
      "step": 4080
    },
    {
      "epoch": 6.4,
      "learning_rate": 1.9992175273865416e-05,
      "loss": 0.0073,
      "step": 4090
    },
    {
      "epoch": 6.42,
      "learning_rate": 1.9796557120500783e-05,
      "loss": 0.0068,
      "step": 4100
    },
    {
      "epoch": 6.42,
      "eval_accuracy": 0.9290670778720124,
      "eval_loss": 0.3051210343837738,
      "eval_runtime": 43.4448,
      "eval_samples_per_second": 59.708,
      "eval_steps_per_second": 7.481,
      "step": 4100
    },
    {
      "epoch": 6.43,
      "learning_rate": 1.9600938967136152e-05,
      "loss": 0.0067,
      "step": 4110
    },
    {
      "epoch": 6.45,
      "learning_rate": 1.940532081377152e-05,
      "loss": 0.0382,
      "step": 4120
    },
    {
      "epoch": 6.46,
      "learning_rate": 1.9209702660406885e-05,
      "loss": 0.0071,
      "step": 4130
    },
    {
      "epoch": 6.48,
      "learning_rate": 1.9014084507042255e-05,
      "loss": 0.0155,
      "step": 4140
    },
    {
      "epoch": 6.49,
      "learning_rate": 1.881846635367762e-05,
      "loss": 0.0075,
      "step": 4150
    },
    {
      "epoch": 6.51,
      "learning_rate": 1.862284820031299e-05,
      "loss": 0.0073,
      "step": 4160
    },
    {
      "epoch": 6.53,
      "learning_rate": 1.8427230046948357e-05,
      "loss": 0.0076,
      "step": 4170
    },
    {
      "epoch": 6.54,
      "learning_rate": 1.8231611893583723e-05,
      "loss": 0.0064,
      "step": 4180
    },
    {
      "epoch": 6.56,
      "learning_rate": 1.8035993740219093e-05,
      "loss": 0.0097,
      "step": 4190
    },
    {
      "epoch": 6.57,
      "learning_rate": 1.784037558685446e-05,
      "loss": 0.006,
      "step": 4200
    },
    {
      "epoch": 6.57,
      "eval_accuracy": 0.9298380878951427,
      "eval_loss": 0.3000366985797882,
      "eval_runtime": 42.6162,
      "eval_samples_per_second": 60.869,
      "eval_steps_per_second": 7.626,
      "step": 4200
    },
    {
      "epoch": 6.59,
      "learning_rate": 1.764475743348983e-05,
      "loss": 0.0077,
      "step": 4210
    },
    {
      "epoch": 6.6,
      "learning_rate": 1.7449139280125195e-05,
      "loss": 0.0076,
      "step": 4220
    },
    {
      "epoch": 6.62,
      "learning_rate": 1.7253521126760565e-05,
      "loss": 0.0069,
      "step": 4230
    },
    {
      "epoch": 6.64,
      "learning_rate": 1.705790297339593e-05,
      "loss": 0.0067,
      "step": 4240
    },
    {
      "epoch": 6.65,
      "learning_rate": 1.6862284820031298e-05,
      "loss": 0.0075,
      "step": 4250
    },
    {
      "epoch": 6.67,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.0068,
      "step": 4260
    },
    {
      "epoch": 6.68,
      "learning_rate": 1.6471048513302034e-05,
      "loss": 0.0062,
      "step": 4270
    },
    {
      "epoch": 6.7,
      "learning_rate": 1.6275430359937403e-05,
      "loss": 0.0066,
      "step": 4280
    },
    {
      "epoch": 6.71,
      "learning_rate": 1.607981220657277e-05,
      "loss": 0.0072,
      "step": 4290
    },
    {
      "epoch": 6.73,
      "learning_rate": 1.5884194053208136e-05,
      "loss": 0.0075,
      "step": 4300
    },
    {
      "epoch": 6.73,
      "eval_accuracy": 0.930994602929838,
      "eval_loss": 0.30098453164100647,
      "eval_runtime": 42.908,
      "eval_samples_per_second": 60.455,
      "eval_steps_per_second": 7.574,
      "step": 4300
    },
    {
      "epoch": 6.74,
      "learning_rate": 1.5688575899843506e-05,
      "loss": 0.0066,
      "step": 4310
    },
    {
      "epoch": 6.76,
      "learning_rate": 1.5492957746478872e-05,
      "loss": 0.0061,
      "step": 4320
    },
    {
      "epoch": 6.78,
      "learning_rate": 1.5297339593114242e-05,
      "loss": 0.0073,
      "step": 4330
    },
    {
      "epoch": 6.79,
      "learning_rate": 1.5101721439749608e-05,
      "loss": 0.013,
      "step": 4340
    },
    {
      "epoch": 6.81,
      "learning_rate": 1.4906103286384976e-05,
      "loss": 0.0071,
      "step": 4350
    },
    {
      "epoch": 6.82,
      "learning_rate": 1.4710485133020346e-05,
      "loss": 0.0063,
      "step": 4360
    },
    {
      "epoch": 6.84,
      "learning_rate": 1.4514866979655714e-05,
      "loss": 0.0071,
      "step": 4370
    },
    {
      "epoch": 6.85,
      "learning_rate": 1.4319248826291082e-05,
      "loss": 0.0067,
      "step": 4380
    },
    {
      "epoch": 6.87,
      "learning_rate": 1.412363067292645e-05,
      "loss": 0.0475,
      "step": 4390
    },
    {
      "epoch": 6.89,
      "learning_rate": 1.3928012519561818e-05,
      "loss": 0.0057,
      "step": 4400
    },
    {
      "epoch": 6.89,
      "eval_accuracy": 0.9298380878951427,
      "eval_loss": 0.3036852180957794,
      "eval_runtime": 42.308,
      "eval_samples_per_second": 61.312,
      "eval_steps_per_second": 7.682,
      "step": 4400
    },
    {
      "epoch": 6.9,
      "learning_rate": 1.3732394366197184e-05,
      "loss": 0.0058,
      "step": 4410
    },
    {
      "epoch": 6.92,
      "learning_rate": 1.3536776212832552e-05,
      "loss": 0.0068,
      "step": 4420
    },
    {
      "epoch": 6.93,
      "learning_rate": 1.334115805946792e-05,
      "loss": 0.0061,
      "step": 4430
    },
    {
      "epoch": 6.95,
      "learning_rate": 1.3145539906103288e-05,
      "loss": 0.0245,
      "step": 4440
    },
    {
      "epoch": 6.96,
      "learning_rate": 1.2949921752738656e-05,
      "loss": 0.0304,
      "step": 4450
    },
    {
      "epoch": 6.98,
      "learning_rate": 1.2754303599374023e-05,
      "loss": 0.0068,
      "step": 4460
    },
    {
      "epoch": 7.0,
      "learning_rate": 1.255868544600939e-05,
      "loss": 0.0066,
      "step": 4470
    },
    {
      "epoch": 7.01,
      "learning_rate": 1.2363067292644757e-05,
      "loss": 0.0071,
      "step": 4480
    },
    {
      "epoch": 7.03,
      "learning_rate": 1.2167449139280125e-05,
      "loss": 0.0053,
      "step": 4490
    },
    {
      "epoch": 7.04,
      "learning_rate": 1.1971830985915493e-05,
      "loss": 0.0058,
      "step": 4500
    },
    {
      "epoch": 7.04,
      "eval_accuracy": 0.9279105628373169,
      "eval_loss": 0.30713140964508057,
      "eval_runtime": 42.5583,
      "eval_samples_per_second": 60.952,
      "eval_steps_per_second": 7.637,
      "step": 4500
    },
    {
      "epoch": 7.06,
      "learning_rate": 1.1776212832550863e-05,
      "loss": 0.0061,
      "step": 4510
    },
    {
      "epoch": 7.07,
      "learning_rate": 1.1580594679186229e-05,
      "loss": 0.0064,
      "step": 4520
    },
    {
      "epoch": 7.09,
      "learning_rate": 1.1384976525821597e-05,
      "loss": 0.0069,
      "step": 4530
    },
    {
      "epoch": 7.1,
      "learning_rate": 1.1189358372456965e-05,
      "loss": 0.0057,
      "step": 4540
    },
    {
      "epoch": 7.12,
      "learning_rate": 1.0993740219092333e-05,
      "loss": 0.0061,
      "step": 4550
    },
    {
      "epoch": 7.14,
      "learning_rate": 1.0798122065727701e-05,
      "loss": 0.0062,
      "step": 4560
    },
    {
      "epoch": 7.15,
      "learning_rate": 1.0602503912363067e-05,
      "loss": 0.0062,
      "step": 4570
    },
    {
      "epoch": 7.17,
      "learning_rate": 1.0406885758998435e-05,
      "loss": 0.0058,
      "step": 4580
    },
    {
      "epoch": 7.18,
      "learning_rate": 1.0211267605633803e-05,
      "loss": 0.0073,
      "step": 4590
    },
    {
      "epoch": 7.2,
      "learning_rate": 1.0015649452269172e-05,
      "loss": 0.0075,
      "step": 4600
    },
    {
      "epoch": 7.2,
      "eval_accuracy": 0.9282960678488821,
      "eval_loss": 0.307522177696228,
      "eval_runtime": 43.0156,
      "eval_samples_per_second": 60.304,
      "eval_steps_per_second": 7.555,
      "step": 4600
    },
    {
      "epoch": 7.21,
      "learning_rate": 9.82003129890454e-06,
      "loss": 0.0059,
      "step": 4610
    },
    {
      "epoch": 7.23,
      "learning_rate": 9.624413145539906e-06,
      "loss": 0.0064,
      "step": 4620
    },
    {
      "epoch": 7.25,
      "learning_rate": 9.428794992175274e-06,
      "loss": 0.0063,
      "step": 4630
    },
    {
      "epoch": 7.26,
      "learning_rate": 9.233176838810642e-06,
      "loss": 0.0052,
      "step": 4640
    },
    {
      "epoch": 7.28,
      "learning_rate": 9.03755868544601e-06,
      "loss": 0.0053,
      "step": 4650
    },
    {
      "epoch": 7.29,
      "learning_rate": 8.841940532081378e-06,
      "loss": 0.0059,
      "step": 4660
    },
    {
      "epoch": 7.31,
      "learning_rate": 8.646322378716746e-06,
      "loss": 0.0059,
      "step": 4670
    },
    {
      "epoch": 7.32,
      "learning_rate": 8.450704225352112e-06,
      "loss": 0.0058,
      "step": 4680
    },
    {
      "epoch": 7.34,
      "learning_rate": 8.25508607198748e-06,
      "loss": 0.0054,
      "step": 4690
    },
    {
      "epoch": 7.36,
      "learning_rate": 8.059467918622848e-06,
      "loss": 0.0066,
      "step": 4700
    },
    {
      "epoch": 7.36,
      "eval_accuracy": 0.9294525828835775,
      "eval_loss": 0.30765867233276367,
      "eval_runtime": 43.157,
      "eval_samples_per_second": 60.106,
      "eval_steps_per_second": 7.531,
      "step": 4700
    },
    {
      "epoch": 7.37,
      "learning_rate": 7.863849765258216e-06,
      "loss": 0.0054,
      "step": 4710
    },
    {
      "epoch": 7.39,
      "learning_rate": 7.668231611893584e-06,
      "loss": 0.0064,
      "step": 4720
    },
    {
      "epoch": 7.4,
      "learning_rate": 7.4726134585289515e-06,
      "loss": 0.0059,
      "step": 4730
    },
    {
      "epoch": 7.42,
      "learning_rate": 7.2769953051643195e-06,
      "loss": 0.0057,
      "step": 4740
    },
    {
      "epoch": 7.43,
      "learning_rate": 7.081377151799687e-06,
      "loss": 0.0066,
      "step": 4750
    },
    {
      "epoch": 7.45,
      "learning_rate": 6.885758998435055e-06,
      "loss": 0.0113,
      "step": 4760
    },
    {
      "epoch": 7.46,
      "learning_rate": 6.690140845070423e-06,
      "loss": 0.0059,
      "step": 4770
    },
    {
      "epoch": 7.48,
      "learning_rate": 6.49452269170579e-06,
      "loss": 0.0061,
      "step": 4780
    },
    {
      "epoch": 7.5,
      "learning_rate": 6.298904538341158e-06,
      "loss": 0.0069,
      "step": 4790
    },
    {
      "epoch": 7.51,
      "learning_rate": 6.103286384976526e-06,
      "loss": 0.0056,
      "step": 4800
    },
    {
      "epoch": 7.51,
      "eval_accuracy": 0.9294525828835775,
      "eval_loss": 0.30838659405708313,
      "eval_runtime": 43.0465,
      "eval_samples_per_second": 60.26,
      "eval_steps_per_second": 7.55,
      "step": 4800
    },
    {
      "epoch": 7.53,
      "learning_rate": 5.907668231611894e-06,
      "loss": 0.0059,
      "step": 4810
    },
    {
      "epoch": 7.54,
      "learning_rate": 5.712050078247261e-06,
      "loss": 0.0054,
      "step": 4820
    },
    {
      "epoch": 7.56,
      "learning_rate": 5.516431924882629e-06,
      "loss": 0.0054,
      "step": 4830
    },
    {
      "epoch": 7.57,
      "learning_rate": 5.320813771517997e-06,
      "loss": 0.0059,
      "step": 4840
    },
    {
      "epoch": 7.59,
      "learning_rate": 5.125195618153364e-06,
      "loss": 0.0061,
      "step": 4850
    },
    {
      "epoch": 7.61,
      "learning_rate": 4.929577464788732e-06,
      "loss": 0.0062,
      "step": 4860
    },
    {
      "epoch": 7.62,
      "learning_rate": 4.7339593114241e-06,
      "loss": 0.0215,
      "step": 4870
    },
    {
      "epoch": 7.64,
      "learning_rate": 4.538341158059468e-06,
      "loss": 0.0057,
      "step": 4880
    },
    {
      "epoch": 7.65,
      "learning_rate": 4.342723004694836e-06,
      "loss": 0.0061,
      "step": 4890
    },
    {
      "epoch": 7.67,
      "learning_rate": 4.1471048513302035e-06,
      "loss": 0.0053,
      "step": 4900
    },
    {
      "epoch": 7.67,
      "eval_accuracy": 0.930994602929838,
      "eval_loss": 0.3063901364803314,
      "eval_runtime": 42.9974,
      "eval_samples_per_second": 60.329,
      "eval_steps_per_second": 7.559,
      "step": 4900
    },
    {
      "epoch": 7.68,
      "learning_rate": 3.9514866979655715e-06,
      "loss": 0.006,
      "step": 4910
    },
    {
      "epoch": 7.7,
      "learning_rate": 3.755868544600939e-06,
      "loss": 0.0053,
      "step": 4920
    },
    {
      "epoch": 7.72,
      "learning_rate": 3.560250391236307e-06,
      "loss": 0.0051,
      "step": 4930
    },
    {
      "epoch": 7.73,
      "learning_rate": 3.3646322378716747e-06,
      "loss": 0.0057,
      "step": 4940
    },
    {
      "epoch": 7.75,
      "learning_rate": 3.1690140845070423e-06,
      "loss": 0.0259,
      "step": 4950
    },
    {
      "epoch": 7.76,
      "learning_rate": 2.97339593114241e-06,
      "loss": 0.0055,
      "step": 4960
    },
    {
      "epoch": 7.78,
      "learning_rate": 2.777777777777778e-06,
      "loss": 0.0051,
      "step": 4970
    },
    {
      "epoch": 7.79,
      "learning_rate": 2.582159624413146e-06,
      "loss": 0.006,
      "step": 4980
    },
    {
      "epoch": 7.81,
      "learning_rate": 2.3865414710485135e-06,
      "loss": 0.0062,
      "step": 4990
    },
    {
      "epoch": 7.82,
      "learning_rate": 2.190923317683881e-06,
      "loss": 0.0057,
      "step": 5000
    },
    {
      "epoch": 7.82,
      "eval_accuracy": 0.9317656129529683,
      "eval_loss": 0.3068486750125885,
      "eval_runtime": 42.7527,
      "eval_samples_per_second": 60.675,
      "eval_steps_per_second": 7.602,
      "step": 5000
    },
    {
      "epoch": 7.84,
      "learning_rate": 1.995305164319249e-06,
      "loss": 0.0057,
      "step": 5010
    },
    {
      "epoch": 7.86,
      "learning_rate": 1.7996870109546167e-06,
      "loss": 0.0063,
      "step": 5020
    },
    {
      "epoch": 7.87,
      "learning_rate": 1.6040688575899843e-06,
      "loss": 0.0063,
      "step": 5030
    },
    {
      "epoch": 7.89,
      "learning_rate": 1.4084507042253521e-06,
      "loss": 0.0061,
      "step": 5040
    },
    {
      "epoch": 7.9,
      "learning_rate": 1.21283255086072e-06,
      "loss": 0.0051,
      "step": 5050
    },
    {
      "epoch": 7.92,
      "learning_rate": 1.0172143974960877e-06,
      "loss": 0.0061,
      "step": 5060
    },
    {
      "epoch": 7.93,
      "learning_rate": 8.215962441314555e-07,
      "loss": 0.0061,
      "step": 5070
    },
    {
      "epoch": 7.95,
      "learning_rate": 6.259780907668232e-07,
      "loss": 0.0049,
      "step": 5080
    },
    {
      "epoch": 7.97,
      "learning_rate": 4.303599374021909e-07,
      "loss": 0.0049,
      "step": 5090
    },
    {
      "epoch": 7.98,
      "learning_rate": 2.347417840375587e-07,
      "loss": 0.0055,
      "step": 5100
    },
    {
      "epoch": 7.98,
      "eval_accuracy": 0.9317656129529683,
      "eval_loss": 0.30683887004852295,
      "eval_runtime": 43.5191,
      "eval_samples_per_second": 59.606,
      "eval_steps_per_second": 7.468,
      "step": 5100
    },
    {
      "epoch": 8.0,
      "learning_rate": 3.912363067292645e-08,
      "loss": 0.0062,
      "step": 5110
    },
    {
      "epoch": 8.0,
      "step": 5112,
      "total_flos": 6.337884979995771e+18,
      "train_loss": 0.5634685837152139,
      "train_runtime": 5666.239,
      "train_samples_per_second": 14.415,
      "train_steps_per_second": 0.902
    }
  ],
  "max_steps": 5112,
  "num_train_epochs": 8,
  "total_flos": 6.337884979995771e+18,
  "trial_name": null,
  "trial_params": null
}