{
  "best_metric": 0.9086,
  "best_model_checkpoint": "./results/checkpoint-4680",
  "epoch": 9.9984,
  "global_step": 4680,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 1.0683760683760685e-06,
      "loss": 2.3089,
      "step": 10
    },
    {
      "epoch": 0.04,
      "learning_rate": 2.136752136752137e-06,
      "loss": 2.2658,
      "step": 20
    },
    {
      "epoch": 0.06,
      "learning_rate": 3.205128205128205e-06,
      "loss": 2.1767,
      "step": 30
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.273504273504274e-06,
      "loss": 2.0595,
      "step": 40
    },
    {
      "epoch": 0.11,
      "learning_rate": 5.341880341880342e-06,
      "loss": 1.8958,
      "step": 50
    },
    {
      "epoch": 0.13,
      "learning_rate": 6.41025641025641e-06,
      "loss": 1.753,
      "step": 60
    },
    {
      "epoch": 0.15,
      "learning_rate": 7.478632478632479e-06,
      "loss": 1.4999,
      "step": 70
    },
    {
      "epoch": 0.17,
      "learning_rate": 8.547008547008548e-06,
      "loss": 1.2355,
      "step": 80
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.615384615384616e-06,
      "loss": 1.0098,
      "step": 90
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.0683760683760684e-05,
      "loss": 0.7874,
      "step": 100
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.1752136752136752e-05,
      "loss": 0.7019,
      "step": 110
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.282051282051282e-05,
      "loss": 0.6812,
      "step": 120
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.388888888888889e-05,
      "loss": 0.5752,
      "step": 130
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.4957264957264958e-05,
      "loss": 0.6161,
      "step": 140
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.602564102564103e-05,
      "loss": 0.5919,
      "step": 150
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.7094017094017095e-05,
      "loss": 0.5563,
      "step": 160
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.8162393162393162e-05,
      "loss": 0.5661,
      "step": 170
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.923076923076923e-05,
      "loss": 0.5844,
      "step": 180
    },
    {
      "epoch": 0.41,
      "learning_rate": 2.02991452991453e-05,
      "loss": 0.5954,
      "step": 190
    },
    {
      "epoch": 0.43,
      "learning_rate": 2.1367521367521368e-05,
      "loss": 0.5298,
      "step": 200
    },
    {
      "epoch": 0.45,
      "learning_rate": 2.2435897435897437e-05,
      "loss": 0.5206,
      "step": 210
    },
    {
      "epoch": 0.47,
      "learning_rate": 2.3504273504273504e-05,
      "loss": 0.5294,
      "step": 220
    },
    {
      "epoch": 0.49,
      "learning_rate": 2.4572649572649573e-05,
      "loss": 0.5235,
      "step": 230
    },
    {
      "epoch": 0.51,
      "learning_rate": 2.564102564102564e-05,
      "loss": 0.4706,
      "step": 240
    },
    {
      "epoch": 0.53,
      "learning_rate": 2.670940170940171e-05,
      "loss": 0.5064,
      "step": 250
    },
    {
      "epoch": 0.55,
      "learning_rate": 2.777777777777778e-05,
      "loss": 0.4844,
      "step": 260
    },
    {
      "epoch": 0.58,
      "learning_rate": 2.8846153846153845e-05,
      "loss": 0.481,
      "step": 270
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.9914529914529915e-05,
      "loss": 0.4895,
      "step": 280
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.098290598290599e-05,
      "loss": 0.5137,
      "step": 290
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.205128205128206e-05,
      "loss": 0.4769,
      "step": 300
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.311965811965812e-05,
      "loss": 0.5685,
      "step": 310
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.418803418803419e-05,
      "loss": 0.4679,
      "step": 320
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.525641025641026e-05,
      "loss": 0.4553,
      "step": 330
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.6324786324786323e-05,
      "loss": 0.4842,
      "step": 340
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.739316239316239e-05,
      "loss": 0.5318,
      "step": 350
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.846153846153846e-05,
      "loss": 0.4787,
      "step": 360
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.952991452991453e-05,
      "loss": 0.4367,
      "step": 370
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.05982905982906e-05,
      "loss": 0.467,
      "step": 380
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.166666666666667e-05,
      "loss": 0.4721,
      "step": 390
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.2735042735042735e-05,
      "loss": 0.4258,
      "step": 400
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.3803418803418805e-05,
      "loss": 0.4811,
      "step": 410
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.4871794871794874e-05,
      "loss": 0.4632,
      "step": 420
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.594017094017094e-05,
      "loss": 0.4413,
      "step": 430
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.700854700854701e-05,
      "loss": 0.4483,
      "step": 440
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.8076923076923084e-05,
      "loss": 0.4529,
      "step": 450
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.9145299145299147e-05,
      "loss": 0.4201,
      "step": 460
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8306,
      "eval_loss": 0.4693240821361542,
      "eval_runtime": 47.1099,
      "eval_samples_per_second": 212.27,
      "eval_steps_per_second": 6.644,
      "step": 468
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.997625830959164e-05,
      "loss": 0.4572,
      "step": 470
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.985754985754986e-05,
      "loss": 0.4523,
      "step": 480
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.9738841405508076e-05,
      "loss": 0.4367,
      "step": 490
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.962013295346629e-05,
      "loss": 0.4355,
      "step": 500
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.95014245014245e-05,
      "loss": 0.4326,
      "step": 510
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.938271604938271e-05,
      "loss": 0.4255,
      "step": 520
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.926400759734093e-05,
      "loss": 0.4132,
      "step": 530
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.9145299145299147e-05,
      "loss": 0.4226,
      "step": 540
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.902659069325736e-05,
      "loss": 0.4443,
      "step": 550
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.890788224121557e-05,
      "loss": 0.4309,
      "step": 560
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.878917378917379e-05,
      "loss": 0.4062,
      "step": 570
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.867046533713201e-05,
      "loss": 0.406,
      "step": 580
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.8551756885090224e-05,
      "loss": 0.4252,
      "step": 590
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.8433048433048433e-05,
      "loss": 0.4016,
      "step": 600
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.831433998100665e-05,
      "loss": 0.411,
      "step": 610
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.819563152896486e-05,
      "loss": 0.4265,
      "step": 620
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.8076923076923084e-05,
      "loss": 0.4245,
      "step": 630
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.7958214624881294e-05,
      "loss": 0.3896,
      "step": 640
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.783950617283951e-05,
      "loss": 0.3938,
      "step": 650
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.772079772079772e-05,
      "loss": 0.4191,
      "step": 660
    },
    {
      "epoch": 1.43,
      "learning_rate": 4.760208926875594e-05,
      "loss": 0.4494,
      "step": 670
    },
    {
      "epoch": 1.45,
      "learning_rate": 4.7483380816714154e-05,
      "loss": 0.4286,
      "step": 680
    },
    {
      "epoch": 1.47,
      "learning_rate": 4.736467236467237e-05,
      "loss": 0.3869,
      "step": 690
    },
    {
      "epoch": 1.49,
      "learning_rate": 4.724596391263058e-05,
      "loss": 0.365,
      "step": 700
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.71272554605888e-05,
      "loss": 0.4056,
      "step": 710
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.700854700854701e-05,
      "loss": 0.4182,
      "step": 720
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.688983855650523e-05,
      "loss": 0.3884,
      "step": 730
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.677113010446344e-05,
      "loss": 0.3964,
      "step": 740
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.665242165242166e-05,
      "loss": 0.3833,
      "step": 750
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.653371320037987e-05,
      "loss": 0.397,
      "step": 760
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.6415004748338084e-05,
      "loss": 0.3863,
      "step": 770
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.62962962962963e-05,
      "loss": 0.3379,
      "step": 780
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.617758784425451e-05,
      "loss": 0.3488,
      "step": 790
    },
    {
      "epoch": 1.71,
      "learning_rate": 4.605887939221273e-05,
      "loss": 0.4028,
      "step": 800
    },
    {
      "epoch": 1.73,
      "learning_rate": 4.594017094017094e-05,
      "loss": 0.3853,
      "step": 810
    },
    {
      "epoch": 1.75,
      "learning_rate": 4.582146248812916e-05,
      "loss": 0.3498,
      "step": 820
    },
    {
      "epoch": 1.77,
      "learning_rate": 4.570275403608737e-05,
      "loss": 0.3778,
      "step": 830
    },
    {
      "epoch": 1.79,
      "learning_rate": 4.558404558404559e-05,
      "loss": 0.4053,
      "step": 840
    },
    {
      "epoch": 1.81,
      "learning_rate": 4.54653371320038e-05,
      "loss": 0.3552,
      "step": 850
    },
    {
      "epoch": 1.84,
      "learning_rate": 4.5346628679962014e-05,
      "loss": 0.3976,
      "step": 860
    },
    {
      "epoch": 1.86,
      "learning_rate": 4.522792022792023e-05,
      "loss": 0.3513,
      "step": 870
    },
    {
      "epoch": 1.88,
      "learning_rate": 4.510921177587845e-05,
      "loss": 0.3319,
      "step": 880
    },
    {
      "epoch": 1.9,
      "learning_rate": 4.499050332383666e-05,
      "loss": 0.3563,
      "step": 890
    },
    {
      "epoch": 1.92,
      "learning_rate": 4.4871794871794874e-05,
      "loss": 0.3692,
      "step": 900
    },
    {
      "epoch": 1.94,
      "learning_rate": 4.4753086419753084e-05,
      "loss": 0.362,
      "step": 910
    },
    {
      "epoch": 1.96,
      "learning_rate": 4.463437796771131e-05,
      "loss": 0.3649,
      "step": 920
    },
    {
      "epoch": 1.99,
      "learning_rate": 4.451566951566952e-05,
      "loss": 0.3531,
      "step": 930
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8684,
      "eval_loss": 0.3599550426006317,
      "eval_runtime": 47.3884,
      "eval_samples_per_second": 211.022,
      "eval_steps_per_second": 6.605,
      "step": 936
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.4396961063627735e-05,
      "loss": 0.4153,
      "step": 940
    },
    {
      "epoch": 2.03,
      "learning_rate": 4.4278252611585945e-05,
      "loss": 0.3254,
      "step": 950
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.415954415954416e-05,
      "loss": 0.3572,
      "step": 960
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.404083570750238e-05,
      "loss": 0.3116,
      "step": 970
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.3922127255460595e-05,
      "loss": 0.3853,
      "step": 980
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.3803418803418805e-05,
      "loss": 0.3605,
      "step": 990
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.368471035137702e-05,
      "loss": 0.3495,
      "step": 1000
    },
    {
      "epoch": 2.16,
      "learning_rate": 4.356600189933523e-05,
      "loss": 0.3736,
      "step": 1010
    },
    {
      "epoch": 2.18,
      "learning_rate": 4.344729344729345e-05,
      "loss": 0.3913,
      "step": 1020
    },
    {
      "epoch": 2.2,
      "learning_rate": 4.3328584995251665e-05,
      "loss": 0.3568,
      "step": 1030
    },
    {
      "epoch": 2.22,
      "learning_rate": 4.3209876543209875e-05,
      "loss": 0.3522,
      "step": 1040
    },
    {
      "epoch": 2.24,
      "learning_rate": 4.309116809116809e-05,
      "loss": 0.3462,
      "step": 1050
    },
    {
      "epoch": 2.26,
      "learning_rate": 4.297245963912631e-05,
      "loss": 0.34,
      "step": 1060
    },
    {
      "epoch": 2.29,
      "learning_rate": 4.2853751187084525e-05,
      "loss": 0.3718,
      "step": 1070
    },
    {
      "epoch": 2.31,
      "learning_rate": 4.2735042735042735e-05,
      "loss": 0.3828,
      "step": 1080
    },
    {
      "epoch": 2.33,
      "learning_rate": 4.261633428300095e-05,
      "loss": 0.3283,
      "step": 1090
    },
    {
      "epoch": 2.35,
      "learning_rate": 4.249762583095916e-05,
      "loss": 0.4015,
      "step": 1100
    },
    {
      "epoch": 2.37,
      "learning_rate": 4.2378917378917385e-05,
      "loss": 0.3469,
      "step": 1110
    },
    {
      "epoch": 2.39,
      "learning_rate": 4.2260208926875595e-05,
      "loss": 0.3562,
      "step": 1120
    },
    {
      "epoch": 2.41,
      "learning_rate": 4.214150047483381e-05,
      "loss": 0.3386,
      "step": 1130
    },
    {
      "epoch": 2.44,
      "learning_rate": 4.202279202279202e-05,
      "loss": 0.3469,
      "step": 1140
    },
    {
      "epoch": 2.46,
      "learning_rate": 4.190408357075024e-05,
      "loss": 0.3164,
      "step": 1150
    },
    {
      "epoch": 2.48,
      "learning_rate": 4.1785375118708455e-05,
      "loss": 0.3688,
      "step": 1160
    },
    {
      "epoch": 2.5,
      "learning_rate": 4.166666666666667e-05,
      "loss": 0.342,
      "step": 1170
    },
    {
      "epoch": 2.52,
      "learning_rate": 4.154795821462488e-05,
      "loss": 0.3089,
      "step": 1180
    },
    {
      "epoch": 2.54,
      "learning_rate": 4.14292497625831e-05,
      "loss": 0.3229,
      "step": 1190
    },
    {
      "epoch": 2.56,
      "learning_rate": 4.131054131054131e-05,
      "loss": 0.3834,
      "step": 1200
    },
    {
      "epoch": 2.58,
      "learning_rate": 4.119183285849953e-05,
      "loss": 0.3545,
      "step": 1210
    },
    {
      "epoch": 2.61,
      "learning_rate": 4.107312440645774e-05,
      "loss": 0.3156,
      "step": 1220
    },
    {
      "epoch": 2.63,
      "learning_rate": 4.095441595441596e-05,
      "loss": 0.3213,
      "step": 1230
    },
    {
      "epoch": 2.65,
      "learning_rate": 4.083570750237417e-05,
      "loss": 0.3634,
      "step": 1240
    },
    {
      "epoch": 2.67,
      "learning_rate": 4.0716999050332386e-05,
      "loss": 0.3813,
      "step": 1250
    },
    {
      "epoch": 2.69,
      "learning_rate": 4.05982905982906e-05,
      "loss": 0.3542,
      "step": 1260
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.047958214624881e-05,
      "loss": 0.3294,
      "step": 1270
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.036087369420703e-05,
      "loss": 0.3166,
      "step": 1280
    },
    {
      "epoch": 2.76,
      "learning_rate": 4.024216524216524e-05,
      "loss": 0.3437,
      "step": 1290
    },
    {
      "epoch": 2.78,
      "learning_rate": 4.012345679012346e-05,
      "loss": 0.3741,
      "step": 1300
    },
    {
      "epoch": 2.8,
      "learning_rate": 4.000474833808167e-05,
      "loss": 0.3283,
      "step": 1310
    },
    {
      "epoch": 2.82,
      "learning_rate": 3.988603988603989e-05,
      "loss": 0.3057,
      "step": 1320
    },
    {
      "epoch": 2.84,
      "learning_rate": 3.97673314339981e-05,
      "loss": 0.3714,
      "step": 1330
    },
    {
      "epoch": 2.86,
      "learning_rate": 3.9648622981956316e-05,
      "loss": 0.3622,
      "step": 1340
    },
    {
      "epoch": 2.88,
      "learning_rate": 3.952991452991453e-05,
      "loss": 0.3114,
      "step": 1350
    },
    {
      "epoch": 2.9,
      "learning_rate": 3.941120607787275e-05,
      "loss": 0.303,
      "step": 1360
    },
    {
      "epoch": 2.93,
      "learning_rate": 3.929249762583096e-05,
      "loss": 0.3716,
      "step": 1370
    },
    {
      "epoch": 2.95,
      "learning_rate": 3.9173789173789176e-05,
      "loss": 0.302,
      "step": 1380
    },
    {
      "epoch": 2.97,
      "learning_rate": 3.9055080721747386e-05,
      "loss": 0.3431,
      "step": 1390
    },
    {
      "epoch": 2.99,
      "learning_rate": 3.893637226970561e-05,
      "loss": 0.3058,
      "step": 1400
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.869,
      "eval_loss": 0.38148394227027893,
      "eval_runtime": 47.5558,
      "eval_samples_per_second": 210.279,
      "eval_steps_per_second": 6.582,
      "step": 1404
    },
    {
      "epoch": 3.01,
      "learning_rate": 3.881766381766382e-05,
      "loss": 0.3738,
      "step": 1410
    },
    {
      "epoch": 3.03,
      "learning_rate": 3.8698955365622036e-05,
      "loss": 0.3314,
      "step": 1420
    },
    {
      "epoch": 3.06,
      "learning_rate": 3.8580246913580246e-05,
      "loss": 0.324,
      "step": 1430
    },
    {
      "epoch": 3.08,
      "learning_rate": 3.846153846153846e-05,
      "loss": 0.3121,
      "step": 1440
    },
    {
      "epoch": 3.1,
      "learning_rate": 3.834283000949668e-05,
      "loss": 0.3253,
      "step": 1450
    },
    {
      "epoch": 3.12,
      "learning_rate": 3.8224121557454896e-05,
      "loss": 0.2757,
      "step": 1460
    },
    {
      "epoch": 3.14,
      "learning_rate": 3.8105413105413106e-05,
      "loss": 0.3142,
      "step": 1470
    },
    {
      "epoch": 3.16,
      "learning_rate": 3.798670465337132e-05,
      "loss": 0.3552,
      "step": 1480
    },
    {
      "epoch": 3.18,
      "learning_rate": 3.786799620132954e-05,
      "loss": 0.2739,
      "step": 1490
    },
    {
      "epoch": 3.2,
      "learning_rate": 3.774928774928775e-05,
      "loss": 0.3667,
      "step": 1500
    },
    {
      "epoch": 3.23,
      "learning_rate": 3.7630579297245966e-05,
      "loss": 0.3323,
      "step": 1510
    },
    {
      "epoch": 3.25,
      "learning_rate": 3.7511870845204176e-05,
      "loss": 0.3214,
      "step": 1520
    },
    {
      "epoch": 3.27,
      "learning_rate": 3.739316239316239e-05,
      "loss": 0.3369,
      "step": 1530
    },
    {
      "epoch": 3.29,
      "learning_rate": 3.727445394112061e-05,
      "loss": 0.3375,
      "step": 1540
    },
    {
      "epoch": 3.31,
      "learning_rate": 3.7155745489078827e-05,
      "loss": 0.3019,
      "step": 1550
    },
    {
      "epoch": 3.33,
      "learning_rate": 3.7037037037037037e-05,
      "loss": 0.3012,
      "step": 1560
    },
    {
      "epoch": 3.35,
      "learning_rate": 3.691832858499525e-05,
      "loss": 0.3172,
      "step": 1570
    },
    {
      "epoch": 3.38,
      "learning_rate": 3.679962013295346e-05,
      "loss": 0.2997,
      "step": 1580
    },
    {
      "epoch": 3.4,
      "learning_rate": 3.668091168091169e-05,
      "loss": 0.3282,
      "step": 1590
    },
    {
      "epoch": 3.42,
      "learning_rate": 3.65622032288699e-05,
      "loss": 0.3156,
      "step": 1600
    },
    {
      "epoch": 3.44,
      "learning_rate": 3.6443494776828113e-05,
      "loss": 0.3159,
      "step": 1610
    },
    {
      "epoch": 3.46,
      "learning_rate": 3.6324786324786323e-05,
      "loss": 0.3228,
      "step": 1620
    },
    {
      "epoch": 3.48,
      "learning_rate": 3.620607787274454e-05,
      "loss": 0.3314,
      "step": 1630
    },
    {
      "epoch": 3.5,
      "learning_rate": 3.608736942070276e-05,
      "loss": 0.3157,
      "step": 1640
    },
    {
      "epoch": 3.52,
      "learning_rate": 3.5968660968660974e-05,
      "loss": 0.3322,
      "step": 1650
    },
    {
      "epoch": 3.55,
      "learning_rate": 3.5849952516619184e-05,
      "loss": 0.3038,
      "step": 1660
    },
    {
      "epoch": 3.57,
      "learning_rate": 3.57312440645774e-05,
      "loss": 0.3524,
      "step": 1670
    },
    {
      "epoch": 3.59,
      "learning_rate": 3.561253561253561e-05,
      "loss": 0.3029,
      "step": 1680
    },
    {
      "epoch": 3.61,
      "learning_rate": 3.5493827160493834e-05,
      "loss": 0.304,
      "step": 1690
    },
    {
      "epoch": 3.63,
      "learning_rate": 3.5375118708452044e-05,
      "loss": 0.2968,
      "step": 1700
    },
    {
      "epoch": 3.65,
      "learning_rate": 3.525641025641026e-05,
      "loss": 0.3294,
      "step": 1710
    },
    {
      "epoch": 3.67,
      "learning_rate": 3.513770180436847e-05,
      "loss": 0.2952,
      "step": 1720
    },
    {
      "epoch": 3.7,
      "learning_rate": 3.501899335232669e-05,
      "loss": 0.3556,
      "step": 1730
    },
    {
      "epoch": 3.72,
      "learning_rate": 3.4900284900284904e-05,
      "loss": 0.2956,
      "step": 1740
    },
    {
      "epoch": 3.74,
      "learning_rate": 3.4781576448243114e-05,
      "loss": 0.3118,
      "step": 1750
    },
    {
      "epoch": 3.76,
      "learning_rate": 3.466286799620133e-05,
      "loss": 0.3057,
      "step": 1760
    },
    {
      "epoch": 3.78,
      "learning_rate": 3.454415954415954e-05,
      "loss": 0.2832,
      "step": 1770
    },
    {
      "epoch": 3.8,
      "learning_rate": 3.4425451092117764e-05,
      "loss": 0.3224,
      "step": 1780
    },
    {
      "epoch": 3.82,
      "learning_rate": 3.4306742640075974e-05,
      "loss": 0.3237,
      "step": 1790
    },
    {
      "epoch": 3.84,
      "learning_rate": 3.418803418803419e-05,
      "loss": 0.2871,
      "step": 1800
    },
    {
      "epoch": 3.87,
      "learning_rate": 3.40693257359924e-05,
      "loss": 0.3196,
      "step": 1810
    },
    {
      "epoch": 3.89,
      "learning_rate": 3.395061728395062e-05,
      "loss": 0.3489,
      "step": 1820
    },
    {
      "epoch": 3.91,
      "learning_rate": 3.3831908831908834e-05,
      "loss": 0.3108,
      "step": 1830
    },
    {
      "epoch": 3.93,
      "learning_rate": 3.371320037986705e-05,
      "loss": 0.3285,
      "step": 1840
    },
    {
      "epoch": 3.95,
      "learning_rate": 3.359449192782526e-05,
      "loss": 0.2986,
      "step": 1850
    },
    {
      "epoch": 3.97,
      "learning_rate": 3.347578347578348e-05,
      "loss": 0.301,
      "step": 1860
    },
    {
      "epoch": 3.99,
      "learning_rate": 3.335707502374169e-05,
      "loss": 0.2819,
      "step": 1870
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8852,
      "eval_loss": 0.32173556089401245,
      "eval_runtime": 47.4022,
      "eval_samples_per_second": 210.961,
      "eval_steps_per_second": 6.603,
      "step": 1872
    },
    {
      "epoch": 4.02,
      "learning_rate": 3.323836657169991e-05,
      "loss": 0.309,
      "step": 1880
    },
    {
      "epoch": 4.04,
      "learning_rate": 3.311965811965812e-05,
      "loss": 0.3257,
      "step": 1890
    },
    {
      "epoch": 4.06,
      "learning_rate": 3.300094966761634e-05,
      "loss": 0.2827,
      "step": 1900
    },
    {
      "epoch": 4.08,
      "learning_rate": 3.288224121557455e-05,
      "loss": 0.3234,
      "step": 1910
    },
    {
      "epoch": 4.1,
      "learning_rate": 3.2763532763532764e-05,
      "loss": 0.3303,
      "step": 1920
    },
    {
      "epoch": 4.12,
      "learning_rate": 3.264482431149098e-05,
      "loss": 0.2984,
      "step": 1930
    },
    {
      "epoch": 4.15,
      "learning_rate": 3.25261158594492e-05,
      "loss": 0.2868,
      "step": 1940
    },
    {
      "epoch": 4.17,
      "learning_rate": 3.240740740740741e-05,
      "loss": 0.3303,
      "step": 1950
    },
    {
      "epoch": 4.19,
      "learning_rate": 3.2288698955365625e-05,
      "loss": 0.2711,
      "step": 1960
    },
    {
      "epoch": 4.21,
      "learning_rate": 3.216999050332384e-05,
      "loss": 0.3173,
      "step": 1970
    },
    {
      "epoch": 4.23,
      "learning_rate": 3.205128205128206e-05,
      "loss": 0.3103,
      "step": 1980
    },
    {
      "epoch": 4.25,
      "learning_rate": 3.193257359924027e-05,
      "loss": 0.3114,
      "step": 1990
    },
    {
      "epoch": 4.27,
      "learning_rate": 3.181386514719848e-05,
      "loss": 0.271,
      "step": 2000
    },
    {
      "epoch": 4.29,
      "learning_rate": 3.1695156695156695e-05,
      "loss": 0.2834,
      "step": 2010
    },
    {
      "epoch": 4.32,
      "learning_rate": 3.157644824311491e-05,
      "loss": 0.2812,
      "step": 2020
    },
    {
      "epoch": 4.34,
      "learning_rate": 3.145773979107313e-05,
      "loss": 0.3213,
      "step": 2030
    },
    {
      "epoch": 4.36,
      "learning_rate": 3.133903133903134e-05,
      "loss": 0.2381,
      "step": 2040
    },
    {
      "epoch": 4.38,
      "learning_rate": 3.1220322886989555e-05,
      "loss": 0.3131,
      "step": 2050
    },
    {
      "epoch": 4.4,
      "learning_rate": 3.1101614434947765e-05,
      "loss": 0.2921,
      "step": 2060
    },
    {
      "epoch": 4.42,
      "learning_rate": 3.098290598290599e-05,
      "loss": 0.3018,
      "step": 2070
    },
    {
      "epoch": 4.44,
      "learning_rate": 3.08641975308642e-05,
      "loss": 0.304,
      "step": 2080
    },
    {
      "epoch": 4.47,
      "learning_rate": 3.0745489078822415e-05,
      "loss": 0.3204,
      "step": 2090
    },
    {
      "epoch": 4.49,
      "learning_rate": 3.0626780626780625e-05,
      "loss": 0.2933,
      "step": 2100
    },
    {
      "epoch": 4.51,
      "learning_rate": 3.0508072174738845e-05,
      "loss": 0.2772,
      "step": 2110
    },
    {
      "epoch": 4.53,
      "learning_rate": 3.0389363722697055e-05,
      "loss": 0.2849,
      "step": 2120
    },
    {
      "epoch": 4.55,
      "learning_rate": 3.0270655270655275e-05,
      "loss": 0.3125,
      "step": 2130
    },
    {
      "epoch": 4.57,
      "learning_rate": 3.0151946818613485e-05,
      "loss": 0.3005,
      "step": 2140
    },
    {
      "epoch": 4.59,
      "learning_rate": 3.0033238366571702e-05,
      "loss": 0.2965,
      "step": 2150
    },
    {
      "epoch": 4.61,
      "learning_rate": 2.9914529914529915e-05,
      "loss": 0.2634,
      "step": 2160
    },
    {
      "epoch": 4.64,
      "learning_rate": 2.9795821462488132e-05,
      "loss": 0.3093,
      "step": 2170
    },
    {
      "epoch": 4.66,
      "learning_rate": 2.9677113010446345e-05,
      "loss": 0.2865,
      "step": 2180
    },
    {
      "epoch": 4.68,
      "learning_rate": 2.9558404558404562e-05,
      "loss": 0.2939,
      "step": 2190
    },
    {
      "epoch": 4.7,
      "learning_rate": 2.9439696106362775e-05,
      "loss": 0.3056,
      "step": 2200
    },
    {
      "epoch": 4.72,
      "learning_rate": 2.9320987654320992e-05,
      "loss": 0.2601,
      "step": 2210
    },
    {
      "epoch": 4.74,
      "learning_rate": 2.9202279202279202e-05,
      "loss": 0.2709,
      "step": 2220
    },
    {
      "epoch": 4.76,
      "learning_rate": 2.9083570750237422e-05,
      "loss": 0.2963,
      "step": 2230
    },
    {
      "epoch": 4.79,
      "learning_rate": 2.8964862298195632e-05,
      "loss": 0.2637,
      "step": 2240
    },
    {
      "epoch": 4.81,
      "learning_rate": 2.8846153846153845e-05,
      "loss": 0.2667,
      "step": 2250
    },
    {
      "epoch": 4.83,
      "learning_rate": 2.8727445394112062e-05,
      "loss": 0.3028,
      "step": 2260
    },
    {
      "epoch": 4.85,
      "learning_rate": 2.8608736942070276e-05,
      "loss": 0.289,
      "step": 2270
    },
    {
      "epoch": 4.87,
      "learning_rate": 2.8490028490028492e-05,
      "loss": 0.2918,
      "step": 2280
    },
    {
      "epoch": 4.89,
      "learning_rate": 2.8371320037986702e-05,
      "loss": 0.2907,
      "step": 2290
    },
    {
      "epoch": 4.91,
      "learning_rate": 2.8252611585944922e-05,
      "loss": 0.2824,
      "step": 2300
    },
    {
      "epoch": 4.93,
      "learning_rate": 2.8133903133903132e-05,
      "loss": 0.2747,
      "step": 2310
    },
    {
      "epoch": 4.96,
      "learning_rate": 2.8015194681861352e-05,
      "loss": 0.281,
      "step": 2320
    },
    {
      "epoch": 4.98,
      "learning_rate": 2.7896486229819562e-05,
      "loss": 0.2534,
      "step": 2330
    },
    {
      "epoch": 5.0,
      "learning_rate": 2.777777777777778e-05,
      "loss": 0.2799,
      "step": 2340
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.8945,
      "eval_loss": 0.2917763590812683,
      "eval_runtime": 47.4876,
      "eval_samples_per_second": 210.581,
      "eval_steps_per_second": 6.591,
      "step": 2340
    },
    {
      "epoch": 5.02,
      "learning_rate": 2.7659069325735992e-05,
      "loss": 0.3173,
      "step": 2350
    },
    {
      "epoch": 5.04,
      "learning_rate": 2.754036087369421e-05,
      "loss": 0.2615,
      "step": 2360
    },
    {
      "epoch": 5.06,
      "learning_rate": 2.7421652421652423e-05,
      "loss": 0.2444,
      "step": 2370
    },
    {
      "epoch": 5.09,
      "learning_rate": 2.730294396961064e-05,
      "loss": 0.2709,
      "step": 2380
    },
    {
      "epoch": 5.11,
      "learning_rate": 2.7184235517568853e-05,
      "loss": 0.2531,
      "step": 2390
    },
    {
      "epoch": 5.13,
      "learning_rate": 2.706552706552707e-05,
      "loss": 0.2695,
      "step": 2400
    },
    {
      "epoch": 5.15,
      "learning_rate": 2.694681861348528e-05,
      "loss": 0.2852,
      "step": 2410
    },
    {
      "epoch": 5.17,
      "learning_rate": 2.68281101614435e-05,
      "loss": 0.2897,
      "step": 2420
    },
    {
      "epoch": 5.19,
      "learning_rate": 2.670940170940171e-05,
      "loss": 0.313,
      "step": 2430
    },
    {
      "epoch": 5.21,
      "learning_rate": 2.6590693257359926e-05,
      "loss": 0.2665,
      "step": 2440
    },
    {
      "epoch": 5.23,
      "learning_rate": 2.647198480531814e-05,
      "loss": 0.2549,
      "step": 2450
    },
    {
      "epoch": 5.26,
      "learning_rate": 2.6353276353276356e-05,
      "loss": 0.292,
      "step": 2460
    },
    {
      "epoch": 5.28,
      "learning_rate": 2.623456790123457e-05,
      "loss": 0.2575,
      "step": 2470
    },
    {
      "epoch": 5.3,
      "learning_rate": 2.611585944919278e-05,
      "loss": 0.2421,
      "step": 2480
    },
    {
      "epoch": 5.32,
      "learning_rate": 2.5997150997151e-05,
      "loss": 0.226,
      "step": 2490
    },
    {
      "epoch": 5.34,
      "learning_rate": 2.587844254510921e-05,
      "loss": 0.2378,
      "step": 2500
    },
    {
      "epoch": 5.36,
      "learning_rate": 2.575973409306743e-05,
      "loss": 0.2576,
      "step": 2510
    },
    {
      "epoch": 5.38,
      "learning_rate": 2.564102564102564e-05,
      "loss": 0.2227,
      "step": 2520
    },
    {
      "epoch": 5.41,
      "learning_rate": 2.5522317188983856e-05,
      "loss": 0.2704,
      "step": 2530
    },
    {
      "epoch": 5.43,
      "learning_rate": 2.540360873694207e-05,
      "loss": 0.2817,
      "step": 2540
    },
    {
      "epoch": 5.45,
      "learning_rate": 2.5284900284900286e-05,
      "loss": 0.3046,
      "step": 2550
    },
    {
      "epoch": 5.47,
      "learning_rate": 2.51661918328585e-05,
      "loss": 0.2461,
      "step": 2560
    },
    {
      "epoch": 5.49,
      "learning_rate": 2.5047483380816717e-05,
      "loss": 0.2875,
      "step": 2570
    },
    {
      "epoch": 5.51,
      "learning_rate": 2.492877492877493e-05,
      "loss": 0.2858,
      "step": 2580
    },
    {
      "epoch": 5.53,
      "learning_rate": 2.4810066476733147e-05,
      "loss": 0.2619,
      "step": 2590
    },
    {
      "epoch": 5.55,
      "learning_rate": 2.4691358024691357e-05,
      "loss": 0.2492,
      "step": 2600
    },
    {
      "epoch": 5.58,
      "learning_rate": 2.4572649572649573e-05,
      "loss": 0.26,
      "step": 2610
    },
    {
      "epoch": 5.6,
      "learning_rate": 2.4453941120607787e-05,
      "loss": 0.2629,
      "step": 2620
    },
    {
      "epoch": 5.62,
      "learning_rate": 2.4335232668566003e-05,
      "loss": 0.2853,
      "step": 2630
    },
    {
      "epoch": 5.64,
      "learning_rate": 2.4216524216524217e-05,
      "loss": 0.2677,
      "step": 2640
    },
    {
      "epoch": 5.66,
      "learning_rate": 2.409781576448243e-05,
      "loss": 0.2771,
      "step": 2650
    },
    {
      "epoch": 5.68,
      "learning_rate": 2.3979107312440647e-05,
      "loss": 0.3032,
      "step": 2660
    },
    {
      "epoch": 5.7,
      "learning_rate": 2.386039886039886e-05,
      "loss": 0.2739,
      "step": 2670
    },
    {
      "epoch": 5.73,
      "learning_rate": 2.3741690408357077e-05,
      "loss": 0.2901,
      "step": 2680
    },
    {
      "epoch": 5.75,
      "learning_rate": 2.362298195631529e-05,
      "loss": 0.2492,
      "step": 2690
    },
    {
      "epoch": 5.77,
      "learning_rate": 2.3504273504273504e-05,
      "loss": 0.2888,
      "step": 2700
    },
    {
      "epoch": 5.79,
      "learning_rate": 2.338556505223172e-05,
      "loss": 0.2704,
      "step": 2710
    },
    {
      "epoch": 5.81,
      "learning_rate": 2.3266856600189934e-05,
      "loss": 0.2928,
      "step": 2720
    },
    {
      "epoch": 5.83,
      "learning_rate": 2.314814814814815e-05,
      "loss": 0.2615,
      "step": 2730
    },
    {
      "epoch": 5.85,
      "learning_rate": 2.3029439696106364e-05,
      "loss": 0.2785,
      "step": 2740
    },
    {
      "epoch": 5.87,
      "learning_rate": 2.291073124406458e-05,
      "loss": 0.2686,
      "step": 2750
    },
    {
      "epoch": 5.9,
      "learning_rate": 2.2792022792022794e-05,
      "loss": 0.2736,
      "step": 2760
    },
    {
      "epoch": 5.92,
      "learning_rate": 2.2673314339981007e-05,
      "loss": 0.2616,
      "step": 2770
    },
    {
      "epoch": 5.94,
      "learning_rate": 2.2554605887939224e-05,
      "loss": 0.2593,
      "step": 2780
    },
    {
      "epoch": 5.96,
      "learning_rate": 2.2435897435897437e-05,
      "loss": 0.2561,
      "step": 2790
    },
    {
      "epoch": 5.98,
      "learning_rate": 2.2317188983855654e-05,
      "loss": 0.2804,
      "step": 2800
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.8983,
      "eval_loss": 0.2862694263458252,
      "eval_runtime": 47.5296,
      "eval_samples_per_second": 210.395,
      "eval_steps_per_second": 6.585,
      "step": 2808
    },
    {
      "epoch": 6.0,
      "learning_rate": 2.2198480531813867e-05,
      "loss": 0.2877,
      "step": 2810
    },
    {
      "epoch": 6.03,
      "learning_rate": 2.207977207977208e-05,
      "loss": 0.245,
      "step": 2820
    },
    {
      "epoch": 6.05,
      "learning_rate": 2.1961063627730297e-05,
      "loss": 0.2603,
      "step": 2830
    },
    {
      "epoch": 6.07,
      "learning_rate": 2.184235517568851e-05,
      "loss": 0.2685,
      "step": 2840
    },
    {
      "epoch": 6.09,
      "learning_rate": 2.1723646723646724e-05,
      "loss": 0.2556,
      "step": 2850
    },
    {
      "epoch": 6.11,
      "learning_rate": 2.1604938271604937e-05,
      "loss": 0.2698,
      "step": 2860
    },
    {
      "epoch": 6.13,
      "learning_rate": 2.1486229819563154e-05,
      "loss": 0.2491,
      "step": 2870
    },
    {
      "epoch": 6.15,
      "learning_rate": 2.1367521367521368e-05,
      "loss": 0.2746,
      "step": 2880
    },
    {
      "epoch": 6.17,
      "learning_rate": 2.124881291547958e-05,
      "loss": 0.2613,
      "step": 2890
    },
    {
      "epoch": 6.2,
      "learning_rate": 2.1130104463437798e-05,
      "loss": 0.2991,
      "step": 2900
    },
    {
      "epoch": 6.22,
      "learning_rate": 2.101139601139601e-05,
      "loss": 0.302,
      "step": 2910
    },
    {
      "epoch": 6.24,
      "learning_rate": 2.0892687559354228e-05,
      "loss": 0.2531,
      "step": 2920
    },
    {
      "epoch": 6.26,
      "learning_rate": 2.077397910731244e-05,
      "loss": 0.2409,
      "step": 2930
    },
    {
      "epoch": 6.28,
      "learning_rate": 2.0655270655270654e-05,
      "loss": 0.2682,
      "step": 2940
    },
    {
      "epoch": 6.3,
      "learning_rate": 2.053656220322887e-05,
      "loss": 0.2561,
      "step": 2950
    },
    {
      "epoch": 6.32,
      "learning_rate": 2.0417853751187084e-05,
      "loss": 0.2318,
      "step": 2960
    },
    {
      "epoch": 6.35,
      "learning_rate": 2.02991452991453e-05,
      "loss": 0.2551,
      "step": 2970
    },
    {
      "epoch": 6.37,
      "learning_rate": 2.0180436847103515e-05,
      "loss": 0.2669,
      "step": 2980
    },
    {
      "epoch": 6.39,
      "learning_rate": 2.006172839506173e-05,
      "loss": 0.2757,
      "step": 2990
    },
    {
      "epoch": 6.41,
      "learning_rate": 1.9943019943019945e-05,
      "loss": 0.2824,
      "step": 3000
    },
    {
      "epoch": 6.43,
      "learning_rate": 1.9824311490978158e-05,
      "loss": 0.2473,
      "step": 3010
    },
    {
      "epoch": 6.45,
      "learning_rate": 1.9705603038936375e-05,
      "loss": 0.2987,
      "step": 3020
    },
    {
      "epoch": 6.47,
      "learning_rate": 1.9586894586894588e-05,
      "loss": 0.2605,
      "step": 3030
    },
    {
      "epoch": 6.49,
      "learning_rate": 1.9468186134852805e-05,
      "loss": 0.2253,
      "step": 3040
    },
    {
      "epoch": 6.52,
      "learning_rate": 1.9349477682811018e-05,
      "loss": 0.2572,
      "step": 3050
    },
    {
      "epoch": 6.54,
      "learning_rate": 1.923076923076923e-05,
      "loss": 0.2489,
      "step": 3060
    },
    {
      "epoch": 6.56,
      "learning_rate": 1.9112060778727448e-05,
      "loss": 0.2797,
      "step": 3070
    },
    {
      "epoch": 6.58,
      "learning_rate": 1.899335232668566e-05,
      "loss": 0.2779,
      "step": 3080
    },
    {
      "epoch": 6.6,
      "learning_rate": 1.8874643874643875e-05,
      "loss": 0.2607,
      "step": 3090
    },
    {
      "epoch": 6.62,
      "learning_rate": 1.8755935422602088e-05,
      "loss": 0.2576,
      "step": 3100
    },
    {
      "epoch": 6.64,
      "learning_rate": 1.8637226970560305e-05,
      "loss": 0.2666,
      "step": 3110
    },
    {
      "epoch": 6.67,
      "learning_rate": 1.8518518518518518e-05,
      "loss": 0.2659,
      "step": 3120
    },
    {
      "epoch": 6.69,
      "learning_rate": 1.839981006647673e-05,
      "loss": 0.2257,
      "step": 3130
    },
    {
      "epoch": 6.71,
      "learning_rate": 1.828110161443495e-05,
      "loss": 0.2591,
      "step": 3140
    },
    {
      "epoch": 6.73,
      "learning_rate": 1.8162393162393162e-05,
      "loss": 0.2346,
      "step": 3150
    },
    {
      "epoch": 6.75,
      "learning_rate": 1.804368471035138e-05,
      "loss": 0.2618,
      "step": 3160
    },
    {
      "epoch": 6.77,
      "learning_rate": 1.7924976258309592e-05,
      "loss": 0.2356,
      "step": 3170
    },
    {
      "epoch": 6.79,
      "learning_rate": 1.7806267806267805e-05,
      "loss": 0.2741,
      "step": 3180
    },
    {
      "epoch": 6.81,
      "learning_rate": 1.7687559354226022e-05,
      "loss": 0.2494,
      "step": 3190
    },
    {
      "epoch": 6.84,
      "learning_rate": 1.7568850902184235e-05,
      "loss": 0.286,
      "step": 3200
    },
    {
      "epoch": 6.86,
      "learning_rate": 1.7450142450142452e-05,
      "loss": 0.2277,
      "step": 3210
    },
    {
      "epoch": 6.88,
      "learning_rate": 1.7331433998100665e-05,
      "loss": 0.2308,
      "step": 3220
    },
    {
      "epoch": 6.9,
      "learning_rate": 1.7212725546058882e-05,
      "loss": 0.2491,
      "step": 3230
    },
    {
      "epoch": 6.92,
      "learning_rate": 1.7094017094017095e-05,
      "loss": 0.2251,
      "step": 3240
    },
    {
      "epoch": 6.94,
      "learning_rate": 1.697530864197531e-05,
      "loss": 0.2406,
      "step": 3250
    },
    {
      "epoch": 6.96,
      "learning_rate": 1.6856600189933525e-05,
      "loss": 0.2626,
      "step": 3260
    },
    {
      "epoch": 6.99,
      "learning_rate": 1.673789173789174e-05,
      "loss": 0.235,
      "step": 3270
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.8954,
      "eval_loss": 0.2886597216129303,
      "eval_runtime": 47.6094,
      "eval_samples_per_second": 210.043,
      "eval_steps_per_second": 6.574,
      "step": 3276
    },
    {
      "epoch": 7.01,
      "learning_rate": 1.6619183285849956e-05,
      "loss": 0.2382,
      "step": 3280
    },
    {
      "epoch": 7.03,
      "learning_rate": 1.650047483380817e-05,
      "loss": 0.2688,
      "step": 3290
    },
    {
      "epoch": 7.05,
      "learning_rate": 1.6381766381766382e-05,
      "loss": 0.2459,
      "step": 3300
    },
    {
      "epoch": 7.07,
      "learning_rate": 1.62630579297246e-05,
      "loss": 0.2309,
      "step": 3310
    },
    {
      "epoch": 7.09,
      "learning_rate": 1.6144349477682812e-05,
      "loss": 0.2354,
      "step": 3320
    },
    {
      "epoch": 7.12,
      "learning_rate": 1.602564102564103e-05,
      "loss": 0.2357,
      "step": 3330
    },
    {
      "epoch": 7.14,
      "learning_rate": 1.590693257359924e-05,
      "loss": 0.2388,
      "step": 3340
    },
    {
      "epoch": 7.16,
      "learning_rate": 1.5788224121557456e-05,
      "loss": 0.2699,
      "step": 3350
    },
    {
      "epoch": 7.18,
      "learning_rate": 1.566951566951567e-05,
      "loss": 0.2446,
      "step": 3360
    },
    {
      "epoch": 7.2,
      "learning_rate": 1.5550807217473882e-05,
      "loss": 0.292,
      "step": 3370
    },
    {
      "epoch": 7.22,
      "learning_rate": 1.54320987654321e-05,
      "loss": 0.2085,
      "step": 3380
    },
    {
      "epoch": 7.24,
      "learning_rate": 1.5313390313390312e-05,
      "loss": 0.2597,
      "step": 3390
    },
    {
      "epoch": 7.26,
      "learning_rate": 1.5194681861348528e-05,
      "loss": 0.2508,
      "step": 3400
    },
    {
      "epoch": 7.29,
      "learning_rate": 1.5075973409306743e-05,
      "loss": 0.2794,
      "step": 3410
    },
    {
      "epoch": 7.31,
      "learning_rate": 1.4957264957264958e-05,
      "loss": 0.2556,
      "step": 3420
    },
    {
      "epoch": 7.33,
      "learning_rate": 1.4838556505223173e-05,
      "loss": 0.2471,
      "step": 3430
    },
    {
      "epoch": 7.35,
      "learning_rate": 1.4719848053181388e-05,
      "loss": 0.2261,
      "step": 3440
    },
    {
      "epoch": 7.37,
      "learning_rate": 1.4601139601139601e-05,
      "loss": 0.2405,
      "step": 3450
    },
    {
      "epoch": 7.39,
      "learning_rate": 1.4482431149097816e-05,
      "loss": 0.2278,
      "step": 3460
    },
    {
      "epoch": 7.41,
      "learning_rate": 1.4363722697056031e-05,
      "loss": 0.2672,
      "step": 3470
    },
    {
      "epoch": 7.44,
      "learning_rate": 1.4245014245014246e-05,
      "loss": 0.2362,
      "step": 3480
    },
    {
      "epoch": 7.46,
      "learning_rate": 1.4126305792972461e-05,
      "loss": 0.2845,
      "step": 3490
    },
    {
      "epoch": 7.48,
      "learning_rate": 1.4007597340930676e-05,
      "loss": 0.2431,
      "step": 3500
    },
    {
      "epoch": 7.5,
      "learning_rate": 1.388888888888889e-05,
      "loss": 0.238,
      "step": 3510
    },
    {
      "epoch": 7.52,
      "learning_rate": 1.3770180436847105e-05,
      "loss": 0.246,
      "step": 3520
    },
    {
      "epoch": 7.54,
      "learning_rate": 1.365147198480532e-05,
      "loss": 0.2536,
      "step": 3530
    },
    {
      "epoch": 7.56,
      "learning_rate": 1.3532763532763535e-05,
      "loss": 0.2625,
      "step": 3540
    },
    {
      "epoch": 7.58,
      "learning_rate": 1.341405508072175e-05,
      "loss": 0.2571,
      "step": 3550
    },
    {
      "epoch": 7.61,
      "learning_rate": 1.3295346628679963e-05,
      "loss": 0.2519,
      "step": 3560
    },
    {
      "epoch": 7.63,
      "learning_rate": 1.3176638176638178e-05,
      "loss": 0.2648,
      "step": 3570
    },
    {
      "epoch": 7.65,
      "learning_rate": 1.305792972459639e-05,
      "loss": 0.2178,
      "step": 3580
    },
    {
      "epoch": 7.67,
      "learning_rate": 1.2939221272554605e-05,
      "loss": 0.2561,
      "step": 3590
    },
    {
      "epoch": 7.69,
      "learning_rate": 1.282051282051282e-05,
      "loss": 0.2007,
      "step": 3600
    },
    {
      "epoch": 7.71,
      "learning_rate": 1.2701804368471035e-05,
      "loss": 0.2175,
      "step": 3610
    },
    {
      "epoch": 7.73,
      "learning_rate": 1.258309591642925e-05,
      "loss": 0.2447,
      "step": 3620
    },
    {
      "epoch": 7.76,
      "learning_rate": 1.2464387464387465e-05,
      "loss": 0.2207,
      "step": 3630
    },
    {
      "epoch": 7.78,
      "learning_rate": 1.2345679012345678e-05,
      "loss": 0.2517,
      "step": 3640
    },
    {
      "epoch": 7.8,
      "learning_rate": 1.2226970560303893e-05,
      "loss": 0.2707,
      "step": 3650
    },
    {
      "epoch": 7.82,
      "learning_rate": 1.2108262108262108e-05,
      "loss": 0.2298,
      "step": 3660
    },
    {
      "epoch": 7.84,
      "learning_rate": 1.1989553656220323e-05,
      "loss": 0.214,
      "step": 3670
    },
    {
      "epoch": 7.86,
      "learning_rate": 1.1870845204178538e-05,
      "loss": 0.226,
      "step": 3680
    },
    {
      "epoch": 7.88,
      "learning_rate": 1.1752136752136752e-05,
      "loss": 0.2001,
      "step": 3690
    },
    {
      "epoch": 7.9,
      "learning_rate": 1.1633428300094967e-05,
      "loss": 0.2323,
      "step": 3700
    },
    {
      "epoch": 7.93,
      "learning_rate": 1.1514719848053182e-05,
      "loss": 0.2276,
      "step": 3710
    },
    {
      "epoch": 7.95,
      "learning_rate": 1.1396011396011397e-05,
      "loss": 0.2236,
      "step": 3720
    },
    {
      "epoch": 7.97,
      "learning_rate": 1.1277302943969612e-05,
      "loss": 0.2241,
      "step": 3730
    },
    {
      "epoch": 7.99,
      "learning_rate": 1.1158594491927827e-05,
      "loss": 0.2684,
      "step": 3740
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.8965,
      "eval_loss": 0.2847461998462677,
      "eval_runtime": 47.6902,
      "eval_samples_per_second": 209.687,
      "eval_steps_per_second": 6.563,
      "step": 3744
    },
    {
      "epoch": 8.01,
      "learning_rate": 1.103988603988604e-05,
      "loss": 0.2173,
      "step": 3750
    },
    {
      "epoch": 8.03,
      "learning_rate": 1.0921177587844255e-05,
      "loss": 0.2211,
      "step": 3760
    },
    {
      "epoch": 8.06,
      "learning_rate": 1.0802469135802469e-05,
      "loss": 0.2381,
      "step": 3770
    },
    {
      "epoch": 8.08,
      "learning_rate": 1.0683760683760684e-05,
      "loss": 0.2001,
      "step": 3780
    },
    {
      "epoch": 8.1,
      "learning_rate": 1.0565052231718899e-05,
      "loss": 0.2414,
      "step": 3790
    },
    {
      "epoch": 8.12,
      "learning_rate": 1.0446343779677114e-05,
      "loss": 0.2017,
      "step": 3800
    },
    {
      "epoch": 8.14,
      "learning_rate": 1.0327635327635327e-05,
      "loss": 0.2335,
      "step": 3810
    },
    {
      "epoch": 8.16,
      "learning_rate": 1.0208926875593542e-05,
      "loss": 0.2428,
      "step": 3820
    },
    {
      "epoch": 8.18,
      "learning_rate": 1.0090218423551757e-05,
      "loss": 0.232,
      "step": 3830
    },
    {
      "epoch": 8.2,
      "learning_rate": 9.971509971509972e-06,
      "loss": 0.218,
      "step": 3840
    },
    {
      "epoch": 8.23,
      "learning_rate": 9.852801519468187e-06,
      "loss": 0.2295,
      "step": 3850
    },
    {
      "epoch": 8.25,
      "learning_rate": 9.734093067426402e-06,
      "loss": 0.2432,
      "step": 3860
    },
    {
      "epoch": 8.27,
      "learning_rate": 9.615384615384616e-06,
      "loss": 0.216,
      "step": 3870
    },
    {
      "epoch": 8.29,
      "learning_rate": 9.49667616334283e-06,
      "loss": 0.2608,
      "step": 3880
    },
    {
      "epoch": 8.31,
      "learning_rate": 9.377967711301044e-06,
      "loss": 0.2347,
      "step": 3890
    },
    {
      "epoch": 8.33,
      "learning_rate": 9.259259259259259e-06,
      "loss": 0.2477,
      "step": 3900
    },
    {
      "epoch": 8.35,
      "learning_rate": 9.140550807217474e-06,
      "loss": 0.2198,
      "step": 3910
    },
    {
      "epoch": 8.38,
      "learning_rate": 9.02184235517569e-06,
      "loss": 0.215,
      "step": 3920
    },
    {
      "epoch": 8.4,
      "learning_rate": 8.903133903133903e-06,
      "loss": 0.215,
      "step": 3930
    },
    {
      "epoch": 8.42,
      "learning_rate": 8.784425451092118e-06,
      "loss": 0.2208,
      "step": 3940
    },
    {
      "epoch": 8.44,
      "learning_rate": 8.665716999050333e-06,
      "loss": 0.2278,
      "step": 3950
    },
    {
      "epoch": 8.46,
      "learning_rate": 8.547008547008548e-06,
      "loss": 0.2671,
      "step": 3960
    },
    {
      "epoch": 8.48,
      "learning_rate": 8.428300094966763e-06,
      "loss": 0.2441,
      "step": 3970
    },
    {
      "epoch": 8.5,
      "learning_rate": 8.309591642924978e-06,
      "loss": 0.2254,
      "step": 3980
    },
    {
      "epoch": 8.52,
      "learning_rate": 8.190883190883191e-06,
      "loss": 0.2045,
      "step": 3990
    },
    {
      "epoch": 8.55,
      "learning_rate": 8.072174738841406e-06,
      "loss": 0.2381,
      "step": 4000
    },
    {
      "epoch": 8.57,
      "learning_rate": 7.95346628679962e-06,
      "loss": 0.2158,
      "step": 4010
    },
    {
      "epoch": 8.59,
      "learning_rate": 7.834757834757835e-06,
      "loss": 0.2409,
      "step": 4020
    },
    {
      "epoch": 8.61,
      "learning_rate": 7.71604938271605e-06,
      "loss": 0.2086,
      "step": 4030
    },
    {
      "epoch": 8.63,
      "learning_rate": 7.597340930674264e-06,
      "loss": 0.2148,
      "step": 4040
    },
    {
      "epoch": 8.65,
      "learning_rate": 7.478632478632479e-06,
      "loss": 0.2218,
      "step": 4050
    },
    {
      "epoch": 8.67,
      "learning_rate": 7.359924026590694e-06,
      "loss": 0.2449,
      "step": 4060
    },
    {
      "epoch": 8.7,
      "learning_rate": 7.241215574548908e-06,
      "loss": 0.242,
      "step": 4070
    },
    {
      "epoch": 8.72,
      "learning_rate": 7.122507122507123e-06,
      "loss": 0.2034,
      "step": 4080
    },
    {
      "epoch": 8.74,
      "learning_rate": 7.003798670465338e-06,
      "loss": 0.2272,
      "step": 4090
    },
    {
      "epoch": 8.76,
      "learning_rate": 6.885090218423552e-06,
      "loss": 0.1943,
      "step": 4100
    },
    {
      "epoch": 8.78,
      "learning_rate": 6.766381766381767e-06,
      "loss": 0.2313,
      "step": 4110
    },
    {
      "epoch": 8.8,
      "learning_rate": 6.6476733143399815e-06,
      "loss": 0.2217,
      "step": 4120
    },
    {
      "epoch": 8.82,
      "learning_rate": 6.528964862298195e-06,
      "loss": 0.255,
      "step": 4130
    },
    {
      "epoch": 8.84,
      "learning_rate": 6.41025641025641e-06,
      "loss": 0.2623,
      "step": 4140
    },
    {
      "epoch": 8.87,
      "learning_rate": 6.291547958214625e-06,
      "loss": 0.2427,
      "step": 4150
    },
    {
      "epoch": 8.89,
      "learning_rate": 6.172839506172839e-06,
      "loss": 0.2157,
      "step": 4160
    },
    {
      "epoch": 8.91,
      "learning_rate": 6.054131054131054e-06,
      "loss": 0.2486,
      "step": 4170
    },
    {
      "epoch": 8.93,
      "learning_rate": 5.935422602089269e-06,
      "loss": 0.2439,
      "step": 4180
    },
    {
      "epoch": 8.95,
      "learning_rate": 5.816714150047483e-06,
      "loss": 0.2476,
      "step": 4190
    },
    {
      "epoch": 8.97,
      "learning_rate": 5.6980056980056985e-06,
      "loss": 0.2464,
      "step": 4200
    },
    {
      "epoch": 8.99,
      "learning_rate": 5.5792972459639135e-06,
      "loss": 0.2276,
      "step": 4210
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.9072,
      "eval_loss": 0.2635934352874756,
      "eval_runtime": 47.7546,
      "eval_samples_per_second": 209.404,
      "eval_steps_per_second": 6.554,
      "step": 4212
    },
    {
      "epoch": 9.02,
      "learning_rate": 5.460588793922128e-06,
      "loss": 0.2186,
      "step": 4220
    },
    {
      "epoch": 9.04,
      "learning_rate": 5.341880341880342e-06,
      "loss": 0.2298,
      "step": 4230
    },
    {
      "epoch": 9.06,
      "learning_rate": 5.223171889838557e-06,
      "loss": 0.2186,
      "step": 4240
    },
    {
      "epoch": 9.08,
      "learning_rate": 5.104463437796771e-06,
      "loss": 0.2331,
      "step": 4250
    },
    {
      "epoch": 9.1,
      "learning_rate": 4.985754985754986e-06,
      "loss": 0.1939,
      "step": 4260
    },
    {
      "epoch": 9.12,
      "learning_rate": 4.867046533713201e-06,
      "loss": 0.1823,
      "step": 4270
    },
    {
      "epoch": 9.15,
      "learning_rate": 4.748338081671415e-06,
      "loss": 0.2161,
      "step": 4280
    },
    {
      "epoch": 9.17,
      "learning_rate": 4.6296296296296296e-06,
      "loss": 0.205,
      "step": 4290
    },
    {
      "epoch": 9.19,
      "learning_rate": 4.510921177587845e-06,
      "loss": 0.2441,
      "step": 4300
    },
    {
      "epoch": 9.21,
      "learning_rate": 4.392212725546059e-06,
      "loss": 0.205,
      "step": 4310
    },
    {
      "epoch": 9.23,
      "learning_rate": 4.273504273504274e-06,
      "loss": 0.2395,
      "step": 4320
    },
    {
      "epoch": 9.25,
      "learning_rate": 4.154795821462489e-06,
      "loss": 0.2253,
      "step": 4330
    },
    {
      "epoch": 9.27,
      "learning_rate": 4.036087369420703e-06,
      "loss": 0.2011,
      "step": 4340
    },
    {
      "epoch": 9.29,
      "learning_rate": 3.917378917378917e-06,
      "loss": 0.2047,
      "step": 4350
    },
    {
      "epoch": 9.32,
      "learning_rate": 3.798670465337132e-06,
      "loss": 0.2147,
      "step": 4360
    },
    {
      "epoch": 9.34,
      "learning_rate": 3.679962013295347e-06,
      "loss": 0.238,
      "step": 4370
    },
    {
      "epoch": 9.36,
      "learning_rate": 3.5612535612535615e-06,
      "loss": 0.1856,
      "step": 4380
    },
    {
      "epoch": 9.38,
      "learning_rate": 3.442545109211776e-06,
      "loss": 0.2375,
      "step": 4390
    },
    {
      "epoch": 9.4,
      "learning_rate": 3.3238366571699908e-06,
      "loss": 0.2077,
      "step": 4400
    },
    {
      "epoch": 9.42,
      "learning_rate": 3.205128205128205e-06,
      "loss": 0.2377,
      "step": 4410
    },
    {
      "epoch": 9.44,
      "learning_rate": 3.0864197530864196e-06,
      "loss": 0.2098,
      "step": 4420
    },
    {
      "epoch": 9.47,
      "learning_rate": 2.9677113010446346e-06,
      "loss": 0.2114,
      "step": 4430
    },
    {
      "epoch": 9.49,
      "learning_rate": 2.8490028490028492e-06,
      "loss": 0.2236,
      "step": 4440
    },
    {
      "epoch": 9.51,
      "learning_rate": 2.730294396961064e-06,
      "loss": 0.2106,
      "step": 4450
    },
    {
      "epoch": 9.53,
      "learning_rate": 2.6115859449192785e-06,
      "loss": 0.2144,
      "step": 4460
    },
    {
      "epoch": 9.55,
      "learning_rate": 2.492877492877493e-06,
      "loss": 0.213,
      "step": 4470
    },
    {
      "epoch": 9.57,
      "learning_rate": 2.3741690408357077e-06,
      "loss": 0.2156,
      "step": 4480
    },
    {
      "epoch": 9.59,
      "learning_rate": 2.2554605887939223e-06,
      "loss": 0.1937,
      "step": 4490
    },
    {
      "epoch": 9.61,
      "learning_rate": 2.136752136752137e-06,
      "loss": 0.2047,
      "step": 4500
    },
    {
      "epoch": 9.64,
      "learning_rate": 2.0180436847103515e-06,
      "loss": 0.2212,
      "step": 4510
    },
    {
      "epoch": 9.66,
      "learning_rate": 1.899335232668566e-06,
      "loss": 0.243,
      "step": 4520
    },
    {
      "epoch": 9.68,
      "learning_rate": 1.7806267806267808e-06,
      "loss": 0.1946,
      "step": 4530
    },
    {
      "epoch": 9.7,
      "learning_rate": 1.6619183285849954e-06,
      "loss": 0.2191,
      "step": 4540
    },
    {
      "epoch": 9.72,
      "learning_rate": 1.5432098765432098e-06,
      "loss": 0.2316,
      "step": 4550
    },
    {
      "epoch": 9.74,
      "learning_rate": 1.4245014245014246e-06,
      "loss": 0.2298,
      "step": 4560
    },
    {
      "epoch": 9.76,
      "learning_rate": 1.3057929724596392e-06,
      "loss": 0.2217,
      "step": 4570
    },
    {
      "epoch": 9.79,
      "learning_rate": 1.1870845204178538e-06,
      "loss": 0.1999,
      "step": 4580
    },
    {
      "epoch": 9.81,
      "learning_rate": 1.0683760683760685e-06,
      "loss": 0.2415,
      "step": 4590
    },
    {
      "epoch": 9.83,
      "learning_rate": 9.49667616334283e-07,
      "loss": 0.2058,
      "step": 4600
    },
    {
      "epoch": 9.85,
      "learning_rate": 8.309591642924977e-07,
      "loss": 0.1963,
      "step": 4610
    },
    {
      "epoch": 9.87,
      "learning_rate": 7.122507122507123e-07,
      "loss": 0.2147,
      "step": 4620
    },
    {
      "epoch": 9.89,
      "learning_rate": 5.935422602089269e-07,
      "loss": 0.2069,
      "step": 4630
    },
    {
      "epoch": 9.91,
      "learning_rate": 4.748338081671415e-07,
      "loss": 0.2068,
      "step": 4640
    },
    {
      "epoch": 9.93,
      "learning_rate": 3.5612535612535615e-07,
      "loss": 0.2256,
      "step": 4650
    },
    {
      "epoch": 9.96,
      "learning_rate": 2.3741690408357074e-07,
      "loss": 0.1996,
      "step": 4660
    },
    {
      "epoch": 9.98,
      "learning_rate": 1.1870845204178537e-07,
      "loss": 0.2189,
      "step": 4670
    },
    {
      "epoch": 10.0,
      "learning_rate": 0.0,
      "loss": 0.204,
      "step": 4680
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.9086,
      "eval_loss": 0.2588711977005005,
      "eval_runtime": 47.7546,
      "eval_samples_per_second": 209.404,
      "eval_steps_per_second": 6.554,
      "step": 4680
    },
    {
      "epoch": 10.0,
      "step": 4680,
      "total_flos": 4.700425507906506e+19,
      "train_loss": 0.3335707022848292,
      "train_runtime": 7122.0116,
      "train_samples_per_second": 84.246,
      "train_steps_per_second": 0.657
    }
  ],
  "max_steps": 4680,
  "num_train_epochs": 10,
  "total_flos": 4.700425507906506e+19,
  "trial_name": null,
  "trial_params": null
}