{
  "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": 3.344017094017094e-05,
      "loss": 0.5116,
      "step": 10
    },
    {
      "epoch": 0.04,
      "learning_rate": 3.450854700854701e-05,
      "loss": 0.4826,
      "step": 20
    },
    {
      "epoch": 0.06,
      "learning_rate": 3.557692307692308e-05,
      "loss": 0.4609,
      "step": 30
    },
    {
      "epoch": 0.09,
      "learning_rate": 3.664529914529915e-05,
      "loss": 0.4828,
      "step": 40
    },
    {
      "epoch": 0.11,
      "learning_rate": 3.771367521367522e-05,
      "loss": 0.515,
      "step": 50
    },
    {
      "epoch": 0.13,
      "learning_rate": 3.878205128205129e-05,
      "loss": 0.4694,
      "step": 60
    },
    {
      "epoch": 0.15,
      "learning_rate": 3.985042735042736e-05,
      "loss": 0.4676,
      "step": 70
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.091880341880342e-05,
      "loss": 0.4992,
      "step": 80
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.198717948717949e-05,
      "loss": 0.4541,
      "step": 90
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.305555555555556e-05,
      "loss": 0.4417,
      "step": 100
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.412393162393162e-05,
      "loss": 0.4958,
      "step": 110
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.519230769230769e-05,
      "loss": 0.5027,
      "step": 120
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.626068376068376e-05,
      "loss": 0.4531,
      "step": 130
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.732905982905983e-05,
      "loss": 0.4671,
      "step": 140
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.83974358974359e-05,
      "loss": 0.4696,
      "step": 150
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.946581196581197e-05,
      "loss": 0.4496,
      "step": 160
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.994064577397911e-05,
      "loss": 0.4481,
      "step": 170
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.982193732193732e-05,
      "loss": 0.4559,
      "step": 180
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.970322886989554e-05,
      "loss": 0.426,
      "step": 190
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.9584520417853754e-05,
      "loss": 0.4454,
      "step": 200
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.946581196581197e-05,
      "loss": 0.4364,
      "step": 210
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.934710351377018e-05,
      "loss": 0.4569,
      "step": 220
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.92283950617284e-05,
      "loss": 0.4139,
      "step": 230
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.9109686609686614e-05,
      "loss": 0.387,
      "step": 240
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.899097815764483e-05,
      "loss": 0.4149,
      "step": 250
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.887226970560304e-05,
      "loss": 0.4132,
      "step": 260
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.875356125356126e-05,
      "loss": 0.4662,
      "step": 270
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.863485280151947e-05,
      "loss": 0.4325,
      "step": 280
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.8516144349477684e-05,
      "loss": 0.4455,
      "step": 290
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.83974358974359e-05,
      "loss": 0.453,
      "step": 300
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.827872744539411e-05,
      "loss": 0.4062,
      "step": 310
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.816001899335233e-05,
      "loss": 0.4098,
      "step": 320
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.8041310541310545e-05,
      "loss": 0.442,
      "step": 330
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.792260208926876e-05,
      "loss": 0.426,
      "step": 340
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.780389363722697e-05,
      "loss": 0.4603,
      "step": 350
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.768518518518519e-05,
      "loss": 0.3436,
      "step": 360
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.75664767331434e-05,
      "loss": 0.4108,
      "step": 370
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.7447768281101615e-05,
      "loss": 0.3932,
      "step": 380
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.732905982905983e-05,
      "loss": 0.396,
      "step": 390
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.721035137701805e-05,
      "loss": 0.4101,
      "step": 400
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.709164292497626e-05,
      "loss": 0.4116,
      "step": 410
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.6972934472934475e-05,
      "loss": 0.4133,
      "step": 420
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.685422602089269e-05,
      "loss": 0.3968,
      "step": 430
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.673551756885091e-05,
      "loss": 0.3688,
      "step": 440
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.661680911680912e-05,
      "loss": 0.424,
      "step": 450
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.6498100664767335e-05,
      "loss": 0.3952,
      "step": 460
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8584,
      "eval_loss": 0.3884398937225342,
      "eval_runtime": 42.2861,
      "eval_samples_per_second": 236.484,
      "eval_steps_per_second": 7.402,
      "step": 468
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.6379392212725545e-05,
      "loss": 0.4125,
      "step": 470
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.626068376068376e-05,
      "loss": 0.3419,
      "step": 480
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.614197530864198e-05,
      "loss": 0.3792,
      "step": 490
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.6023266856600195e-05,
      "loss": 0.3655,
      "step": 500
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.5904558404558405e-05,
      "loss": 0.3539,
      "step": 510
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.578584995251662e-05,
      "loss": 0.3653,
      "step": 520
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.566714150047484e-05,
      "loss": 0.3749,
      "step": 530
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.554843304843305e-05,
      "loss": 0.3775,
      "step": 540
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.5429724596391265e-05,
      "loss": 0.3337,
      "step": 550
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.5311016144349475e-05,
      "loss": 0.3581,
      "step": 560
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.519230769230769e-05,
      "loss": 0.3829,
      "step": 570
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.507359924026591e-05,
      "loss": 0.3598,
      "step": 580
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.4954890788224125e-05,
      "loss": 0.3287,
      "step": 590
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.4836182336182335e-05,
      "loss": 0.3547,
      "step": 600
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.471747388414055e-05,
      "loss": 0.3787,
      "step": 610
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.459876543209877e-05,
      "loss": 0.3901,
      "step": 620
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.4480056980056986e-05,
      "loss": 0.3816,
      "step": 630
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.4361348528015196e-05,
      "loss": 0.3091,
      "step": 640
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.424264007597341e-05,
      "loss": 0.306,
      "step": 650
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.412393162393162e-05,
      "loss": 0.3591,
      "step": 660
    },
    {
      "epoch": 1.43,
      "learning_rate": 4.400522317188984e-05,
      "loss": 0.3772,
      "step": 670
    },
    {
      "epoch": 1.45,
      "learning_rate": 4.3886514719848056e-05,
      "loss": 0.3808,
      "step": 680
    },
    {
      "epoch": 1.47,
      "learning_rate": 4.376780626780627e-05,
      "loss": 0.3741,
      "step": 690
    },
    {
      "epoch": 1.49,
      "learning_rate": 4.364909781576448e-05,
      "loss": 0.3368,
      "step": 700
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.35303893637227e-05,
      "loss": 0.3816,
      "step": 710
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.3411680911680916e-05,
      "loss": 0.3531,
      "step": 720
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.329297245963913e-05,
      "loss": 0.3347,
      "step": 730
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.317426400759734e-05,
      "loss": 0.3551,
      "step": 740
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.305555555555556e-05,
      "loss": 0.3297,
      "step": 750
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.293684710351377e-05,
      "loss": 0.3055,
      "step": 760
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.281813865147199e-05,
      "loss": 0.3743,
      "step": 770
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.26994301994302e-05,
      "loss": 0.3774,
      "step": 780
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.258072174738841e-05,
      "loss": 0.349,
      "step": 790
    },
    {
      "epoch": 1.71,
      "learning_rate": 4.246201329534663e-05,
      "loss": 0.3853,
      "step": 800
    },
    {
      "epoch": 1.73,
      "learning_rate": 4.2343304843304846e-05,
      "loss": 0.3304,
      "step": 810
    },
    {
      "epoch": 1.75,
      "learning_rate": 4.222459639126306e-05,
      "loss": 0.3238,
      "step": 820
    },
    {
      "epoch": 1.77,
      "learning_rate": 4.210588793922127e-05,
      "loss": 0.3073,
      "step": 830
    },
    {
      "epoch": 1.79,
      "learning_rate": 4.198717948717949e-05,
      "loss": 0.3225,
      "step": 840
    },
    {
      "epoch": 1.81,
      "learning_rate": 4.18684710351377e-05,
      "loss": 0.3293,
      "step": 850
    },
    {
      "epoch": 1.84,
      "learning_rate": 4.1749762583095916e-05,
      "loss": 0.3265,
      "step": 860
    },
    {
      "epoch": 1.86,
      "learning_rate": 4.163105413105413e-05,
      "loss": 0.3882,
      "step": 870
    },
    {
      "epoch": 1.88,
      "learning_rate": 4.151234567901235e-05,
      "loss": 0.3513,
      "step": 880
    },
    {
      "epoch": 1.9,
      "learning_rate": 4.139363722697056e-05,
      "loss": 0.3394,
      "step": 890
    },
    {
      "epoch": 1.92,
      "learning_rate": 4.1274928774928776e-05,
      "loss": 0.3552,
      "step": 900
    },
    {
      "epoch": 1.94,
      "learning_rate": 4.115622032288699e-05,
      "loss": 0.3723,
      "step": 910
    },
    {
      "epoch": 1.96,
      "learning_rate": 4.103751187084521e-05,
      "loss": 0.3605,
      "step": 920
    },
    {
      "epoch": 1.99,
      "learning_rate": 4.091880341880342e-05,
      "loss": 0.3348,
      "step": 930
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8691,
      "eval_loss": 0.36213183403015137,
      "eval_runtime": 42.3927,
      "eval_samples_per_second": 235.89,
      "eval_steps_per_second": 7.383,
      "step": 936
    },
    {
      "epoch": 2.01,
      "learning_rate": 4.0800094966761637e-05,
      "loss": 0.3809,
      "step": 940
    },
    {
      "epoch": 2.03,
      "learning_rate": 4.0681386514719847e-05,
      "loss": 0.3514,
      "step": 950
    },
    {
      "epoch": 2.05,
      "learning_rate": 4.056267806267807e-05,
      "loss": 0.3182,
      "step": 960
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.044396961063628e-05,
      "loss": 0.3321,
      "step": 970
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.03252611585945e-05,
      "loss": 0.3473,
      "step": 980
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.020655270655271e-05,
      "loss": 0.3174,
      "step": 990
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.0087844254510923e-05,
      "loss": 0.3256,
      "step": 1000
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.996913580246914e-05,
      "loss": 0.3478,
      "step": 1010
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.985042735042736e-05,
      "loss": 0.3255,
      "step": 1020
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.973171889838557e-05,
      "loss": 0.3322,
      "step": 1030
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.961301044634378e-05,
      "loss": 0.3269,
      "step": 1040
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.9494301994301994e-05,
      "loss": 0.3229,
      "step": 1050
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.937559354226021e-05,
      "loss": 0.3673,
      "step": 1060
    },
    {
      "epoch": 2.29,
      "learning_rate": 3.925688509021843e-05,
      "loss": 0.273,
      "step": 1070
    },
    {
      "epoch": 2.31,
      "learning_rate": 3.913817663817664e-05,
      "loss": 0.3712,
      "step": 1080
    },
    {
      "epoch": 2.33,
      "learning_rate": 3.9019468186134854e-05,
      "loss": 0.3051,
      "step": 1090
    },
    {
      "epoch": 2.35,
      "learning_rate": 3.890075973409307e-05,
      "loss": 0.3494,
      "step": 1100
    },
    {
      "epoch": 2.37,
      "learning_rate": 3.878205128205129e-05,
      "loss": 0.3198,
      "step": 1110
    },
    {
      "epoch": 2.39,
      "learning_rate": 3.86633428300095e-05,
      "loss": 0.3397,
      "step": 1120
    },
    {
      "epoch": 2.41,
      "learning_rate": 3.8544634377967714e-05,
      "loss": 0.2907,
      "step": 1130
    },
    {
      "epoch": 2.44,
      "learning_rate": 3.8425925925925924e-05,
      "loss": 0.3441,
      "step": 1140
    },
    {
      "epoch": 2.46,
      "learning_rate": 3.830721747388414e-05,
      "loss": 0.3356,
      "step": 1150
    },
    {
      "epoch": 2.48,
      "learning_rate": 3.818850902184236e-05,
      "loss": 0.2924,
      "step": 1160
    },
    {
      "epoch": 2.5,
      "learning_rate": 3.8069800569800574e-05,
      "loss": 0.3179,
      "step": 1170
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.7951092117758784e-05,
      "loss": 0.3183,
      "step": 1180
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.7832383665717e-05,
      "loss": 0.3186,
      "step": 1190
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.771367521367522e-05,
      "loss": 0.3192,
      "step": 1200
    },
    {
      "epoch": 2.58,
      "learning_rate": 3.7594966761633434e-05,
      "loss": 0.3431,
      "step": 1210
    },
    {
      "epoch": 2.61,
      "learning_rate": 3.7476258309591644e-05,
      "loss": 0.2914,
      "step": 1220
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.735754985754986e-05,
      "loss": 0.2794,
      "step": 1230
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.723884140550807e-05,
      "loss": 0.2934,
      "step": 1240
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.7120132953466294e-05,
      "loss": 0.3627,
      "step": 1250
    },
    {
      "epoch": 2.69,
      "learning_rate": 3.7001424501424504e-05,
      "loss": 0.3087,
      "step": 1260
    },
    {
      "epoch": 2.71,
      "learning_rate": 3.6882716049382714e-05,
      "loss": 0.3198,
      "step": 1270
    },
    {
      "epoch": 2.73,
      "learning_rate": 3.676400759734093e-05,
      "loss": 0.3219,
      "step": 1280
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.664529914529915e-05,
      "loss": 0.2951,
      "step": 1290
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.6526590693257364e-05,
      "loss": 0.3525,
      "step": 1300
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.6407882241215574e-05,
      "loss": 0.3241,
      "step": 1310
    },
    {
      "epoch": 2.82,
      "learning_rate": 3.628917378917379e-05,
      "loss": 0.3315,
      "step": 1320
    },
    {
      "epoch": 2.84,
      "learning_rate": 3.6170465337132e-05,
      "loss": 0.3155,
      "step": 1330
    },
    {
      "epoch": 2.86,
      "learning_rate": 3.605175688509022e-05,
      "loss": 0.291,
      "step": 1340
    },
    {
      "epoch": 2.88,
      "learning_rate": 3.5933048433048435e-05,
      "loss": 0.3543,
      "step": 1350
    },
    {
      "epoch": 2.9,
      "learning_rate": 3.581433998100665e-05,
      "loss": 0.3039,
      "step": 1360
    },
    {
      "epoch": 2.93,
      "learning_rate": 3.569563152896486e-05,
      "loss": 0.3036,
      "step": 1370
    },
    {
      "epoch": 2.95,
      "learning_rate": 3.557692307692308e-05,
      "loss": 0.3079,
      "step": 1380
    },
    {
      "epoch": 2.97,
      "learning_rate": 3.5458214624881295e-05,
      "loss": 0.297,
      "step": 1390
    },
    {
      "epoch": 2.99,
      "learning_rate": 3.533950617283951e-05,
      "loss": 0.2632,
      "step": 1400
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.882,
      "eval_loss": 0.33032089471817017,
      "eval_runtime": 42.3115,
      "eval_samples_per_second": 236.343,
      "eval_steps_per_second": 7.398,
      "step": 1404
    },
    {
      "epoch": 3.01,
      "learning_rate": 3.522079772079772e-05,
      "loss": 0.3317,
      "step": 1410
    },
    {
      "epoch": 3.03,
      "learning_rate": 3.510208926875594e-05,
      "loss": 0.3198,
      "step": 1420
    },
    {
      "epoch": 3.06,
      "learning_rate": 3.498338081671415e-05,
      "loss": 0.2977,
      "step": 1430
    },
    {
      "epoch": 3.08,
      "learning_rate": 3.486467236467237e-05,
      "loss": 0.3448,
      "step": 1440
    },
    {
      "epoch": 3.1,
      "learning_rate": 3.474596391263058e-05,
      "loss": 0.2444,
      "step": 1450
    },
    {
      "epoch": 3.12,
      "learning_rate": 3.46272554605888e-05,
      "loss": 0.289,
      "step": 1460
    },
    {
      "epoch": 3.14,
      "learning_rate": 3.450854700854701e-05,
      "loss": 0.3191,
      "step": 1470
    },
    {
      "epoch": 3.16,
      "learning_rate": 3.4389838556505225e-05,
      "loss": 0.314,
      "step": 1480
    },
    {
      "epoch": 3.18,
      "learning_rate": 3.427113010446344e-05,
      "loss": 0.3063,
      "step": 1490
    },
    {
      "epoch": 3.2,
      "learning_rate": 3.415242165242166e-05,
      "loss": 0.3068,
      "step": 1500
    },
    {
      "epoch": 3.23,
      "learning_rate": 3.403371320037987e-05,
      "loss": 0.307,
      "step": 1510
    },
    {
      "epoch": 3.25,
      "learning_rate": 3.391500474833808e-05,
      "loss": 0.293,
      "step": 1520
    },
    {
      "epoch": 3.27,
      "learning_rate": 3.3796296296296295e-05,
      "loss": 0.3224,
      "step": 1530
    },
    {
      "epoch": 3.29,
      "learning_rate": 3.367758784425451e-05,
      "loss": 0.2709,
      "step": 1540
    },
    {
      "epoch": 3.31,
      "learning_rate": 3.355887939221273e-05,
      "loss": 0.266,
      "step": 1550
    },
    {
      "epoch": 3.33,
      "learning_rate": 3.344017094017094e-05,
      "loss": 0.2894,
      "step": 1560
    },
    {
      "epoch": 3.35,
      "learning_rate": 3.3321462488129155e-05,
      "loss": 0.2826,
      "step": 1570
    },
    {
      "epoch": 3.38,
      "learning_rate": 3.320275403608737e-05,
      "loss": 0.3215,
      "step": 1580
    },
    {
      "epoch": 3.4,
      "learning_rate": 3.308404558404559e-05,
      "loss": 0.2729,
      "step": 1590
    },
    {
      "epoch": 3.42,
      "learning_rate": 3.29653371320038e-05,
      "loss": 0.2911,
      "step": 1600
    },
    {
      "epoch": 3.44,
      "learning_rate": 3.2846628679962015e-05,
      "loss": 0.3218,
      "step": 1610
    },
    {
      "epoch": 3.46,
      "learning_rate": 3.2727920227920225e-05,
      "loss": 0.2757,
      "step": 1620
    },
    {
      "epoch": 3.48,
      "learning_rate": 3.260921177587844e-05,
      "loss": 0.3144,
      "step": 1630
    },
    {
      "epoch": 3.5,
      "learning_rate": 3.249050332383666e-05,
      "loss": 0.2541,
      "step": 1640
    },
    {
      "epoch": 3.52,
      "learning_rate": 3.2371794871794876e-05,
      "loss": 0.2588,
      "step": 1650
    },
    {
      "epoch": 3.55,
      "learning_rate": 3.2253086419753086e-05,
      "loss": 0.2614,
      "step": 1660
    },
    {
      "epoch": 3.57,
      "learning_rate": 3.21343779677113e-05,
      "loss": 0.3001,
      "step": 1670
    },
    {
      "epoch": 3.59,
      "learning_rate": 3.201566951566952e-05,
      "loss": 0.2864,
      "step": 1680
    },
    {
      "epoch": 3.61,
      "learning_rate": 3.1896961063627736e-05,
      "loss": 0.3021,
      "step": 1690
    },
    {
      "epoch": 3.63,
      "learning_rate": 3.1778252611585946e-05,
      "loss": 0.2616,
      "step": 1700
    },
    {
      "epoch": 3.65,
      "learning_rate": 3.165954415954416e-05,
      "loss": 0.249,
      "step": 1710
    },
    {
      "epoch": 3.67,
      "learning_rate": 3.154083570750237e-05,
      "loss": 0.3208,
      "step": 1720
    },
    {
      "epoch": 3.7,
      "learning_rate": 3.1422127255460596e-05,
      "loss": 0.263,
      "step": 1730
    },
    {
      "epoch": 3.72,
      "learning_rate": 3.1303418803418806e-05,
      "loss": 0.2769,
      "step": 1740
    },
    {
      "epoch": 3.74,
      "learning_rate": 3.118471035137702e-05,
      "loss": 0.3022,
      "step": 1750
    },
    {
      "epoch": 3.76,
      "learning_rate": 3.106600189933523e-05,
      "loss": 0.2739,
      "step": 1760
    },
    {
      "epoch": 3.78,
      "learning_rate": 3.094729344729345e-05,
      "loss": 0.311,
      "step": 1770
    },
    {
      "epoch": 3.8,
      "learning_rate": 3.0828584995251666e-05,
      "loss": 0.2914,
      "step": 1780
    },
    {
      "epoch": 3.82,
      "learning_rate": 3.0709876543209876e-05,
      "loss": 0.2774,
      "step": 1790
    },
    {
      "epoch": 3.84,
      "learning_rate": 3.059116809116809e-05,
      "loss": 0.3084,
      "step": 1800
    },
    {
      "epoch": 3.87,
      "learning_rate": 3.0472459639126306e-05,
      "loss": 0.3092,
      "step": 1810
    },
    {
      "epoch": 3.89,
      "learning_rate": 3.0353751187084523e-05,
      "loss": 0.3016,
      "step": 1820
    },
    {
      "epoch": 3.91,
      "learning_rate": 3.0235042735042736e-05,
      "loss": 0.2995,
      "step": 1830
    },
    {
      "epoch": 3.93,
      "learning_rate": 3.0116334283000953e-05,
      "loss": 0.2761,
      "step": 1840
    },
    {
      "epoch": 3.95,
      "learning_rate": 2.9997625830959163e-05,
      "loss": 0.2694,
      "step": 1850
    },
    {
      "epoch": 3.97,
      "learning_rate": 2.9878917378917383e-05,
      "loss": 0.3259,
      "step": 1860
    },
    {
      "epoch": 3.99,
      "learning_rate": 2.9760208926875593e-05,
      "loss": 0.3201,
      "step": 1870
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8946,
      "eval_loss": 0.29688265919685364,
      "eval_runtime": 42.699,
      "eval_samples_per_second": 234.197,
      "eval_steps_per_second": 7.33,
      "step": 1872
    },
    {
      "epoch": 4.02,
      "learning_rate": 2.964150047483381e-05,
      "loss": 0.3168,
      "step": 1880
    },
    {
      "epoch": 4.04,
      "learning_rate": 2.9522792022792023e-05,
      "loss": 0.2752,
      "step": 1890
    },
    {
      "epoch": 4.06,
      "learning_rate": 2.940408357075024e-05,
      "loss": 0.2339,
      "step": 1900
    },
    {
      "epoch": 4.08,
      "learning_rate": 2.9285375118708453e-05,
      "loss": 0.2997,
      "step": 1910
    },
    {
      "epoch": 4.1,
      "learning_rate": 2.916666666666667e-05,
      "loss": 0.2515,
      "step": 1920
    },
    {
      "epoch": 4.12,
      "learning_rate": 2.9047958214624883e-05,
      "loss": 0.2478,
      "step": 1930
    },
    {
      "epoch": 4.15,
      "learning_rate": 2.89292497625831e-05,
      "loss": 0.3037,
      "step": 1940
    },
    {
      "epoch": 4.17,
      "learning_rate": 2.881054131054131e-05,
      "loss": 0.3028,
      "step": 1950
    },
    {
      "epoch": 4.19,
      "learning_rate": 2.869183285849953e-05,
      "loss": 0.2785,
      "step": 1960
    },
    {
      "epoch": 4.21,
      "learning_rate": 2.857312440645774e-05,
      "loss": 0.2959,
      "step": 1970
    },
    {
      "epoch": 4.23,
      "learning_rate": 2.845441595441596e-05,
      "loss": 0.2934,
      "step": 1980
    },
    {
      "epoch": 4.25,
      "learning_rate": 2.833570750237417e-05,
      "loss": 0.2764,
      "step": 1990
    },
    {
      "epoch": 4.27,
      "learning_rate": 2.8216999050332387e-05,
      "loss": 0.2965,
      "step": 2000
    },
    {
      "epoch": 4.29,
      "learning_rate": 2.80982905982906e-05,
      "loss": 0.2902,
      "step": 2010
    },
    {
      "epoch": 4.32,
      "learning_rate": 2.7979582146248813e-05,
      "loss": 0.2761,
      "step": 2020
    },
    {
      "epoch": 4.34,
      "learning_rate": 2.786087369420703e-05,
      "loss": 0.2683,
      "step": 2030
    },
    {
      "epoch": 4.36,
      "learning_rate": 2.774216524216524e-05,
      "loss": 0.2646,
      "step": 2040
    },
    {
      "epoch": 4.38,
      "learning_rate": 2.762345679012346e-05,
      "loss": 0.2721,
      "step": 2050
    },
    {
      "epoch": 4.4,
      "learning_rate": 2.750474833808167e-05,
      "loss": 0.2621,
      "step": 2060
    },
    {
      "epoch": 4.42,
      "learning_rate": 2.7386039886039887e-05,
      "loss": 0.2512,
      "step": 2070
    },
    {
      "epoch": 4.44,
      "learning_rate": 2.72673314339981e-05,
      "loss": 0.2847,
      "step": 2080
    },
    {
      "epoch": 4.47,
      "learning_rate": 2.7148622981956317e-05,
      "loss": 0.271,
      "step": 2090
    },
    {
      "epoch": 4.49,
      "learning_rate": 2.702991452991453e-05,
      "loss": 0.2589,
      "step": 2100
    },
    {
      "epoch": 4.51,
      "learning_rate": 2.6911206077872747e-05,
      "loss": 0.2466,
      "step": 2110
    },
    {
      "epoch": 4.53,
      "learning_rate": 2.679249762583096e-05,
      "loss": 0.2518,
      "step": 2120
    },
    {
      "epoch": 4.55,
      "learning_rate": 2.6673789173789177e-05,
      "loss": 0.3062,
      "step": 2130
    },
    {
      "epoch": 4.57,
      "learning_rate": 2.6555080721747387e-05,
      "loss": 0.2562,
      "step": 2140
    },
    {
      "epoch": 4.59,
      "learning_rate": 2.6436372269705607e-05,
      "loss": 0.2557,
      "step": 2150
    },
    {
      "epoch": 4.61,
      "learning_rate": 2.6317663817663817e-05,
      "loss": 0.2822,
      "step": 2160
    },
    {
      "epoch": 4.64,
      "learning_rate": 2.6198955365622034e-05,
      "loss": 0.2695,
      "step": 2170
    },
    {
      "epoch": 4.66,
      "learning_rate": 2.6080246913580247e-05,
      "loss": 0.2642,
      "step": 2180
    },
    {
      "epoch": 4.68,
      "learning_rate": 2.5961538461538464e-05,
      "loss": 0.2701,
      "step": 2190
    },
    {
      "epoch": 4.7,
      "learning_rate": 2.5842830009496677e-05,
      "loss": 0.2606,
      "step": 2200
    },
    {
      "epoch": 4.72,
      "learning_rate": 2.5724121557454894e-05,
      "loss": 0.2426,
      "step": 2210
    },
    {
      "epoch": 4.74,
      "learning_rate": 2.5605413105413107e-05,
      "loss": 0.2631,
      "step": 2220
    },
    {
      "epoch": 4.76,
      "learning_rate": 2.5486704653371324e-05,
      "loss": 0.2542,
      "step": 2230
    },
    {
      "epoch": 4.79,
      "learning_rate": 2.5367996201329537e-05,
      "loss": 0.2673,
      "step": 2240
    },
    {
      "epoch": 4.81,
      "learning_rate": 2.5249287749287747e-05,
      "loss": 0.2668,
      "step": 2250
    },
    {
      "epoch": 4.83,
      "learning_rate": 2.5130579297245964e-05,
      "loss": 0.2886,
      "step": 2260
    },
    {
      "epoch": 4.85,
      "learning_rate": 2.5011870845204177e-05,
      "loss": 0.263,
      "step": 2270
    },
    {
      "epoch": 4.87,
      "learning_rate": 2.4893162393162394e-05,
      "loss": 0.2513,
      "step": 2280
    },
    {
      "epoch": 4.89,
      "learning_rate": 2.477445394112061e-05,
      "loss": 0.2672,
      "step": 2290
    },
    {
      "epoch": 4.91,
      "learning_rate": 2.4655745489078824e-05,
      "loss": 0.2367,
      "step": 2300
    },
    {
      "epoch": 4.93,
      "learning_rate": 2.4537037037037038e-05,
      "loss": 0.2684,
      "step": 2310
    },
    {
      "epoch": 4.96,
      "learning_rate": 2.4418328584995254e-05,
      "loss": 0.2861,
      "step": 2320
    },
    {
      "epoch": 4.98,
      "learning_rate": 2.4299620132953468e-05,
      "loss": 0.2599,
      "step": 2330
    },
    {
      "epoch": 5.0,
      "learning_rate": 2.4180911680911684e-05,
      "loss": 0.2428,
      "step": 2340
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.8933,
      "eval_loss": 0.2944689691066742,
      "eval_runtime": 42.2523,
      "eval_samples_per_second": 236.673,
      "eval_steps_per_second": 7.408,
      "step": 2340
    },
    {
      "epoch": 5.02,
      "learning_rate": 2.4062203228869898e-05,
      "loss": 0.2381,
      "step": 2350
    },
    {
      "epoch": 5.04,
      "learning_rate": 2.394349477682811e-05,
      "loss": 0.2344,
      "step": 2360
    },
    {
      "epoch": 5.06,
      "learning_rate": 2.3824786324786324e-05,
      "loss": 0.2607,
      "step": 2370
    },
    {
      "epoch": 5.09,
      "learning_rate": 2.3706077872744538e-05,
      "loss": 0.2201,
      "step": 2380
    },
    {
      "epoch": 5.11,
      "learning_rate": 2.3587369420702755e-05,
      "loss": 0.2867,
      "step": 2390
    },
    {
      "epoch": 5.13,
      "learning_rate": 2.3468660968660968e-05,
      "loss": 0.2694,
      "step": 2400
    },
    {
      "epoch": 5.15,
      "learning_rate": 2.3349952516619185e-05,
      "loss": 0.2258,
      "step": 2410
    },
    {
      "epoch": 5.17,
      "learning_rate": 2.3231244064577398e-05,
      "loss": 0.2523,
      "step": 2420
    },
    {
      "epoch": 5.19,
      "learning_rate": 2.311253561253561e-05,
      "loss": 0.2311,
      "step": 2430
    },
    {
      "epoch": 5.21,
      "learning_rate": 2.2993827160493828e-05,
      "loss": 0.2695,
      "step": 2440
    },
    {
      "epoch": 5.23,
      "learning_rate": 2.287511870845204e-05,
      "loss": 0.2417,
      "step": 2450
    },
    {
      "epoch": 5.26,
      "learning_rate": 2.2756410256410258e-05,
      "loss": 0.2969,
      "step": 2460
    },
    {
      "epoch": 5.28,
      "learning_rate": 2.263770180436847e-05,
      "loss": 0.2712,
      "step": 2470
    },
    {
      "epoch": 5.3,
      "learning_rate": 2.2518993352326688e-05,
      "loss": 0.256,
      "step": 2480
    },
    {
      "epoch": 5.32,
      "learning_rate": 2.24002849002849e-05,
      "loss": 0.2402,
      "step": 2490
    },
    {
      "epoch": 5.34,
      "learning_rate": 2.2281576448243115e-05,
      "loss": 0.2604,
      "step": 2500
    },
    {
      "epoch": 5.36,
      "learning_rate": 2.216286799620133e-05,
      "loss": 0.247,
      "step": 2510
    },
    {
      "epoch": 5.38,
      "learning_rate": 2.2044159544159545e-05,
      "loss": 0.274,
      "step": 2520
    },
    {
      "epoch": 5.41,
      "learning_rate": 2.1925451092117762e-05,
      "loss": 0.2573,
      "step": 2530
    },
    {
      "epoch": 5.43,
      "learning_rate": 2.1806742640075975e-05,
      "loss": 0.2785,
      "step": 2540
    },
    {
      "epoch": 5.45,
      "learning_rate": 2.168803418803419e-05,
      "loss": 0.2378,
      "step": 2550
    },
    {
      "epoch": 5.47,
      "learning_rate": 2.1569325735992405e-05,
      "loss": 0.2102,
      "step": 2560
    },
    {
      "epoch": 5.49,
      "learning_rate": 2.145061728395062e-05,
      "loss": 0.2627,
      "step": 2570
    },
    {
      "epoch": 5.51,
      "learning_rate": 2.1331908831908835e-05,
      "loss": 0.2672,
      "step": 2580
    },
    {
      "epoch": 5.53,
      "learning_rate": 2.121320037986705e-05,
      "loss": 0.2716,
      "step": 2590
    },
    {
      "epoch": 5.55,
      "learning_rate": 2.1094491927825262e-05,
      "loss": 0.244,
      "step": 2600
    },
    {
      "epoch": 5.58,
      "learning_rate": 2.097578347578348e-05,
      "loss": 0.2784,
      "step": 2610
    },
    {
      "epoch": 5.6,
      "learning_rate": 2.085707502374169e-05,
      "loss": 0.256,
      "step": 2620
    },
    {
      "epoch": 5.62,
      "learning_rate": 2.0738366571699905e-05,
      "loss": 0.261,
      "step": 2630
    },
    {
      "epoch": 5.64,
      "learning_rate": 2.061965811965812e-05,
      "loss": 0.2395,
      "step": 2640
    },
    {
      "epoch": 5.66,
      "learning_rate": 2.0500949667616335e-05,
      "loss": 0.2911,
      "step": 2650
    },
    {
      "epoch": 5.68,
      "learning_rate": 2.038224121557455e-05,
      "loss": 0.2633,
      "step": 2660
    },
    {
      "epoch": 5.7,
      "learning_rate": 2.0263532763532762e-05,
      "loss": 0.2364,
      "step": 2670
    },
    {
      "epoch": 5.73,
      "learning_rate": 2.014482431149098e-05,
      "loss": 0.273,
      "step": 2680
    },
    {
      "epoch": 5.75,
      "learning_rate": 2.0026115859449192e-05,
      "loss": 0.2546,
      "step": 2690
    },
    {
      "epoch": 5.77,
      "learning_rate": 1.990740740740741e-05,
      "loss": 0.25,
      "step": 2700
    },
    {
      "epoch": 5.79,
      "learning_rate": 1.9788698955365622e-05,
      "loss": 0.2913,
      "step": 2710
    },
    {
      "epoch": 5.81,
      "learning_rate": 1.966999050332384e-05,
      "loss": 0.2486,
      "step": 2720
    },
    {
      "epoch": 5.83,
      "learning_rate": 1.9551282051282052e-05,
      "loss": 0.2468,
      "step": 2730
    },
    {
      "epoch": 5.85,
      "learning_rate": 1.9432573599240266e-05,
      "loss": 0.2427,
      "step": 2740
    },
    {
      "epoch": 5.87,
      "learning_rate": 1.9313865147198482e-05,
      "loss": 0.2451,
      "step": 2750
    },
    {
      "epoch": 5.9,
      "learning_rate": 1.9195156695156696e-05,
      "loss": 0.2488,
      "step": 2760
    },
    {
      "epoch": 5.92,
      "learning_rate": 1.9076448243114913e-05,
      "loss": 0.2391,
      "step": 2770
    },
    {
      "epoch": 5.94,
      "learning_rate": 1.8957739791073126e-05,
      "loss": 0.2391,
      "step": 2780
    },
    {
      "epoch": 5.96,
      "learning_rate": 1.883903133903134e-05,
      "loss": 0.255,
      "step": 2790
    },
    {
      "epoch": 5.98,
      "learning_rate": 1.8720322886989556e-05,
      "loss": 0.2008,
      "step": 2800
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.8953,
      "eval_loss": 0.2888832092285156,
      "eval_runtime": 41.9713,
      "eval_samples_per_second": 238.258,
      "eval_steps_per_second": 7.457,
      "step": 2808
    },
    {
      "epoch": 6.0,
      "learning_rate": 1.860161443494777e-05,
      "loss": 0.2604,
      "step": 2810
    },
    {
      "epoch": 6.03,
      "learning_rate": 1.8482905982905986e-05,
      "loss": 0.2691,
      "step": 2820
    },
    {
      "epoch": 6.05,
      "learning_rate": 1.83641975308642e-05,
      "loss": 0.2085,
      "step": 2830
    },
    {
      "epoch": 6.07,
      "learning_rate": 1.8245489078822413e-05,
      "loss": 0.2429,
      "step": 2840
    },
    {
      "epoch": 6.09,
      "learning_rate": 1.812678062678063e-05,
      "loss": 0.249,
      "step": 2850
    },
    {
      "epoch": 6.11,
      "learning_rate": 1.800807217473884e-05,
      "loss": 0.2608,
      "step": 2860
    },
    {
      "epoch": 6.13,
      "learning_rate": 1.7889363722697056e-05,
      "loss": 0.2088,
      "step": 2870
    },
    {
      "epoch": 6.15,
      "learning_rate": 1.777065527065527e-05,
      "loss": 0.2382,
      "step": 2880
    },
    {
      "epoch": 6.17,
      "learning_rate": 1.7651946818613486e-05,
      "loss": 0.2389,
      "step": 2890
    },
    {
      "epoch": 6.2,
      "learning_rate": 1.75332383665717e-05,
      "loss": 0.2403,
      "step": 2900
    },
    {
      "epoch": 6.22,
      "learning_rate": 1.7414529914529913e-05,
      "loss": 0.2207,
      "step": 2910
    },
    {
      "epoch": 6.24,
      "learning_rate": 1.729582146248813e-05,
      "loss": 0.2641,
      "step": 2920
    },
    {
      "epoch": 6.26,
      "learning_rate": 1.7177113010446343e-05,
      "loss": 0.2301,
      "step": 2930
    },
    {
      "epoch": 6.28,
      "learning_rate": 1.705840455840456e-05,
      "loss": 0.2135,
      "step": 2940
    },
    {
      "epoch": 6.3,
      "learning_rate": 1.6939696106362773e-05,
      "loss": 0.2033,
      "step": 2950
    },
    {
      "epoch": 6.32,
      "learning_rate": 1.682098765432099e-05,
      "loss": 0.2382,
      "step": 2960
    },
    {
      "epoch": 6.35,
      "learning_rate": 1.6702279202279203e-05,
      "loss": 0.2336,
      "step": 2970
    },
    {
      "epoch": 6.37,
      "learning_rate": 1.6583570750237416e-05,
      "loss": 0.2248,
      "step": 2980
    },
    {
      "epoch": 6.39,
      "learning_rate": 1.6464862298195633e-05,
      "loss": 0.2411,
      "step": 2990
    },
    {
      "epoch": 6.41,
      "learning_rate": 1.6346153846153847e-05,
      "loss": 0.2421,
      "step": 3000
    },
    {
      "epoch": 6.43,
      "learning_rate": 1.6227445394112063e-05,
      "loss": 0.2169,
      "step": 3010
    },
    {
      "epoch": 6.45,
      "learning_rate": 1.6108736942070277e-05,
      "loss": 0.2248,
      "step": 3020
    },
    {
      "epoch": 6.47,
      "learning_rate": 1.599002849002849e-05,
      "loss": 0.2065,
      "step": 3030
    },
    {
      "epoch": 6.49,
      "learning_rate": 1.5871320037986707e-05,
      "loss": 0.242,
      "step": 3040
    },
    {
      "epoch": 6.52,
      "learning_rate": 1.575261158594492e-05,
      "loss": 0.281,
      "step": 3050
    },
    {
      "epoch": 6.54,
      "learning_rate": 1.5633903133903137e-05,
      "loss": 0.2757,
      "step": 3060
    },
    {
      "epoch": 6.56,
      "learning_rate": 1.551519468186135e-05,
      "loss": 0.2273,
      "step": 3070
    },
    {
      "epoch": 6.58,
      "learning_rate": 1.5396486229819563e-05,
      "loss": 0.2139,
      "step": 3080
    },
    {
      "epoch": 6.6,
      "learning_rate": 1.527777777777778e-05,
      "loss": 0.2429,
      "step": 3090
    },
    {
      "epoch": 6.62,
      "learning_rate": 1.5159069325735995e-05,
      "loss": 0.2349,
      "step": 3100
    },
    {
      "epoch": 6.64,
      "learning_rate": 1.5040360873694207e-05,
      "loss": 0.2232,
      "step": 3110
    },
    {
      "epoch": 6.67,
      "learning_rate": 1.4921652421652422e-05,
      "loss": 0.2326,
      "step": 3120
    },
    {
      "epoch": 6.69,
      "learning_rate": 1.4802943969610635e-05,
      "loss": 0.2384,
      "step": 3130
    },
    {
      "epoch": 6.71,
      "learning_rate": 1.468423551756885e-05,
      "loss": 0.238,
      "step": 3140
    },
    {
      "epoch": 6.73,
      "learning_rate": 1.4565527065527065e-05,
      "loss": 0.2324,
      "step": 3150
    },
    {
      "epoch": 6.75,
      "learning_rate": 1.444681861348528e-05,
      "loss": 0.2347,
      "step": 3160
    },
    {
      "epoch": 6.77,
      "learning_rate": 1.4328110161443495e-05,
      "loss": 0.2259,
      "step": 3170
    },
    {
      "epoch": 6.79,
      "learning_rate": 1.4209401709401709e-05,
      "loss": 0.2186,
      "step": 3180
    },
    {
      "epoch": 6.81,
      "learning_rate": 1.4090693257359924e-05,
      "loss": 0.2062,
      "step": 3190
    },
    {
      "epoch": 6.84,
      "learning_rate": 1.3971984805318139e-05,
      "loss": 0.2366,
      "step": 3200
    },
    {
      "epoch": 6.86,
      "learning_rate": 1.3853276353276354e-05,
      "loss": 0.2137,
      "step": 3210
    },
    {
      "epoch": 6.88,
      "learning_rate": 1.3734567901234569e-05,
      "loss": 0.248,
      "step": 3220
    },
    {
      "epoch": 6.9,
      "learning_rate": 1.3615859449192784e-05,
      "loss": 0.2154,
      "step": 3230
    },
    {
      "epoch": 6.92,
      "learning_rate": 1.3497150997150997e-05,
      "loss": 0.2273,
      "step": 3240
    },
    {
      "epoch": 6.94,
      "learning_rate": 1.3378442545109212e-05,
      "loss": 0.2286,
      "step": 3250
    },
    {
      "epoch": 6.96,
      "learning_rate": 1.3259734093067427e-05,
      "loss": 0.2281,
      "step": 3260
    },
    {
      "epoch": 6.99,
      "learning_rate": 1.3141025641025642e-05,
      "loss": 0.2439,
      "step": 3270
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.902,
      "eval_loss": 0.2731182277202606,
      "eval_runtime": 41.8304,
      "eval_samples_per_second": 239.06,
      "eval_steps_per_second": 7.483,
      "step": 3276
    },
    {
      "epoch": 7.01,
      "learning_rate": 1.3022317188983857e-05,
      "loss": 0.2641,
      "step": 3280
    },
    {
      "epoch": 7.03,
      "learning_rate": 1.290360873694207e-05,
      "loss": 0.2683,
      "step": 3290
    },
    {
      "epoch": 7.05,
      "learning_rate": 1.2784900284900286e-05,
      "loss": 0.2466,
      "step": 3300
    },
    {
      "epoch": 7.07,
      "learning_rate": 1.2666191832858501e-05,
      "loss": 0.2384,
      "step": 3310
    },
    {
      "epoch": 7.09,
      "learning_rate": 1.2547483380816716e-05,
      "loss": 0.2094,
      "step": 3320
    },
    {
      "epoch": 7.12,
      "learning_rate": 1.242877492877493e-05,
      "loss": 0.2172,
      "step": 3330
    },
    {
      "epoch": 7.14,
      "learning_rate": 1.2310066476733144e-05,
      "loss": 0.2357,
      "step": 3340
    },
    {
      "epoch": 7.16,
      "learning_rate": 1.219135802469136e-05,
      "loss": 0.2062,
      "step": 3350
    },
    {
      "epoch": 7.18,
      "learning_rate": 1.2072649572649573e-05,
      "loss": 0.2054,
      "step": 3360
    },
    {
      "epoch": 7.2,
      "learning_rate": 1.1953941120607788e-05,
      "loss": 0.2283,
      "step": 3370
    },
    {
      "epoch": 7.22,
      "learning_rate": 1.1835232668566003e-05,
      "loss": 0.1973,
      "step": 3380
    },
    {
      "epoch": 7.24,
      "learning_rate": 1.1716524216524218e-05,
      "loss": 0.2173,
      "step": 3390
    },
    {
      "epoch": 7.26,
      "learning_rate": 1.1597815764482433e-05,
      "loss": 0.2233,
      "step": 3400
    },
    {
      "epoch": 7.29,
      "learning_rate": 1.1479107312440646e-05,
      "loss": 0.2262,
      "step": 3410
    },
    {
      "epoch": 7.31,
      "learning_rate": 1.136039886039886e-05,
      "loss": 0.235,
      "step": 3420
    },
    {
      "epoch": 7.33,
      "learning_rate": 1.1241690408357075e-05,
      "loss": 0.2152,
      "step": 3430
    },
    {
      "epoch": 7.35,
      "learning_rate": 1.112298195631529e-05,
      "loss": 0.2014,
      "step": 3440
    },
    {
      "epoch": 7.37,
      "learning_rate": 1.1004273504273505e-05,
      "loss": 0.2182,
      "step": 3450
    },
    {
      "epoch": 7.39,
      "learning_rate": 1.088556505223172e-05,
      "loss": 0.2176,
      "step": 3460
    },
    {
      "epoch": 7.41,
      "learning_rate": 1.0766856600189935e-05,
      "loss": 0.2076,
      "step": 3470
    },
    {
      "epoch": 7.44,
      "learning_rate": 1.0648148148148148e-05,
      "loss": 0.2182,
      "step": 3480
    },
    {
      "epoch": 7.46,
      "learning_rate": 1.0529439696106363e-05,
      "loss": 0.2279,
      "step": 3490
    },
    {
      "epoch": 7.48,
      "learning_rate": 1.0410731244064578e-05,
      "loss": 0.2278,
      "step": 3500
    },
    {
      "epoch": 7.5,
      "learning_rate": 1.0292022792022793e-05,
      "loss": 0.2191,
      "step": 3510
    },
    {
      "epoch": 7.52,
      "learning_rate": 1.0173314339981008e-05,
      "loss": 0.2353,
      "step": 3520
    },
    {
      "epoch": 7.54,
      "learning_rate": 1.0054605887939222e-05,
      "loss": 0.243,
      "step": 3530
    },
    {
      "epoch": 7.56,
      "learning_rate": 9.935897435897435e-06,
      "loss": 0.198,
      "step": 3540
    },
    {
      "epoch": 7.58,
      "learning_rate": 9.81718898385565e-06,
      "loss": 0.1768,
      "step": 3550
    },
    {
      "epoch": 7.61,
      "learning_rate": 9.698480531813865e-06,
      "loss": 0.2213,
      "step": 3560
    },
    {
      "epoch": 7.63,
      "learning_rate": 9.57977207977208e-06,
      "loss": 0.2309,
      "step": 3570
    },
    {
      "epoch": 7.65,
      "learning_rate": 9.461063627730295e-06,
      "loss": 0.2018,
      "step": 3580
    },
    {
      "epoch": 7.67,
      "learning_rate": 9.34235517568851e-06,
      "loss": 0.2154,
      "step": 3590
    },
    {
      "epoch": 7.69,
      "learning_rate": 9.223646723646723e-06,
      "loss": 0.1891,
      "step": 3600
    },
    {
      "epoch": 7.71,
      "learning_rate": 9.104938271604939e-06,
      "loss": 0.2372,
      "step": 3610
    },
    {
      "epoch": 7.73,
      "learning_rate": 8.986229819563154e-06,
      "loss": 0.2471,
      "step": 3620
    },
    {
      "epoch": 7.76,
      "learning_rate": 8.867521367521369e-06,
      "loss": 0.2373,
      "step": 3630
    },
    {
      "epoch": 7.78,
      "learning_rate": 8.748812915479584e-06,
      "loss": 0.2021,
      "step": 3640
    },
    {
      "epoch": 7.8,
      "learning_rate": 8.630104463437797e-06,
      "loss": 0.2023,
      "step": 3650
    },
    {
      "epoch": 7.82,
      "learning_rate": 8.51139601139601e-06,
      "loss": 0.1933,
      "step": 3660
    },
    {
      "epoch": 7.84,
      "learning_rate": 8.392687559354225e-06,
      "loss": 0.2028,
      "step": 3670
    },
    {
      "epoch": 7.86,
      "learning_rate": 8.27397910731244e-06,
      "loss": 0.2253,
      "step": 3680
    },
    {
      "epoch": 7.88,
      "learning_rate": 8.155270655270655e-06,
      "loss": 0.2568,
      "step": 3690
    },
    {
      "epoch": 7.9,
      "learning_rate": 8.03656220322887e-06,
      "loss": 0.2159,
      "step": 3700
    },
    {
      "epoch": 7.93,
      "learning_rate": 7.917853751187086e-06,
      "loss": 0.2225,
      "step": 3710
    },
    {
      "epoch": 7.95,
      "learning_rate": 7.799145299145299e-06,
      "loss": 0.2232,
      "step": 3720
    },
    {
      "epoch": 7.97,
      "learning_rate": 7.680436847103514e-06,
      "loss": 0.1996,
      "step": 3730
    },
    {
      "epoch": 7.99,
      "learning_rate": 7.561728395061729e-06,
      "loss": 0.2099,
      "step": 3740
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.9047,
      "eval_loss": 0.26033690571784973,
      "eval_runtime": 42.0488,
      "eval_samples_per_second": 237.819,
      "eval_steps_per_second": 7.444,
      "step": 3744
    },
    {
      "epoch": 8.01,
      "learning_rate": 7.443019943019944e-06,
      "loss": 0.2122,
      "step": 3750
    },
    {
      "epoch": 8.03,
      "learning_rate": 7.324311490978158e-06,
      "loss": 0.2154,
      "step": 3760
    },
    {
      "epoch": 8.06,
      "learning_rate": 7.205603038936373e-06,
      "loss": 0.1813,
      "step": 3770
    },
    {
      "epoch": 8.08,
      "learning_rate": 7.0868945868945866e-06,
      "loss": 0.2202,
      "step": 3780
    },
    {
      "epoch": 8.1,
      "learning_rate": 6.968186134852802e-06,
      "loss": 0.2215,
      "step": 3790
    },
    {
      "epoch": 8.12,
      "learning_rate": 6.849477682811016e-06,
      "loss": 0.2242,
      "step": 3800
    },
    {
      "epoch": 8.14,
      "learning_rate": 6.730769230769231e-06,
      "loss": 0.1883,
      "step": 3810
    },
    {
      "epoch": 8.16,
      "learning_rate": 6.612060778727446e-06,
      "loss": 0.2232,
      "step": 3820
    },
    {
      "epoch": 8.18,
      "learning_rate": 6.49335232668566e-06,
      "loss": 0.2138,
      "step": 3830
    },
    {
      "epoch": 8.2,
      "learning_rate": 6.374643874643875e-06,
      "loss": 0.2093,
      "step": 3840
    },
    {
      "epoch": 8.23,
      "learning_rate": 6.255935422602089e-06,
      "loss": 0.2264,
      "step": 3850
    },
    {
      "epoch": 8.25,
      "learning_rate": 6.137226970560304e-06,
      "loss": 0.2072,
      "step": 3860
    },
    {
      "epoch": 8.27,
      "learning_rate": 6.0185185185185185e-06,
      "loss": 0.2372,
      "step": 3870
    },
    {
      "epoch": 8.29,
      "learning_rate": 5.8998100664767336e-06,
      "loss": 0.1983,
      "step": 3880
    },
    {
      "epoch": 8.31,
      "learning_rate": 5.781101614434948e-06,
      "loss": 0.1887,
      "step": 3890
    },
    {
      "epoch": 8.33,
      "learning_rate": 5.662393162393163e-06,
      "loss": 0.2393,
      "step": 3900
    },
    {
      "epoch": 8.35,
      "learning_rate": 5.543684710351377e-06,
      "loss": 0.2134,
      "step": 3910
    },
    {
      "epoch": 8.38,
      "learning_rate": 5.424976258309592e-06,
      "loss": 0.1844,
      "step": 3920
    },
    {
      "epoch": 8.4,
      "learning_rate": 5.306267806267807e-06,
      "loss": 0.1963,
      "step": 3930
    },
    {
      "epoch": 8.42,
      "learning_rate": 5.187559354226021e-06,
      "loss": 0.2066,
      "step": 3940
    },
    {
      "epoch": 8.44,
      "learning_rate": 5.0688509021842354e-06,
      "loss": 0.2074,
      "step": 3950
    },
    {
      "epoch": 8.46,
      "learning_rate": 4.9501424501424505e-06,
      "loss": 0.2209,
      "step": 3960
    },
    {
      "epoch": 8.48,
      "learning_rate": 4.8314339981006655e-06,
      "loss": 0.2151,
      "step": 3970
    },
    {
      "epoch": 8.5,
      "learning_rate": 4.71272554605888e-06,
      "loss": 0.2016,
      "step": 3980
    },
    {
      "epoch": 8.52,
      "learning_rate": 4.594017094017095e-06,
      "loss": 0.2167,
      "step": 3990
    },
    {
      "epoch": 8.55,
      "learning_rate": 4.475308641975309e-06,
      "loss": 0.2113,
      "step": 4000
    },
    {
      "epoch": 8.57,
      "learning_rate": 4.356600189933523e-06,
      "loss": 0.2023,
      "step": 4010
    },
    {
      "epoch": 8.59,
      "learning_rate": 4.237891737891738e-06,
      "loss": 0.1946,
      "step": 4020
    },
    {
      "epoch": 8.61,
      "learning_rate": 4.119183285849953e-06,
      "loss": 0.2053,
      "step": 4030
    },
    {
      "epoch": 8.63,
      "learning_rate": 4.000474833808167e-06,
      "loss": 0.2036,
      "step": 4040
    },
    {
      "epoch": 8.65,
      "learning_rate": 3.8817663817663825e-06,
      "loss": 0.2234,
      "step": 4050
    },
    {
      "epoch": 8.67,
      "learning_rate": 3.7630579297245962e-06,
      "loss": 0.2337,
      "step": 4060
    },
    {
      "epoch": 8.7,
      "learning_rate": 3.644349477682811e-06,
      "loss": 0.1812,
      "step": 4070
    },
    {
      "epoch": 8.72,
      "learning_rate": 3.525641025641026e-06,
      "loss": 0.2122,
      "step": 4080
    },
    {
      "epoch": 8.74,
      "learning_rate": 3.4069325735992405e-06,
      "loss": 0.2294,
      "step": 4090
    },
    {
      "epoch": 8.76,
      "learning_rate": 3.288224121557455e-06,
      "loss": 0.194,
      "step": 4100
    },
    {
      "epoch": 8.78,
      "learning_rate": 3.16951566951567e-06,
      "loss": 0.2044,
      "step": 4110
    },
    {
      "epoch": 8.8,
      "learning_rate": 3.0508072174738843e-06,
      "loss": 0.2095,
      "step": 4120
    },
    {
      "epoch": 8.82,
      "learning_rate": 2.932098765432099e-06,
      "loss": 0.1992,
      "step": 4130
    },
    {
      "epoch": 8.84,
      "learning_rate": 2.8133903133903136e-06,
      "loss": 0.2114,
      "step": 4140
    },
    {
      "epoch": 8.87,
      "learning_rate": 2.694681861348528e-06,
      "loss": 0.2198,
      "step": 4150
    },
    {
      "epoch": 8.89,
      "learning_rate": 2.575973409306743e-06,
      "loss": 0.2108,
      "step": 4160
    },
    {
      "epoch": 8.91,
      "learning_rate": 2.4572649572649574e-06,
      "loss": 0.2065,
      "step": 4170
    },
    {
      "epoch": 8.93,
      "learning_rate": 2.338556505223172e-06,
      "loss": 0.1917,
      "step": 4180
    },
    {
      "epoch": 8.95,
      "learning_rate": 2.2198480531813866e-06,
      "loss": 0.2246,
      "step": 4190
    },
    {
      "epoch": 8.97,
      "learning_rate": 2.1011396011396013e-06,
      "loss": 0.2241,
      "step": 4200
    },
    {
      "epoch": 8.99,
      "learning_rate": 1.982431149097816e-06,
      "loss": 0.2002,
      "step": 4210
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.9076,
      "eval_loss": 0.26307567954063416,
      "eval_runtime": 41.9933,
      "eval_samples_per_second": 238.133,
      "eval_steps_per_second": 7.454,
      "step": 4212
    },
    {
      "epoch": 9.02,
      "learning_rate": 1.8637226970560307e-06,
      "loss": 0.2029,
      "step": 4220
    },
    {
      "epoch": 9.04,
      "learning_rate": 1.7450142450142451e-06,
      "loss": 0.2027,
      "step": 4230
    },
    {
      "epoch": 9.06,
      "learning_rate": 1.6263057929724597e-06,
      "loss": 0.2285,
      "step": 4240
    },
    {
      "epoch": 9.08,
      "learning_rate": 1.5075973409306743e-06,
      "loss": 0.213,
      "step": 4250
    },
    {
      "epoch": 9.1,
      "learning_rate": 1.388888888888889e-06,
      "loss": 0.1922,
      "step": 4260
    },
    {
      "epoch": 9.12,
      "learning_rate": 1.2701804368471036e-06,
      "loss": 0.2462,
      "step": 4270
    },
    {
      "epoch": 9.15,
      "learning_rate": 1.1514719848053182e-06,
      "loss": 0.2236,
      "step": 4280
    },
    {
      "epoch": 9.17,
      "learning_rate": 1.0327635327635328e-06,
      "loss": 0.215,
      "step": 4290
    },
    {
      "epoch": 9.19,
      "learning_rate": 9.140550807217474e-07,
      "loss": 0.1831,
      "step": 4300
    },
    {
      "epoch": 9.21,
      "learning_rate": 7.95346628679962e-07,
      "loss": 0.2108,
      "step": 4310
    },
    {
      "epoch": 9.23,
      "learning_rate": 6.766381766381766e-07,
      "loss": 0.1701,
      "step": 4320
    },
    {
      "epoch": 9.25,
      "learning_rate": 5.579297245963913e-07,
      "loss": 0.1931,
      "step": 4330
    },
    {
      "epoch": 9.27,
      "learning_rate": 4.392212725546059e-07,
      "loss": 0.2082,
      "step": 4340
    },
    {
      "epoch": 9.29,
      "learning_rate": 3.205128205128205e-07,
      "loss": 0.2034,
      "step": 4350
    },
    {
      "epoch": 9.32,
      "learning_rate": 2.0180436847103516e-07,
      "loss": 0.1671,
      "step": 4360
    },
    {
      "epoch": 9.34,
      "learning_rate": 8.309591642924977e-08,
      "loss": 0.2223,
      "step": 4370
    },
    {
      "epoch": 9.36,
      "learning_rate": 0.0,
      "loss": 0.1798,
      "step": 4380
    },
    {
      "epoch": 9.38,
      "learning_rate": 0.0,
      "loss": 0.1942,
      "step": 4390
    },
    {
      "epoch": 9.4,
      "learning_rate": 0.0,
      "loss": 0.2108,
      "step": 4400
    },
    {
      "epoch": 9.42,
      "learning_rate": 0.0,
      "loss": 0.1951,
      "step": 4410
    },
    {
      "epoch": 9.44,
      "learning_rate": 0.0,
      "loss": 0.2197,
      "step": 4420
    },
    {
      "epoch": 9.47,
      "learning_rate": 0.0,
      "loss": 0.1848,
      "step": 4430
    },
    {
      "epoch": 9.49,
      "learning_rate": 0.0,
      "loss": 0.177,
      "step": 4440
    },
    {
      "epoch": 9.51,
      "learning_rate": 0.0,
      "loss": 0.1737,
      "step": 4450
    },
    {
      "epoch": 9.53,
      "learning_rate": 0.0,
      "loss": 0.1813,
      "step": 4460
    },
    {
      "epoch": 9.55,
      "learning_rate": 0.0,
      "loss": 0.1726,
      "step": 4470
    },
    {
      "epoch": 9.57,
      "learning_rate": 0.0,
      "loss": 0.1778,
      "step": 4480
    },
    {
      "epoch": 9.59,
      "learning_rate": 0.0,
      "loss": 0.2027,
      "step": 4490
    },
    {
      "epoch": 9.61,
      "learning_rate": 0.0,
      "loss": 0.2117,
      "step": 4500
    },
    {
      "epoch": 9.64,
      "learning_rate": 0.0,
      "loss": 0.2084,
      "step": 4510
    },
    {
      "epoch": 9.66,
      "learning_rate": 0.0,
      "loss": 0.2076,
      "step": 4520
    },
    {
      "epoch": 9.68,
      "learning_rate": 0.0,
      "loss": 0.1976,
      "step": 4530
    },
    {
      "epoch": 9.7,
      "learning_rate": 0.0,
      "loss": 0.2229,
      "step": 4540
    },
    {
      "epoch": 9.72,
      "learning_rate": 0.0,
      "loss": 0.2032,
      "step": 4550
    },
    {
      "epoch": 9.74,
      "learning_rate": 0.0,
      "loss": 0.2014,
      "step": 4560
    },
    {
      "epoch": 9.76,
      "learning_rate": 0.0,
      "loss": 0.1919,
      "step": 4570
    },
    {
      "epoch": 9.79,
      "learning_rate": 0.0,
      "loss": 0.2093,
      "step": 4580
    },
    {
      "epoch": 9.81,
      "learning_rate": 0.0,
      "loss": 0.208,
      "step": 4590
    },
    {
      "epoch": 9.83,
      "learning_rate": 0.0,
      "loss": 0.1757,
      "step": 4600
    },
    {
      "epoch": 9.85,
      "learning_rate": 0.0,
      "loss": 0.1822,
      "step": 4610
    },
    {
      "epoch": 9.87,
      "learning_rate": 0.0,
      "loss": 0.2009,
      "step": 4620
    },
    {
      "epoch": 9.89,
      "learning_rate": 0.0,
      "loss": 0.2252,
      "step": 4630
    },
    {
      "epoch": 9.91,
      "learning_rate": 0.0,
      "loss": 0.1827,
      "step": 4640
    },
    {
      "epoch": 9.93,
      "learning_rate": 0.0,
      "loss": 0.1731,
      "step": 4650
    },
    {
      "epoch": 9.96,
      "learning_rate": 0.0,
      "loss": 0.173,
      "step": 4660
    },
    {
      "epoch": 9.98,
      "learning_rate": 0.0,
      "loss": 0.197,
      "step": 4670
    },
    {
      "epoch": 10.0,
      "learning_rate": 0.0,
      "loss": 0.193,
      "step": 4680
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.9086,
      "eval_loss": 0.25825217366218567,
      "eval_runtime": 42.0394,
      "eval_samples_per_second": 237.872,
      "eval_steps_per_second": 7.445,
      "step": 4680
    },
    {
      "epoch": 10.0,
      "step": 4680,
      "total_flos": 4.650096471071824e+19,
      "train_loss": 0.279335702662794,
      "train_runtime": 6125.6563,
      "train_samples_per_second": 97.949,
      "train_steps_per_second": 0.764
    }
  ],
  "max_steps": 4680,
  "num_train_epochs": 10,
  "total_flos": 4.650096471071824e+19,
  "trial_name": null,
  "trial_params": null
}