{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.998535871156662,
  "eval_steps": 500,
  "global_step": 3072,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 4.9998692735249987e-05,
      "loss": 0.9611,
      "step": 10
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9994771077715244e-05,
      "loss": 0.94,
      "step": 20
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.998823543752733e-05,
      "loss": 0.9521,
      "step": 30
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9979086498191216e-05,
      "loss": 0.9413,
      "step": 40
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9967325216513785e-05,
      "loss": 0.8717,
      "step": 50
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.995295282250373e-05,
      "loss": 0.9099,
      "step": 60
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.993597081924299e-05,
      "loss": 0.9671,
      "step": 70
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.991638098272951e-05,
      "loss": 0.9971,
      "step": 80
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.98941853616915e-05,
      "loss": 0.9215,
      "step": 90
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.98693862773732e-05,
      "loss": 0.9008,
      "step": 100
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.98419863232921e-05,
      "loss": 0.8907,
      "step": 110
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.981198836496775e-05,
      "loss": 0.8838,
      "step": 120
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.9779395539622024e-05,
      "loss": 0.9156,
      "step": 130
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.9744211255851046e-05,
      "loss": 0.8241,
      "step": 140
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.9706439193268736e-05,
      "loss": 0.8777,
      "step": 150
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.966608330212198e-05,
      "loss": 0.8757,
      "step": 160
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.962314780287748e-05,
      "loss": 0.8574,
      "step": 170
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.957763718578041e-05,
      "loss": 0.8419,
      "step": 180
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.952955621038482e-05,
      "loss": 0.8368,
      "step": 190
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.9478909905055845e-05,
      "loss": 0.9646,
      "step": 200
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.942570356644386e-05,
      "loss": 0.9666,
      "step": 210
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.936994275893054e-05,
      "loss": 0.9451,
      "step": 220
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.931163331404694e-05,
      "loss": 0.8937,
      "step": 230
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.9250781329863606e-05,
      "loss": 0.9066,
      "step": 240
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.9187393170352844e-05,
      "loss": 0.8378,
      "step": 250
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.9121475464723175e-05,
      "loss": 0.8696,
      "step": 260
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.9053035106726045e-05,
      "loss": 0.8534,
      "step": 270
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.898207925393484e-05,
      "loss": 0.8417,
      "step": 280
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.890861532699639e-05,
      "loss": 0.8718,
      "step": 290
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.8832651008854845e-05,
      "loss": 0.9024,
      "step": 300
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.8754194243948245e-05,
      "loss": 0.8846,
      "step": 310
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.867325323737765e-05,
      "loss": 0.8569,
      "step": 320
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.858983645404901e-05,
      "loss": 0.863,
      "step": 330
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.8503952617787983e-05,
      "loss": 0.7955,
      "step": 340
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.84156107104275e-05,
      "loss": 0.913,
      "step": 350
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.8324819970868473e-05,
      "loss": 0.8538,
      "step": 360
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.8231589894113584e-05,
      "loss": 0.9323,
      "step": 370
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.813593023027427e-05,
      "loss": 0.8345,
      "step": 380
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.803785098355105e-05,
      "loss": 0.8482,
      "step": 390
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.793736241118728e-05,
      "loss": 0.9007,
      "step": 400
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.783447502239645e-05,
      "loss": 0.8718,
      "step": 410
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.7729199577263065e-05,
      "loss": 0.8733,
      "step": 420
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.7621547085617406e-05,
      "loss": 0.7678,
      "step": 430
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.751152880588409e-05,
      "loss": 0.8189,
      "step": 440
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.739915624390463e-05,
      "loss": 0.9303,
      "step": 450
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.728444115173415e-05,
      "loss": 0.8666,
      "step": 460
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.716739552641237e-05,
      "loss": 0.8999,
      "step": 470
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.7048031608708876e-05,
      "loss": 0.8813,
      "step": 480
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.692636188184305e-05,
      "loss": 0.9063,
      "step": 490
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.680239907017849e-05,
      "loss": 0.8109,
      "step": 500
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.667615613789232e-05,
      "loss": 0.825,
      "step": 510
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.654764628761936e-05,
      "loss": 0.9117,
      "step": 520
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.6416882959071395e-05,
      "loss": 0.8901,
      "step": 530
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.628387982763163e-05,
      "loss": 0.8362,
      "step": 540
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.614865080292449e-05,
      "loss": 0.9148,
      "step": 550
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.601121002736095e-05,
      "loss": 0.9449,
      "step": 560
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.58715718746595e-05,
      "loss": 0.7937,
      "step": 570
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.5729750948342926e-05,
      "loss": 0.835,
      "step": 580
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.558576208021105e-05,
      "loss": 0.9117,
      "step": 590
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.5439620328789593e-05,
      "loss": 0.7855,
      "step": 600
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.529134097775539e-05,
      "loss": 0.8816,
      "step": 610
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.514093953433792e-05,
      "loss": 0.8344,
      "step": 620
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.498843172769763e-05,
      "loss": 0.8936,
      "step": 630
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.4833833507280884e-05,
      "loss": 0.9126,
      "step": 640
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.4677161041152005e-05,
      "loss": 0.8691,
      "step": 650
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.451843071430236e-05,
      "loss": 0.8383,
      "step": 660
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.435765912693686e-05,
      "loss": 0.9212,
      "step": 670
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.419486309273781e-05,
      "loss": 0.8768,
      "step": 680
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.403005963710655e-05,
      "loss": 0.9103,
      "step": 690
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.386326599538295e-05,
      "loss": 0.8308,
      "step": 700
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.369449961104289e-05,
      "loss": 0.9243,
      "step": 710
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.352377813387398e-05,
      "loss": 0.8994,
      "step": 720
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.335111941812975e-05,
      "loss": 0.9656,
      "step": 730
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.3176541520662426e-05,
      "loss": 0.901,
      "step": 740
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.3000062699034544e-05,
      "loss": 0.924,
      "step": 750
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.28217014096095e-05,
      "loss": 0.8821,
      "step": 760
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.264147630562141e-05,
      "loss": 0.8631,
      "step": 770
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.245940623522433e-05,
      "loss": 0.9012,
      "step": 780
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.227551023952104e-05,
      "loss": 0.9543,
      "step": 790
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.208980755057178e-05,
      "loss": 0.8243,
      "step": 800
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.19023175893829e-05,
      "loss": 0.8568,
      "step": 810
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.171305996387577e-05,
      "loss": 0.9434,
      "step": 820
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.1522054466836186e-05,
      "loss": 0.8362,
      "step": 830
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.1329321073844415e-05,
      "loss": 0.8904,
      "step": 840
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.1134879941186135e-05,
      "loss": 0.9025,
      "step": 850
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.093875140374443e-05,
      "loss": 0.872,
      "step": 860
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.074095597287318e-05,
      "loss": 0.8268,
      "step": 870
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.054151433425194e-05,
      "loss": 0.8755,
      "step": 880
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.034044734572262e-05,
      "loss": 0.8389,
      "step": 890
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.013777603510814e-05,
      "loss": 0.8528,
      "step": 900
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.993352159801332e-05,
      "loss": 0.8036,
      "step": 910
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.9727705395608205e-05,
      "loss": 0.8854,
      "step": 920
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.9520348952394115e-05,
      "loss": 0.8261,
      "step": 930
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.931147395395256e-05,
      "loss": 0.842,
      "step": 940
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.9101102244677335e-05,
      "loss": 0.8749,
      "step": 950
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.888925582549006e-05,
      "loss": 0.8799,
      "step": 960
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.867595685153922e-05,
      "loss": 0.8573,
      "step": 970
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.846122762988319e-05,
      "loss": 0.8694,
      "step": 980
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.824509061715738e-05,
      "loss": 0.927,
      "step": 990
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.802756841722559e-05,
      "loss": 0.8816,
      "step": 1000
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.780868377881619e-05,
      "loss": 0.9129,
      "step": 1010
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.758845959314294e-05,
      "loss": 0.8362,
      "step": 1020
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.7366918891511046e-05,
      "loss": 0.8907,
      "step": 1030
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.7144084842908505e-05,
      "loss": 0.9378,
      "step": 1040
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.691998075158306e-05,
      "loss": 0.9064,
      "step": 1050
    },
    {
      "epoch": 1.03,
      "learning_rate": 3.669463005460502e-05,
      "loss": 0.9101,
      "step": 1060
    },
    {
      "epoch": 1.04,
      "learning_rate": 3.646805631941619e-05,
      "loss": 0.8307,
      "step": 1070
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.624028324136517e-05,
      "loss": 0.8958,
      "step": 1080
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.6011334641229215e-05,
      "loss": 0.8681,
      "step": 1090
    },
    {
      "epoch": 1.07,
      "learning_rate": 3.578123446272309e-05,
      "loss": 0.8711,
      "step": 1100
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.5550006769994996e-05,
      "loss": 0.8607,
      "step": 1110
    },
    {
      "epoch": 1.09,
      "learning_rate": 3.5317675745109866e-05,
      "loss": 0.9532,
      "step": 1120
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.508426568552045e-05,
      "loss": 0.848,
      "step": 1130
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.484980100152621e-05,
      "loss": 0.8097,
      "step": 1140
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.461430621372044e-05,
      "loss": 0.8612,
      "step": 1150
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.437780595042595e-05,
      "loss": 0.8281,
      "step": 1160
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.414032494511935e-05,
      "loss": 0.8297,
      "step": 1170
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.390188803384439e-05,
      "loss": 0.8607,
      "step": 1180
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.3662520152614615e-05,
      "loss": 0.8986,
      "step": 1190
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.34222463348055e-05,
      "loss": 0.8884,
      "step": 1200
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.3181091708536435e-05,
      "loss": 0.8901,
      "step": 1210
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.293908149404279e-05,
      "loss": 0.8583,
      "step": 1220
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.2696241001038374e-05,
      "loss": 0.7928,
      "step": 1230
    },
    {
      "epoch": 1.21,
      "learning_rate": 3.245259562606851e-05,
      "loss": 0.8508,
      "step": 1240
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.220817084985398e-05,
      "loss": 0.8245,
      "step": 1250
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.196299223462633e-05,
      "loss": 0.8746,
      "step": 1260
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.171708542145445e-05,
      "loss": 0.8666,
      "step": 1270
    },
    {
      "epoch": 1.25,
      "learning_rate": 3.147047612756302e-05,
      "loss": 0.8807,
      "step": 1280
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.1223190143643014e-05,
      "loss": 0.8557,
      "step": 1290
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.09752533311544e-05,
      "loss": 0.842,
      "step": 1300
    },
    {
      "epoch": 1.28,
      "learning_rate": 3.0726691619621617e-05,
      "loss": 0.8701,
      "step": 1310
    },
    {
      "epoch": 1.29,
      "learning_rate": 3.0477531003921745e-05,
      "loss": 0.8755,
      "step": 1320
    },
    {
      "epoch": 1.3,
      "learning_rate": 3.0227797541566e-05,
      "loss": 0.9009,
      "step": 1330
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.9977517349974544e-05,
      "loss": 0.8578,
      "step": 1340
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.9726716603745157e-05,
      "loss": 0.8067,
      "step": 1350
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.9475421531915827e-05,
      "loss": 0.8139,
      "step": 1360
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.922365841522167e-05,
      "loss": 0.8382,
      "step": 1370
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.8971453583346536e-05,
      "loss": 0.9073,
      "step": 1380
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.871883341216934e-05,
      "loss": 0.8783,
      "step": 1390
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.846582432100567e-05,
      "loss": 0.8555,
      "step": 1400
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.8212452769844834e-05,
      "loss": 0.8236,
      "step": 1410
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.795874525658263e-05,
      "loss": 0.8816,
      "step": 1420
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.7704728314250168e-05,
      "loss": 0.8199,
      "step": 1430
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.7450428508239024e-05,
      "loss": 0.8271,
      "step": 1440
    },
    {
      "epoch": 1.42,
      "learning_rate": 2.7195872433522974e-05,
      "loss": 0.8444,
      "step": 1450
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.694108671187669e-05,
      "loss": 0.8232,
      "step": 1460
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.66860979890916e-05,
      "loss": 0.7959,
      "step": 1470
    },
    {
      "epoch": 1.44,
      "learning_rate": 2.6430932932189224e-05,
      "loss": 0.8368,
      "step": 1480
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.617561822663228e-05,
      "loss": 0.8336,
      "step": 1490
    },
    {
      "epoch": 1.46,
      "learning_rate": 2.5920180573533976e-05,
      "loss": 0.8431,
      "step": 1500
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.566464668686547e-05,
      "loss": 0.8309,
      "step": 1510
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.5409043290662173e-05,
      "loss": 0.9332,
      "step": 1520
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.5153397116228865e-05,
      "loss": 0.8416,
      "step": 1530
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.489773489934413e-05,
      "loss": 0.8978,
      "step": 1540
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.4642083377464305e-05,
      "loss": 0.8724,
      "step": 1550
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.4386469286927196e-05,
      "loss": 0.9087,
      "step": 1560
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.413091936015603e-05,
      "loss": 0.8749,
      "step": 1570
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.3875460322863697e-05,
      "loss": 0.8781,
      "step": 1580
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.3620118891257754e-05,
      "loss": 0.8544,
      "step": 1590
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.3364921769246423e-05,
      "loss": 0.8244,
      "step": 1600
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.3109895645645867e-05,
      "loss": 0.8656,
      "step": 1610
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.2855067191389006e-05,
      "loss": 0.888,
      "step": 1620
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.2600463056736293e-05,
      "loss": 0.8219,
      "step": 1630
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.2346109868488567e-05,
      "loss": 0.8732,
      "step": 1640
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.2092034227202383e-05,
      "loss": 0.8984,
      "step": 1650
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.183826270440812e-05,
      "loss": 0.8474,
      "step": 1660
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.158482183983109e-05,
      "loss": 0.8423,
      "step": 1670
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.1331738138615958e-05,
      "loss": 0.8716,
      "step": 1680
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.1079038068554842e-05,
      "loss": 0.8783,
      "step": 1690
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.082674805731926e-05,
      "loss": 0.7759,
      "step": 1700
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.0574894489696285e-05,
      "loss": 0.8975,
      "step": 1710
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.03235037048292e-05,
      "loss": 0.8425,
      "step": 1720
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.0072601993462946e-05,
      "loss": 0.824,
      "step": 1730
    },
    {
      "epoch": 1.7,
      "learning_rate": 1.982221559519454e-05,
      "loss": 0.7989,
      "step": 1740
    },
    {
      "epoch": 1.71,
      "learning_rate": 1.957237069572898e-05,
      "loss": 0.8145,
      "step": 1750
    },
    {
      "epoch": 1.72,
      "learning_rate": 1.932309342414067e-05,
      "loss": 0.8581,
      "step": 1760
    },
    {
      "epoch": 1.73,
      "learning_rate": 1.907440985014082e-05,
      "loss": 0.8886,
      "step": 1770
    },
    {
      "epoch": 1.74,
      "learning_rate": 1.8826345981351044e-05,
      "loss": 0.8048,
      "step": 1780
    },
    {
      "epoch": 1.75,
      "learning_rate": 1.8578927760583463e-05,
      "loss": 0.8534,
      "step": 1790
    },
    {
      "epoch": 1.76,
      "learning_rate": 1.8332181063127545e-05,
      "loss": 0.8392,
      "step": 1800
    },
    {
      "epoch": 1.77,
      "learning_rate": 1.8086131694044077e-05,
      "loss": 0.8926,
      "step": 1810
    },
    {
      "epoch": 1.78,
      "learning_rate": 1.784080538546642e-05,
      "loss": 0.923,
      "step": 1820
    },
    {
      "epoch": 1.79,
      "learning_rate": 1.7596227793909402e-05,
      "loss": 0.8209,
      "step": 1830
    },
    {
      "epoch": 1.8,
      "learning_rate": 1.7352424497586163e-05,
      "loss": 0.813,
      "step": 1840
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.7109420993733115e-05,
      "loss": 0.8375,
      "step": 1850
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.686724269594343e-05,
      "loss": 0.8477,
      "step": 1860
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.6625914931509262e-05,
      "loss": 0.8263,
      "step": 1870
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.6385462938772992e-05,
      "loss": 0.8816,
      "step": 1880
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.614591186448774e-05,
      "loss": 0.8677,
      "step": 1890
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.5907286761187535e-05,
      "loss": 0.9021,
      "step": 1900
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.5669612584567265e-05,
      "loss": 0.8449,
      "step": 1910
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.5432914190872757e-05,
      "loss": 0.8328,
      "step": 1920
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.5197216334301339e-05,
      "loss": 0.8591,
      "step": 1930
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.4962543664412986e-05,
      "loss": 0.9411,
      "step": 1940
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.4728920723552407e-05,
      "loss": 0.9106,
      "step": 1950
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.4496371944282441e-05,
      "loss": 0.8332,
      "step": 1960
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.4264921646828838e-05,
      "loss": 0.783,
      "step": 1970
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.4034594036536817e-05,
      "loss": 0.8703,
      "step": 1980
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.3805413201339632e-05,
      "loss": 0.8918,
      "step": 1990
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.3577403109239484e-05,
      "loss": 0.8795,
      "step": 2000
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.3350587605800852e-05,
      "loss": 0.8222,
      "step": 2010
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.3124990411656706e-05,
      "loss": 0.8187,
      "step": 2020
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.2900635120027859e-05,
      "loss": 0.8343,
      "step": 2030
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.2677545194255402e-05,
      "loss": 0.9145,
      "step": 2040
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.2455743965347047e-05,
      "loss": 0.913,
      "step": 2050
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.223525462953703e-05,
      "loss": 0.9304,
      "step": 2060
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.201610024586026e-05,
      "loss": 0.82,
      "step": 2070
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.1798303733740802e-05,
      "loss": 0.8232,
      "step": 2080
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.1581887870594884e-05,
      "loss": 0.8997,
      "step": 2090
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.1366875289448845e-05,
      "loss": 0.9386,
      "step": 2100
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.1153288476572126e-05,
      "loss": 0.8515,
      "step": 2110
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.0941149769125674e-05,
      "loss": 0.8724,
      "step": 2120
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.0730481352825823e-05,
      "loss": 0.8616,
      "step": 2130
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.0521305259624129e-05,
      "loss": 0.82,
      "step": 2140
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.0313643365403274e-05,
      "loss": 0.866,
      "step": 2150
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.0107517387689166e-05,
      "loss": 0.7712,
      "step": 2160
    },
    {
      "epoch": 2.12,
      "learning_rate": 9.90294888337981e-06,
      "loss": 0.8451,
      "step": 2170
    },
    {
      "epoch": 2.13,
      "learning_rate": 9.699959246490762e-06,
      "loss": 0.7923,
      "step": 2180
    },
    {
      "epoch": 2.14,
      "learning_rate": 9.498569705917771e-06,
      "loss": 0.8547,
      "step": 2190
    },
    {
      "epoch": 2.15,
      "learning_rate": 9.298801323216647e-06,
      "loss": 0.8624,
      "step": 2200
    },
    {
      "epoch": 2.16,
      "learning_rate": 9.100674990400571e-06,
      "loss": 0.8079,
      "step": 2210
    },
    {
      "epoch": 2.17,
      "learning_rate": 8.904211427755218e-06,
      "loss": 0.8211,
      "step": 2220
    },
    {
      "epoch": 2.18,
      "learning_rate": 8.70943118167179e-06,
      "loss": 0.9058,
      "step": 2230
    },
    {
      "epoch": 2.19,
      "learning_rate": 8.51635462249828e-06,
      "loss": 0.7715,
      "step": 2240
    },
    {
      "epoch": 2.2,
      "learning_rate": 8.325001942409066e-06,
      "loss": 0.8093,
      "step": 2250
    },
    {
      "epoch": 2.21,
      "learning_rate": 8.13539315329322e-06,
      "loss": 0.8812,
      "step": 2260
    },
    {
      "epoch": 2.22,
      "learning_rate": 7.947548084661674e-06,
      "loss": 0.8675,
      "step": 2270
    },
    {
      "epoch": 2.23,
      "learning_rate": 7.761486381573327e-06,
      "loss": 0.7757,
      "step": 2280
    },
    {
      "epoch": 2.24,
      "learning_rate": 7.577227502580667e-06,
      "loss": 0.9026,
      "step": 2290
    },
    {
      "epoch": 2.24,
      "learning_rate": 7.394790717694677e-06,
      "loss": 0.8779,
      "step": 2300
    },
    {
      "epoch": 2.25,
      "learning_rate": 7.214195106369587e-06,
      "loss": 0.8702,
      "step": 2310
    },
    {
      "epoch": 2.26,
      "learning_rate": 7.035459555507548e-06,
      "loss": 0.8602,
      "step": 2320
    },
    {
      "epoch": 2.27,
      "learning_rate": 6.858602757483371e-06,
      "loss": 0.8647,
      "step": 2330
    },
    {
      "epoch": 2.28,
      "learning_rate": 6.683643208189683e-06,
      "loss": 0.8216,
      "step": 2340
    },
    {
      "epoch": 2.29,
      "learning_rate": 6.510599205102594e-06,
      "loss": 0.8674,
      "step": 2350
    },
    {
      "epoch": 2.3,
      "learning_rate": 6.339488845368155e-06,
      "loss": 0.8232,
      "step": 2360
    },
    {
      "epoch": 2.31,
      "learning_rate": 6.17033002390969e-06,
      "loss": 0.8058,
      "step": 2370
    },
    {
      "epoch": 2.32,
      "learning_rate": 6.003140431556345e-06,
      "loss": 0.8183,
      "step": 2380
    },
    {
      "epoch": 2.33,
      "learning_rate": 5.837937553192999e-06,
      "loss": 0.8745,
      "step": 2390
    },
    {
      "epoch": 2.34,
      "learning_rate": 5.674738665931575e-06,
      "loss": 0.7398,
      "step": 2400
    },
    {
      "epoch": 2.35,
      "learning_rate": 5.513560837304304e-06,
      "loss": 0.8321,
      "step": 2410
    },
    {
      "epoch": 2.36,
      "learning_rate": 5.354420923478687e-06,
      "loss": 0.8657,
      "step": 2420
    },
    {
      "epoch": 2.37,
      "learning_rate": 5.197335567494696e-06,
      "loss": 0.8899,
      "step": 2430
    },
    {
      "epoch": 2.38,
      "learning_rate": 5.042321197524213e-06,
      "loss": 0.8185,
      "step": 2440
    },
    {
      "epoch": 2.39,
      "learning_rate": 4.889394025152979e-06,
      "loss": 0.9032,
      "step": 2450
    },
    {
      "epoch": 2.4,
      "learning_rate": 4.738570043685134e-06,
      "loss": 0.8692,
      "step": 2460
    },
    {
      "epoch": 2.41,
      "learning_rate": 4.589865026470619e-06,
      "loss": 0.8571,
      "step": 2470
    },
    {
      "epoch": 2.42,
      "learning_rate": 4.4432945252556284e-06,
      "loss": 0.8563,
      "step": 2480
    },
    {
      "epoch": 2.43,
      "learning_rate": 4.298873868556108e-06,
      "loss": 0.8423,
      "step": 2490
    },
    {
      "epoch": 2.44,
      "learning_rate": 4.156618160054771e-06,
      "loss": 0.8008,
      "step": 2500
    },
    {
      "epoch": 2.45,
      "learning_rate": 4.016542277021465e-06,
      "loss": 0.9192,
      "step": 2510
    },
    {
      "epoch": 2.46,
      "learning_rate": 3.878660868757323e-06,
      "loss": 0.9414,
      "step": 2520
    },
    {
      "epoch": 2.47,
      "learning_rate": 3.742988355062735e-06,
      "loss": 0.8645,
      "step": 2530
    },
    {
      "epoch": 2.48,
      "learning_rate": 3.609538924729269e-06,
      "loss": 0.7851,
      "step": 2540
    },
    {
      "epoch": 2.49,
      "learning_rate": 3.478326534055823e-06,
      "loss": 0.8207,
      "step": 2550
    },
    {
      "epoch": 2.5,
      "learning_rate": 3.3493649053890326e-06,
      "loss": 0.8443,
      "step": 2560
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.2226675256882167e-06,
      "loss": 0.841,
      "step": 2570
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.098247645114838e-06,
      "loss": 0.8591,
      "step": 2580
    },
    {
      "epoch": 2.53,
      "learning_rate": 2.9761182756468165e-06,
      "loss": 0.8146,
      "step": 2590
    },
    {
      "epoch": 2.54,
      "learning_rate": 2.856292189717741e-06,
      "loss": 0.8441,
      "step": 2600
    },
    {
      "epoch": 2.55,
      "learning_rate": 2.7387819188810534e-06,
      "loss": 0.878,
      "step": 2610
    },
    {
      "epoch": 2.56,
      "learning_rate": 2.623599752499553e-06,
      "loss": 0.8543,
      "step": 2620
    },
    {
      "epoch": 2.57,
      "learning_rate": 2.510757736460112e-06,
      "loss": 0.8435,
      "step": 2630
    },
    {
      "epoch": 2.58,
      "learning_rate": 2.4002676719139166e-06,
      "loss": 0.8424,
      "step": 2640
    },
    {
      "epoch": 2.59,
      "learning_rate": 2.2921411140423037e-06,
      "loss": 0.8488,
      "step": 2650
    },
    {
      "epoch": 2.6,
      "learning_rate": 2.1863893708482795e-06,
      "loss": 0.8021,
      "step": 2660
    },
    {
      "epoch": 2.61,
      "learning_rate": 2.0830235019739352e-06,
      "loss": 0.8198,
      "step": 2670
    },
    {
      "epoch": 2.62,
      "learning_rate": 1.9820543175438077e-06,
      "loss": 0.7836,
      "step": 2680
    },
    {
      "epoch": 2.63,
      "learning_rate": 1.8834923770343544e-06,
      "loss": 0.7949,
      "step": 2690
    },
    {
      "epoch": 2.64,
      "learning_rate": 1.787347988169613e-06,
      "loss": 0.8243,
      "step": 2700
    },
    {
      "epoch": 2.65,
      "learning_rate": 1.6936312058432158e-06,
      "loss": 0.8798,
      "step": 2710
    },
    {
      "epoch": 2.65,
      "learning_rate": 1.6023518310668618e-06,
      "loss": 0.9497,
      "step": 2720
    },
    {
      "epoch": 2.66,
      "learning_rate": 1.5135194099452521e-06,
      "loss": 0.8978,
      "step": 2730
    },
    {
      "epoch": 2.67,
      "learning_rate": 1.4271432326778273e-06,
      "loss": 0.8524,
      "step": 2740
    },
    {
      "epoch": 2.68,
      "learning_rate": 1.3432323325871216e-06,
      "loss": 0.7888,
      "step": 2750
    },
    {
      "epoch": 2.69,
      "learning_rate": 1.2617954851740832e-06,
      "loss": 0.8624,
      "step": 2760
    },
    {
      "epoch": 2.7,
      "learning_rate": 1.1828412072003108e-06,
      "loss": 0.8722,
      "step": 2770
    },
    {
      "epoch": 2.71,
      "learning_rate": 1.1063777557973638e-06,
      "loss": 0.8501,
      "step": 2780
    },
    {
      "epoch": 2.72,
      "learning_rate": 1.0324131276032127e-06,
      "loss": 0.8302,
      "step": 2790
    },
    {
      "epoch": 2.73,
      "learning_rate": 9.609550579259496e-07,
      "loss": 0.8424,
      "step": 2800
    },
    {
      "epoch": 2.74,
      "learning_rate": 8.920110199348325e-07,
      "loss": 0.8208,
      "step": 2810
    },
    {
      "epoch": 2.75,
      "learning_rate": 8.255882238786983e-07,
      "loss": 0.8773,
      "step": 2820
    },
    {
      "epoch": 2.76,
      "learning_rate": 7.616936163319404e-07,
      "loss": 0.7867,
      "step": 2830
    },
    {
      "epoch": 2.77,
      "learning_rate": 7.003338794680153e-07,
      "loss": 0.847,
      "step": 2840
    },
    {
      "epoch": 2.78,
      "learning_rate": 6.415154303606036e-07,
      "loss": 0.8479,
      "step": 2850
    },
    {
      "epoch": 2.79,
      "learning_rate": 5.852444203125223e-07,
      "loss": 0.8197,
      "step": 2860
    },
    {
      "epoch": 2.8,
      "learning_rate": 5.315267342124008e-07,
      "loss": 0.8719,
      "step": 2870
    },
    {
      "epoch": 2.81,
      "learning_rate": 4.803679899192392e-07,
      "loss": 0.922,
      "step": 2880
    },
    {
      "epoch": 2.82,
      "learning_rate": 4.3177353767488163e-07,
      "loss": 0.8622,
      "step": 2890
    },
    {
      "epoch": 2.83,
      "learning_rate": 3.8574845954448823e-07,
      "loss": 0.8361,
      "step": 2900
    },
    {
      "epoch": 2.84,
      "learning_rate": 3.4229756888503604e-07,
      "loss": 0.9201,
      "step": 2910
    },
    {
      "epoch": 2.85,
      "learning_rate": 3.014254098419406e-07,
      "loss": 0.834,
      "step": 2920
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.6313625687383046e-07,
      "loss": 0.8569,
      "step": 2930
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.2743411430549976e-07,
      "loss": 0.8411,
      "step": 2940
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.943227159091482e-07,
      "loss": 0.8648,
      "step": 2950
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.6380552451389086e-07,
      "loss": 0.8477,
      "step": 2960
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.3588573164361463e-07,
      "loss": 0.8449,
      "step": 2970
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.1056625718320334e-07,
      "loss": 0.8185,
      "step": 2980
    },
    {
      "epoch": 2.92,
      "learning_rate": 8.784974907317111e-08,
      "loss": 0.8326,
      "step": 2990
    },
    {
      "epoch": 2.93,
      "learning_rate": 6.773858303274483e-08,
      "loss": 0.84,
      "step": 3000
    },
    {
      "epoch": 2.94,
      "learning_rate": 5.0234862311399065e-08,
      "loss": 0.7393,
      "step": 3010
    },
    {
      "epoch": 2.95,
      "learning_rate": 3.5340417468898714e-08,
      "loss": 0.8723,
      "step": 3020
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.3056806183865965e-08,
      "loss": 0.8481,
      "step": 3030
    },
    {
      "epoch": 2.97,
      "learning_rate": 1.3385313090857887e-08,
      "loss": 0.8273,
      "step": 3040
    },
    {
      "epoch": 2.98,
      "learning_rate": 6.326949646040525e-09,
      "loss": 0.899,
      "step": 3050
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.8824540213874255e-09,
      "loss": 0.8056,
      "step": 3060
    },
    {
      "epoch": 3.0,
      "learning_rate": 5.229102749415571e-11,
      "loss": 0.8446,
      "step": 3070
    },
    {
      "epoch": 3.0,
      "step": 3072,
      "total_flos": 1.125146154656809e+18,
      "train_loss": 0.862230001638333,
      "train_runtime": 23418.3393,
      "train_samples_per_second": 2.1,
      "train_steps_per_second": 0.131
    }
  ],
  "logging_steps": 10,
  "max_steps": 3072,
  "num_train_epochs": 3,
  "save_steps": 1000,
  "total_flos": 1.125146154656809e+18,
  "trial_name": null,
  "trial_params": null
}