{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 1484,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 9.999875511956753e-05,
      "loss": 1.7561,
      "step": 10
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.999502054025923e-05,
      "loss": 1.7194,
      "step": 20
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.998879644803927e-05,
      "loss": 1.7705,
      "step": 30
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.998008315283767e-05,
      "loss": 1.5094,
      "step": 40
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.996888108853486e-05,
      "loss": 1.5201,
      "step": 50
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.995519081294009e-05,
      "loss": 1.637,
      "step": 60
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.993901300776359e-05,
      "loss": 1.6338,
      "step": 70
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.992034847858267e-05,
      "loss": 1.5496,
      "step": 80
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.989919815480165e-05,
      "loss": 1.5669,
      "step": 90
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.987556308960549e-05,
      "loss": 1.5622,
      "step": 100
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.984944445990735e-05,
      "loss": 1.5836,
      "step": 110
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.982084356629015e-05,
      "loss": 1.5877,
      "step": 120
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.978976183294153e-05,
      "loss": 1.5465,
      "step": 130
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.97562008075832e-05,
      "loss": 1.4834,
      "step": 140
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.97201621613937e-05,
      "loss": 1.4956,
      "step": 150
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.968164768892526e-05,
      "loss": 1.3877,
      "step": 160
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.964065930801438e-05,
      "loss": 1.3466,
      "step": 170
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.95971990596864e-05,
      "loss": 1.4023,
      "step": 180
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.955126910805386e-05,
      "loss": 1.351,
      "step": 190
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.950287174020863e-05,
      "loss": 1.424,
      "step": 200
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.94520093661082e-05,
      "loss": 1.5038,
      "step": 210
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.939868451845554e-05,
      "loss": 1.3436,
      "step": 220
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.934289985257299e-05,
      "loss": 1.3446,
      "step": 230
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.928465814627014e-05,
      "loss": 1.4063,
      "step": 240
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.922396229970541e-05,
      "loss": 1.3803,
      "step": 250
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.916081533524167e-05,
      "loss": 1.2975,
      "step": 260
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.909522039729571e-05,
      "loss": 1.3689,
      "step": 270
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.902718075218176e-05,
      "loss": 1.4446,
      "step": 280
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.895669978794869e-05,
      "loss": 1.237,
      "step": 290
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.888378101421147e-05,
      "loss": 1.333,
      "step": 300
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.880842806197625e-05,
      "loss": 1.3933,
      "step": 310
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.873064468345969e-05,
      "loss": 1.3735,
      "step": 320
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.865043475190201e-05,
      "loss": 1.172,
      "step": 330
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.856780226137419e-05,
      "loss": 1.2538,
      "step": 340
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.848275132657903e-05,
      "loss": 1.2588,
      "step": 350
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.839528618264633e-05,
      "loss": 1.3531,
      "step": 360
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.830541118492193e-05,
      "loss": 1.4203,
      "step": 370
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.821313080875088e-05,
      "loss": 1.4292,
      "step": 380
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.811844964925454e-05,
      "loss": 1.3098,
      "step": 390
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.802137242110185e-05,
      "loss": 1.2093,
      "step": 400
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.792190395827447e-05,
      "loss": 1.2182,
      "step": 410
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.782004921382612e-05,
      "loss": 1.3148,
      "step": 420
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.771581325963594e-05,
      "loss": 1.4185,
      "step": 430
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.760920128615591e-05,
      "loss": 1.409,
      "step": 440
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.750021860215241e-05,
      "loss": 1.2089,
      "step": 450
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.738887063444188e-05,
      "loss": 1.3114,
      "step": 460
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.727516292762058e-05,
      "loss": 1.2528,
      "step": 470
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.715910114378845e-05,
      "loss": 1.1718,
      "step": 480
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.704069106226727e-05,
      "loss": 1.2825,
      "step": 490
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.691993857931277e-05,
      "loss": 1.1629,
      "step": 500
    },
    {
      "epoch": 0.69,
      "learning_rate": 9.679684970782106e-05,
      "loss": 1.2463,
      "step": 510
    },
    {
      "epoch": 0.7,
      "learning_rate": 9.667143057702926e-05,
      "loss": 1.3297,
      "step": 520
    },
    {
      "epoch": 0.71,
      "learning_rate": 9.654368743221022e-05,
      "loss": 1.1995,
      "step": 530
    },
    {
      "epoch": 0.73,
      "learning_rate": 9.641362663436161e-05,
      "loss": 1.1135,
      "step": 540
    },
    {
      "epoch": 0.74,
      "learning_rate": 9.628125465988913e-05,
      "loss": 1.1311,
      "step": 550
    },
    {
      "epoch": 0.75,
      "learning_rate": 9.614657810028403e-05,
      "loss": 1.2915,
      "step": 560
    },
    {
      "epoch": 0.77,
      "learning_rate": 9.60096036617948e-05,
      "loss": 1.1898,
      "step": 570
    },
    {
      "epoch": 0.78,
      "learning_rate": 9.587033816509341e-05,
      "loss": 1.211,
      "step": 580
    },
    {
      "epoch": 0.8,
      "learning_rate": 9.572878854493553e-05,
      "loss": 1.1389,
      "step": 590
    },
    {
      "epoch": 0.81,
      "learning_rate": 9.558496184981525e-05,
      "loss": 1.0623,
      "step": 600
    },
    {
      "epoch": 0.82,
      "learning_rate": 9.543886524161409e-05,
      "loss": 1.1475,
      "step": 610
    },
    {
      "epoch": 0.84,
      "learning_rate": 9.529050599524443e-05,
      "loss": 1.0662,
      "step": 620
    },
    {
      "epoch": 0.85,
      "learning_rate": 9.513989149828718e-05,
      "loss": 1.2403,
      "step": 630
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.498702925062393e-05,
      "loss": 1.3078,
      "step": 640
    },
    {
      "epoch": 0.88,
      "learning_rate": 9.48319268640635e-05,
      "loss": 1.2456,
      "step": 650
    },
    {
      "epoch": 0.89,
      "learning_rate": 9.467459206196298e-05,
      "loss": 1.1248,
      "step": 660
    },
    {
      "epoch": 0.9,
      "learning_rate": 9.451503267884299e-05,
      "loss": 1.0899,
      "step": 670
    },
    {
      "epoch": 0.92,
      "learning_rate": 9.435325665999771e-05,
      "loss": 1.3189,
      "step": 680
    },
    {
      "epoch": 0.93,
      "learning_rate": 9.418927206109913e-05,
      "loss": 0.9791,
      "step": 690
    },
    {
      "epoch": 0.94,
      "learning_rate": 9.402308704779599e-05,
      "loss": 1.112,
      "step": 700
    },
    {
      "epoch": 0.96,
      "learning_rate": 9.385470989530716e-05,
      "loss": 1.0378,
      "step": 710
    },
    {
      "epoch": 0.97,
      "learning_rate": 9.368414898800952e-05,
      "loss": 1.2503,
      "step": 720
    },
    {
      "epoch": 0.98,
      "learning_rate": 9.35114128190205e-05,
      "loss": 1.0507,
      "step": 730
    },
    {
      "epoch": 1.0,
      "learning_rate": 9.333650998977518e-05,
      "loss": 1.0279,
      "step": 740
    },
    {
      "epoch": 1.01,
      "learning_rate": 9.315944920959798e-05,
      "loss": 0.9083,
      "step": 750
    },
    {
      "epoch": 1.02,
      "learning_rate": 9.298023929526888e-05,
      "loss": 1.0812,
      "step": 760
    },
    {
      "epoch": 1.04,
      "learning_rate": 9.279888917058452e-05,
      "loss": 1.0576,
      "step": 770
    },
    {
      "epoch": 1.05,
      "learning_rate": 9.261540786591377e-05,
      "loss": 1.2718,
      "step": 780
    },
    {
      "epoch": 1.06,
      "learning_rate": 9.242980451774807e-05,
      "loss": 1.2115,
      "step": 790
    },
    {
      "epoch": 1.08,
      "learning_rate": 9.224208836824647e-05,
      "loss": 1.0058,
      "step": 800
    },
    {
      "epoch": 1.09,
      "learning_rate": 9.20522687647754e-05,
      "loss": 1.1806,
      "step": 810
    },
    {
      "epoch": 1.11,
      "learning_rate": 9.18603551594433e-05,
      "loss": 1.1535,
      "step": 820
    },
    {
      "epoch": 1.12,
      "learning_rate": 9.166635710862983e-05,
      "loss": 1.0037,
      "step": 830
    },
    {
      "epoch": 1.13,
      "learning_rate": 9.14702842725101e-05,
      "loss": 0.9461,
      "step": 840
    },
    {
      "epoch": 1.15,
      "learning_rate": 9.127214641457358e-05,
      "loss": 1.1394,
      "step": 850
    },
    {
      "epoch": 1.16,
      "learning_rate": 9.107195340113794e-05,
      "loss": 0.9338,
      "step": 860
    },
    {
      "epoch": 1.17,
      "learning_rate": 9.086971520085783e-05,
      "loss": 1.0181,
      "step": 870
    },
    {
      "epoch": 1.19,
      "learning_rate": 9.066544188422836e-05,
      "loss": 0.9725,
      "step": 880
    },
    {
      "epoch": 1.2,
      "learning_rate": 9.04591436230837e-05,
      "loss": 1.0929,
      "step": 890
    },
    {
      "epoch": 1.21,
      "learning_rate": 9.025083069009063e-05,
      "loss": 1.1141,
      "step": 900
    },
    {
      "epoch": 1.23,
      "learning_rate": 9.004051345823689e-05,
      "loss": 1.1072,
      "step": 910
    },
    {
      "epoch": 1.24,
      "learning_rate": 8.982820240031474e-05,
      "loss": 1.0488,
      "step": 920
    },
    {
      "epoch": 1.25,
      "learning_rate": 8.961390808839945e-05,
      "loss": 1.0832,
      "step": 930
    },
    {
      "epoch": 1.27,
      "learning_rate": 8.939764119332284e-05,
      "loss": 0.9583,
      "step": 940
    },
    {
      "epoch": 1.28,
      "learning_rate": 8.917941248414196e-05,
      "loss": 1.0901,
      "step": 950
    },
    {
      "epoch": 1.29,
      "learning_rate": 8.895923282760279e-05,
      "loss": 1.2054,
      "step": 960
    },
    {
      "epoch": 1.31,
      "learning_rate": 8.873711318759919e-05,
      "loss": 1.1045,
      "step": 970
    },
    {
      "epoch": 1.32,
      "learning_rate": 8.851306462462688e-05,
      "loss": 1.2223,
      "step": 980
    },
    {
      "epoch": 1.33,
      "learning_rate": 8.828709829523276e-05,
      "loss": 1.1103,
      "step": 990
    },
    {
      "epoch": 1.35,
      "learning_rate": 8.805922545145928e-05,
      "loss": 1.1602,
      "step": 1000
    },
    {
      "epoch": 1.36,
      "learning_rate": 8.782945744028424e-05,
      "loss": 1.0639,
      "step": 1010
    },
    {
      "epoch": 1.37,
      "learning_rate": 8.759780570305565e-05,
      "loss": 1.2104,
      "step": 1020
    },
    {
      "epoch": 1.39,
      "learning_rate": 8.736428177492213e-05,
      "loss": 0.85,
      "step": 1030
    },
    {
      "epoch": 1.4,
      "learning_rate": 8.712889728425842e-05,
      "loss": 0.9958,
      "step": 1040
    },
    {
      "epoch": 1.42,
      "learning_rate": 8.689166395208636e-05,
      "loss": 0.9216,
      "step": 1050
    },
    {
      "epoch": 1.43,
      "learning_rate": 8.665259359149132e-05,
      "loss": 0.9329,
      "step": 1060
    },
    {
      "epoch": 1.44,
      "learning_rate": 8.641169810703382e-05,
      "loss": 1.0304,
      "step": 1070
    },
    {
      "epoch": 1.46,
      "learning_rate": 8.616898949415686e-05,
      "loss": 1.0397,
      "step": 1080
    },
    {
      "epoch": 1.47,
      "learning_rate": 8.592447983858858e-05,
      "loss": 1.0056,
      "step": 1090
    },
    {
      "epoch": 1.48,
      "learning_rate": 8.567818131574039e-05,
      "loss": 0.9399,
      "step": 1100
    },
    {
      "epoch": 1.5,
      "learning_rate": 8.543010619010077e-05,
      "loss": 1.0151,
      "step": 1110
    },
    {
      "epoch": 1.51,
      "learning_rate": 8.518026681462448e-05,
      "loss": 0.8214,
      "step": 1120
    },
    {
      "epoch": 1.52,
      "learning_rate": 8.492867563011752e-05,
      "loss": 0.8622,
      "step": 1130
    },
    {
      "epoch": 1.54,
      "learning_rate": 8.467534516461763e-05,
      "loss": 0.962,
      "step": 1140
    },
    {
      "epoch": 1.55,
      "learning_rate": 8.442028803277035e-05,
      "loss": 1.2468,
      "step": 1150
    },
    {
      "epoch": 1.56,
      "learning_rate": 8.4163516935201e-05,
      "loss": 0.762,
      "step": 1160
    },
    {
      "epoch": 1.58,
      "learning_rate": 8.390504465788216e-05,
      "loss": 1.1067,
      "step": 1170
    },
    {
      "epoch": 1.59,
      "learning_rate": 8.364488407149707e-05,
      "loss": 1.0618,
      "step": 1180
    },
    {
      "epoch": 1.6,
      "learning_rate": 8.338304813079865e-05,
      "loss": 0.8968,
      "step": 1190
    },
    {
      "epoch": 1.62,
      "learning_rate": 8.311954987396444e-05,
      "loss": 0.9637,
      "step": 1200
    },
    {
      "epoch": 1.63,
      "learning_rate": 8.285440242194744e-05,
      "loss": 0.8944,
      "step": 1210
    },
    {
      "epoch": 1.64,
      "learning_rate": 8.258761897782264e-05,
      "loss": 0.9935,
      "step": 1220
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.231921282612957e-05,
      "loss": 1.0193,
      "step": 1230
    },
    {
      "epoch": 1.67,
      "learning_rate": 8.204919733221089e-05,
      "loss": 0.9402,
      "step": 1240
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.177758594154681e-05,
      "loss": 0.7653,
      "step": 1250
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.150439217908556e-05,
      "loss": 0.9966,
      "step": 1260
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.122962964856989e-05,
      "loss": 0.8538,
      "step": 1270
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.095331203185968e-05,
      "loss": 1.078,
      "step": 1280
    },
    {
      "epoch": 1.74,
      "learning_rate": 8.067545308825075e-05,
      "loss": 0.8603,
      "step": 1290
    },
    {
      "epoch": 1.75,
      "learning_rate": 8.039606665378956e-05,
      "loss": 0.6651,
      "step": 1300
    },
    {
      "epoch": 1.77,
      "learning_rate": 8.011516664058431e-05,
      "loss": 0.7974,
      "step": 1310
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.983276703611222e-05,
      "loss": 0.8603,
      "step": 1320
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.954888190252292e-05,
      "loss": 0.8481,
      "step": 1330
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.926352537593837e-05,
      "loss": 0.8794,
      "step": 1340
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.897671166574878e-05,
      "loss": 0.6893,
      "step": 1350
    },
    {
      "epoch": 1.83,
      "learning_rate": 7.86884550539052e-05,
      "loss": 1.076,
      "step": 1360
    },
    {
      "epoch": 1.85,
      "learning_rate": 7.839876989420827e-05,
      "loss": 0.8875,
      "step": 1370
    },
    {
      "epoch": 1.86,
      "learning_rate": 7.810767061159341e-05,
      "loss": 0.7543,
      "step": 1380
    },
    {
      "epoch": 1.87,
      "learning_rate": 7.78151717014127e-05,
      "loss": 0.9026,
      "step": 1390
    },
    {
      "epoch": 1.89,
      "learning_rate": 7.752128772871292e-05,
      "loss": 0.7245,
      "step": 1400
    },
    {
      "epoch": 1.9,
      "learning_rate": 7.722603332751034e-05,
      "loss": 1.0003,
      "step": 1410
    },
    {
      "epoch": 1.91,
      "learning_rate": 7.692942320006204e-05,
      "loss": 1.08,
      "step": 1420
    },
    {
      "epoch": 1.93,
      "learning_rate": 7.663147211613378e-05,
      "loss": 0.9045,
      "step": 1430
    },
    {
      "epoch": 1.94,
      "learning_rate": 7.633219491226448e-05,
      "loss": 0.9683,
      "step": 1440
    },
    {
      "epoch": 1.95,
      "learning_rate": 7.60316064910276e-05,
      "loss": 1.0975,
      "step": 1450
    },
    {
      "epoch": 1.97,
      "learning_rate": 7.572972182028885e-05,
      "loss": 0.8337,
      "step": 1460
    },
    {
      "epoch": 1.98,
      "learning_rate": 7.542655593246103e-05,
      "loss": 0.944,
      "step": 1470
    },
    {
      "epoch": 1.99,
      "learning_rate": 7.512212392375539e-05,
      "loss": 0.9337,
      "step": 1480
    }
  ],
  "logging_steps": 10,
  "max_steps": 4452,
  "num_train_epochs": 6,
  "save_steps": 500,
  "total_flos": 7755360321208320.0,
  "trial_name": null,
  "trial_params": null
}