{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 9.999401161746212,
  "global_step": 83490,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.06,
      "learning_rate": 2.9826326506168404e-06,
      "loss": 1.509,
      "step": 500
    },
    {
      "epoch": 0.12,
      "learning_rate": 2.9652653012336808e-06,
      "loss": 1.5092,
      "step": 1000
    },
    {
      "epoch": 0.18,
      "learning_rate": 2.947897951850521e-06,
      "loss": 1.5089,
      "step": 1500
    },
    {
      "epoch": 0.24,
      "learning_rate": 2.9305306024673614e-06,
      "loss": 1.5082,
      "step": 2000
    },
    {
      "epoch": 0.3,
      "learning_rate": 2.9131632530842018e-06,
      "loss": 1.5086,
      "step": 2500
    },
    {
      "epoch": 0.36,
      "learning_rate": 2.895795903701042e-06,
      "loss": 1.5082,
      "step": 3000
    },
    {
      "epoch": 0.42,
      "learning_rate": 2.8784285543178824e-06,
      "loss": 1.5077,
      "step": 3500
    },
    {
      "epoch": 0.48,
      "learning_rate": 2.8610612049347228e-06,
      "loss": 1.5077,
      "step": 4000
    },
    {
      "epoch": 0.54,
      "learning_rate": 2.843693855551563e-06,
      "loss": 1.5067,
      "step": 4500
    },
    {
      "epoch": 0.6,
      "learning_rate": 2.8263265061684034e-06,
      "loss": 1.507,
      "step": 5000
    },
    {
      "epoch": 0.66,
      "learning_rate": 2.8089591567852438e-06,
      "loss": 1.5067,
      "step": 5500
    },
    {
      "epoch": 0.72,
      "learning_rate": 2.791591807402084e-06,
      "loss": 1.5067,
      "step": 6000
    },
    {
      "epoch": 0.78,
      "learning_rate": 2.7742244580189245e-06,
      "loss": 1.5061,
      "step": 6500
    },
    {
      "epoch": 0.84,
      "learning_rate": 2.756857108635765e-06,
      "loss": 1.5054,
      "step": 7000
    },
    {
      "epoch": 0.9,
      "learning_rate": 2.739489759252605e-06,
      "loss": 1.5058,
      "step": 7500
    },
    {
      "epoch": 0.96,
      "learning_rate": 2.7221224098694455e-06,
      "loss": 1.5052,
      "step": 8000
    },
    {
      "epoch": 1.0,
      "eval_loss": 1.5115793943405151,
      "eval_runtime": 99.986,
      "eval_samples_per_second": 300.042,
      "eval_steps_per_second": 1.88,
      "step": 8349
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.704755060486286e-06,
      "loss": 1.5046,
      "step": 8500
    },
    {
      "epoch": 1.08,
      "learning_rate": 2.687387711103126e-06,
      "loss": 1.5045,
      "step": 9000
    },
    {
      "epoch": 1.14,
      "learning_rate": 2.6700203617199665e-06,
      "loss": 1.5038,
      "step": 9500
    },
    {
      "epoch": 1.2,
      "learning_rate": 2.652653012336807e-06,
      "loss": 1.5042,
      "step": 10000
    },
    {
      "epoch": 1.26,
      "learning_rate": 2.635285662953647e-06,
      "loss": 1.5039,
      "step": 10500
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.6179183135704875e-06,
      "loss": 1.5033,
      "step": 11000
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.600550964187328e-06,
      "loss": 1.5028,
      "step": 11500
    },
    {
      "epoch": 1.44,
      "learning_rate": 2.583183614804168e-06,
      "loss": 1.5036,
      "step": 12000
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.5658162654210085e-06,
      "loss": 1.5032,
      "step": 12500
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.548448916037849e-06,
      "loss": 1.5029,
      "step": 13000
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.5310815666546896e-06,
      "loss": 1.5028,
      "step": 13500
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.5137142172715295e-06,
      "loss": 1.5024,
      "step": 14000
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.49634686788837e-06,
      "loss": 1.5022,
      "step": 14500
    },
    {
      "epoch": 1.8,
      "learning_rate": 2.47897951850521e-06,
      "loss": 1.5022,
      "step": 15000
    },
    {
      "epoch": 1.86,
      "learning_rate": 2.4616121691220505e-06,
      "loss": 1.5019,
      "step": 15500
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.444244819738891e-06,
      "loss": 1.502,
      "step": 16000
    },
    {
      "epoch": 1.98,
      "learning_rate": 2.426877470355731e-06,
      "loss": 1.5019,
      "step": 16500
    },
    {
      "epoch": 2.0,
      "eval_loss": 1.509125828742981,
      "eval_runtime": 102.1448,
      "eval_samples_per_second": 293.701,
      "eval_steps_per_second": 1.841,
      "step": 16699
    },
    {
      "epoch": 2.04,
      "learning_rate": 2.4095101209725715e-06,
      "loss": 1.501,
      "step": 17000
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.392142771589412e-06,
      "loss": 1.5004,
      "step": 17500
    },
    {
      "epoch": 2.16,
      "learning_rate": 2.374775422206252e-06,
      "loss": 1.5009,
      "step": 18000
    },
    {
      "epoch": 2.22,
      "learning_rate": 2.3574080728230925e-06,
      "loss": 1.5004,
      "step": 18500
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.340040723439933e-06,
      "loss": 1.5004,
      "step": 19000
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.3226733740567732e-06,
      "loss": 1.5001,
      "step": 19500
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.3053060246736136e-06,
      "loss": 1.4998,
      "step": 20000
    },
    {
      "epoch": 2.46,
      "learning_rate": 2.2879386752904543e-06,
      "loss": 1.4998,
      "step": 20500
    },
    {
      "epoch": 2.52,
      "learning_rate": 2.2705713259072942e-06,
      "loss": 1.4998,
      "step": 21000
    },
    {
      "epoch": 2.58,
      "learning_rate": 2.2532039765241346e-06,
      "loss": 1.4997,
      "step": 21500
    },
    {
      "epoch": 2.63,
      "learning_rate": 2.235836627140975e-06,
      "loss": 1.4992,
      "step": 22000
    },
    {
      "epoch": 2.69,
      "learning_rate": 2.2184692777578152e-06,
      "loss": 1.4991,
      "step": 22500
    },
    {
      "epoch": 2.75,
      "learning_rate": 2.2011019283746556e-06,
      "loss": 1.499,
      "step": 23000
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.183734578991496e-06,
      "loss": 1.4987,
      "step": 23500
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.1663672296083363e-06,
      "loss": 1.4991,
      "step": 24000
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.1489998802251766e-06,
      "loss": 1.4986,
      "step": 24500
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.131632530842017e-06,
      "loss": 1.4989,
      "step": 25000
    },
    {
      "epoch": 3.0,
      "eval_loss": 1.5069462060928345,
      "eval_runtime": 101.4916,
      "eval_samples_per_second": 295.591,
      "eval_steps_per_second": 1.852,
      "step": 25048
    },
    {
      "epoch": 3.05,
      "learning_rate": 2.1142651814588573e-06,
      "loss": 1.498,
      "step": 25500
    },
    {
      "epoch": 3.11,
      "learning_rate": 2.096897832075698e-06,
      "loss": 1.4977,
      "step": 26000
    },
    {
      "epoch": 3.17,
      "learning_rate": 2.079530482692538e-06,
      "loss": 1.4975,
      "step": 26500
    },
    {
      "epoch": 3.23,
      "learning_rate": 2.0621631333093783e-06,
      "loss": 1.4974,
      "step": 27000
    },
    {
      "epoch": 3.29,
      "learning_rate": 2.044795783926219e-06,
      "loss": 1.4976,
      "step": 27500
    },
    {
      "epoch": 3.35,
      "learning_rate": 2.027428434543059e-06,
      "loss": 1.4978,
      "step": 28000
    },
    {
      "epoch": 3.41,
      "learning_rate": 2.0100610851598993e-06,
      "loss": 1.4971,
      "step": 28500
    },
    {
      "epoch": 3.47,
      "learning_rate": 1.99269373577674e-06,
      "loss": 1.4974,
      "step": 29000
    },
    {
      "epoch": 3.53,
      "learning_rate": 1.97532638639358e-06,
      "loss": 1.4972,
      "step": 29500
    },
    {
      "epoch": 3.59,
      "learning_rate": 1.9579590370104203e-06,
      "loss": 1.4969,
      "step": 30000
    },
    {
      "epoch": 3.65,
      "learning_rate": 1.940591687627261e-06,
      "loss": 1.4967,
      "step": 30500
    },
    {
      "epoch": 3.71,
      "learning_rate": 1.923224338244101e-06,
      "loss": 1.4969,
      "step": 31000
    },
    {
      "epoch": 3.77,
      "learning_rate": 1.9058569888609415e-06,
      "loss": 1.497,
      "step": 31500
    },
    {
      "epoch": 3.83,
      "learning_rate": 1.888489639477782e-06,
      "loss": 1.4967,
      "step": 32000
    },
    {
      "epoch": 3.89,
      "learning_rate": 1.871122290094622e-06,
      "loss": 1.4965,
      "step": 32500
    },
    {
      "epoch": 3.95,
      "learning_rate": 1.8537549407114627e-06,
      "loss": 1.4969,
      "step": 33000
    },
    {
      "epoch": 4.0,
      "eval_loss": 1.5056320428848267,
      "eval_runtime": 101.7674,
      "eval_samples_per_second": 294.79,
      "eval_steps_per_second": 1.847,
      "step": 33398
    },
    {
      "epoch": 4.01,
      "learning_rate": 1.8363875913283029e-06,
      "loss": 1.4962,
      "step": 33500
    },
    {
      "epoch": 4.07,
      "learning_rate": 1.8190202419451432e-06,
      "loss": 1.4956,
      "step": 34000
    },
    {
      "epoch": 4.13,
      "learning_rate": 1.8016528925619838e-06,
      "loss": 1.4955,
      "step": 34500
    },
    {
      "epoch": 4.19,
      "learning_rate": 1.7842855431788235e-06,
      "loss": 1.4955,
      "step": 35000
    },
    {
      "epoch": 4.25,
      "learning_rate": 1.7669181937956644e-06,
      "loss": 1.4953,
      "step": 35500
    },
    {
      "epoch": 4.31,
      "learning_rate": 1.7495508444125048e-06,
      "loss": 1.4954,
      "step": 36000
    },
    {
      "epoch": 4.37,
      "learning_rate": 1.7321834950293447e-06,
      "loss": 1.4952,
      "step": 36500
    },
    {
      "epoch": 4.43,
      "learning_rate": 1.714816145646185e-06,
      "loss": 1.4956,
      "step": 37000
    },
    {
      "epoch": 4.49,
      "learning_rate": 1.6974487962630258e-06,
      "loss": 1.4953,
      "step": 37500
    },
    {
      "epoch": 4.55,
      "learning_rate": 1.6800814468798657e-06,
      "loss": 1.4951,
      "step": 38000
    },
    {
      "epoch": 4.61,
      "learning_rate": 1.6627140974967064e-06,
      "loss": 1.4953,
      "step": 38500
    },
    {
      "epoch": 4.67,
      "learning_rate": 1.645346748113547e-06,
      "loss": 1.4954,
      "step": 39000
    },
    {
      "epoch": 4.73,
      "learning_rate": 1.6279793987303871e-06,
      "loss": 1.495,
      "step": 39500
    },
    {
      "epoch": 4.79,
      "learning_rate": 1.6106120493472275e-06,
      "loss": 1.4951,
      "step": 40000
    },
    {
      "epoch": 4.85,
      "learning_rate": 1.5932446999640674e-06,
      "loss": 1.4946,
      "step": 40500
    },
    {
      "epoch": 4.91,
      "learning_rate": 1.5758773505809081e-06,
      "loss": 1.4948,
      "step": 41000
    },
    {
      "epoch": 4.97,
      "learning_rate": 1.5585100011977485e-06,
      "loss": 1.4946,
      "step": 41500
    },
    {
      "epoch": 5.0,
      "eval_loss": 1.504227638244629,
      "eval_runtime": 101.5567,
      "eval_samples_per_second": 295.402,
      "eval_steps_per_second": 1.851,
      "step": 41747
    },
    {
      "epoch": 5.03,
      "learning_rate": 1.5411426518145884e-06,
      "loss": 1.4944,
      "step": 42000
    },
    {
      "epoch": 5.09,
      "learning_rate": 1.5237753024314291e-06,
      "loss": 1.494,
      "step": 42500
    },
    {
      "epoch": 5.15,
      "learning_rate": 1.5064079530482697e-06,
      "loss": 1.4942,
      "step": 43000
    },
    {
      "epoch": 5.21,
      "learning_rate": 1.4890406036651098e-06,
      "loss": 1.4938,
      "step": 43500
    },
    {
      "epoch": 5.27,
      "learning_rate": 1.4716732542819497e-06,
      "loss": 1.4939,
      "step": 44000
    },
    {
      "epoch": 5.33,
      "learning_rate": 1.4543059048987905e-06,
      "loss": 1.4938,
      "step": 44500
    },
    {
      "epoch": 5.39,
      "learning_rate": 1.4369385555156308e-06,
      "loss": 1.4937,
      "step": 45000
    },
    {
      "epoch": 5.45,
      "learning_rate": 1.4195712061324707e-06,
      "loss": 1.494,
      "step": 45500
    },
    {
      "epoch": 5.51,
      "learning_rate": 1.4022038567493115e-06,
      "loss": 1.4936,
      "step": 46000
    },
    {
      "epoch": 5.57,
      "learning_rate": 1.3848365073661518e-06,
      "loss": 1.4936,
      "step": 46500
    },
    {
      "epoch": 5.63,
      "learning_rate": 1.3674691579829922e-06,
      "loss": 1.4934,
      "step": 47000
    },
    {
      "epoch": 5.69,
      "learning_rate": 1.350101808599832e-06,
      "loss": 1.4936,
      "step": 47500
    },
    {
      "epoch": 5.75,
      "learning_rate": 1.3327344592166729e-06,
      "loss": 1.4933,
      "step": 48000
    },
    {
      "epoch": 5.81,
      "learning_rate": 1.3153671098335132e-06,
      "loss": 1.4934,
      "step": 48500
    },
    {
      "epoch": 5.87,
      "learning_rate": 1.297999760450353e-06,
      "loss": 1.493,
      "step": 49000
    },
    {
      "epoch": 5.93,
      "learning_rate": 1.2806324110671939e-06,
      "loss": 1.4932,
      "step": 49500
    },
    {
      "epoch": 5.99,
      "learning_rate": 1.2632650616840342e-06,
      "loss": 1.4933,
      "step": 50000
    },
    {
      "epoch": 6.0,
      "eval_loss": 1.5031880140304565,
      "eval_runtime": 101.923,
      "eval_samples_per_second": 294.34,
      "eval_steps_per_second": 1.845,
      "step": 50097
    },
    {
      "epoch": 6.05,
      "learning_rate": 1.2458977123008745e-06,
      "loss": 1.4928,
      "step": 50500
    },
    {
      "epoch": 6.11,
      "learning_rate": 1.2285303629177145e-06,
      "loss": 1.4928,
      "step": 51000
    },
    {
      "epoch": 6.17,
      "learning_rate": 1.2111630135345552e-06,
      "loss": 1.4927,
      "step": 51500
    },
    {
      "epoch": 6.23,
      "learning_rate": 1.1937956641513955e-06,
      "loss": 1.4926,
      "step": 52000
    },
    {
      "epoch": 6.29,
      "learning_rate": 1.1764283147682355e-06,
      "loss": 1.4929,
      "step": 52500
    },
    {
      "epoch": 6.35,
      "learning_rate": 1.1590609653850762e-06,
      "loss": 1.4927,
      "step": 53000
    },
    {
      "epoch": 6.41,
      "learning_rate": 1.1416936160019166e-06,
      "loss": 1.4924,
      "step": 53500
    },
    {
      "epoch": 6.47,
      "learning_rate": 1.124326266618757e-06,
      "loss": 1.4924,
      "step": 54000
    },
    {
      "epoch": 6.53,
      "learning_rate": 1.1069589172355968e-06,
      "loss": 1.4921,
      "step": 54500
    },
    {
      "epoch": 6.59,
      "learning_rate": 1.0895915678524376e-06,
      "loss": 1.4922,
      "step": 55000
    },
    {
      "epoch": 6.65,
      "learning_rate": 1.072224218469278e-06,
      "loss": 1.4925,
      "step": 55500
    },
    {
      "epoch": 6.71,
      "learning_rate": 1.0548568690861182e-06,
      "loss": 1.4925,
      "step": 56000
    },
    {
      "epoch": 6.77,
      "learning_rate": 1.0374895197029584e-06,
      "loss": 1.4922,
      "step": 56500
    },
    {
      "epoch": 6.83,
      "learning_rate": 1.020122170319799e-06,
      "loss": 1.4924,
      "step": 57000
    },
    {
      "epoch": 6.89,
      "learning_rate": 1.0027548209366393e-06,
      "loss": 1.492,
      "step": 57500
    },
    {
      "epoch": 6.95,
      "learning_rate": 9.853874715534794e-07,
      "loss": 1.4923,
      "step": 58000
    },
    {
      "epoch": 7.0,
      "eval_loss": 1.502358317375183,
      "eval_runtime": 100.8628,
      "eval_samples_per_second": 297.434,
      "eval_steps_per_second": 1.864,
      "step": 58446
    },
    {
      "epoch": 7.01,
      "learning_rate": 9.6802012217032e-07,
      "loss": 1.4919,
      "step": 58500
    },
    {
      "epoch": 7.07,
      "learning_rate": 9.506527727871603e-07,
      "loss": 1.4919,
      "step": 59000
    },
    {
      "epoch": 7.13,
      "learning_rate": 9.332854234040007e-07,
      "loss": 1.4917,
      "step": 59500
    },
    {
      "epoch": 7.19,
      "learning_rate": 9.159180740208407e-07,
      "loss": 1.4917,
      "step": 60000
    },
    {
      "epoch": 7.25,
      "learning_rate": 8.985507246376812e-07,
      "loss": 1.4914,
      "step": 60500
    },
    {
      "epoch": 7.31,
      "learning_rate": 8.811833752545216e-07,
      "loss": 1.4917,
      "step": 61000
    },
    {
      "epoch": 7.37,
      "learning_rate": 8.638160258713617e-07,
      "loss": 1.4917,
      "step": 61500
    },
    {
      "epoch": 7.43,
      "learning_rate": 8.464486764882022e-07,
      "loss": 1.4914,
      "step": 62000
    },
    {
      "epoch": 7.49,
      "learning_rate": 8.290813271050427e-07,
      "loss": 1.4917,
      "step": 62500
    },
    {
      "epoch": 7.55,
      "learning_rate": 8.11713977721883e-07,
      "loss": 1.4914,
      "step": 63000
    },
    {
      "epoch": 7.61,
      "learning_rate": 7.943466283387231e-07,
      "loss": 1.4915,
      "step": 63500
    },
    {
      "epoch": 7.67,
      "learning_rate": 7.769792789555636e-07,
      "loss": 1.4915,
      "step": 64000
    },
    {
      "epoch": 7.73,
      "learning_rate": 7.596119295724041e-07,
      "loss": 1.4912,
      "step": 64500
    },
    {
      "epoch": 7.78,
      "learning_rate": 7.422445801892441e-07,
      "loss": 1.4913,
      "step": 65000
    },
    {
      "epoch": 7.84,
      "learning_rate": 7.248772308060845e-07,
      "loss": 1.4913,
      "step": 65500
    },
    {
      "epoch": 7.9,
      "learning_rate": 7.07509881422925e-07,
      "loss": 1.4912,
      "step": 66000
    },
    {
      "epoch": 7.96,
      "learning_rate": 6.901425320397654e-07,
      "loss": 1.4914,
      "step": 66500
    },
    {
      "epoch": 8.0,
      "eval_loss": 1.5018346309661865,
      "eval_runtime": 100.3999,
      "eval_samples_per_second": 298.805,
      "eval_steps_per_second": 1.873,
      "step": 66796
    },
    {
      "epoch": 8.02,
      "learning_rate": 6.727751826566056e-07,
      "loss": 1.4912,
      "step": 67000
    },
    {
      "epoch": 8.08,
      "learning_rate": 6.554078332734459e-07,
      "loss": 1.4909,
      "step": 67500
    },
    {
      "epoch": 8.14,
      "learning_rate": 6.380404838902863e-07,
      "loss": 1.4912,
      "step": 68000
    },
    {
      "epoch": 8.2,
      "learning_rate": 6.206731345071265e-07,
      "loss": 1.4909,
      "step": 68500
    },
    {
      "epoch": 8.26,
      "learning_rate": 6.033057851239669e-07,
      "loss": 1.4911,
      "step": 69000
    },
    {
      "epoch": 8.32,
      "learning_rate": 5.859384357408073e-07,
      "loss": 1.4908,
      "step": 69500
    },
    {
      "epoch": 8.38,
      "learning_rate": 5.685710863576478e-07,
      "loss": 1.491,
      "step": 70000
    },
    {
      "epoch": 8.44,
      "learning_rate": 5.512037369744878e-07,
      "loss": 1.491,
      "step": 70500
    },
    {
      "epoch": 8.5,
      "learning_rate": 5.338363875913284e-07,
      "loss": 1.491,
      "step": 71000
    },
    {
      "epoch": 8.56,
      "learning_rate": 5.164690382081687e-07,
      "loss": 1.4911,
      "step": 71500
    },
    {
      "epoch": 8.62,
      "learning_rate": 4.991016888250091e-07,
      "loss": 1.4908,
      "step": 72000
    },
    {
      "epoch": 8.68,
      "learning_rate": 4.817343394418493e-07,
      "loss": 1.4907,
      "step": 72500
    },
    {
      "epoch": 8.74,
      "learning_rate": 4.643669900586897e-07,
      "loss": 1.4907,
      "step": 73000
    },
    {
      "epoch": 8.8,
      "learning_rate": 4.4699964067553015e-07,
      "loss": 1.4906,
      "step": 73500
    },
    {
      "epoch": 8.86,
      "learning_rate": 4.296322912923702e-07,
      "loss": 1.4907,
      "step": 74000
    },
    {
      "epoch": 8.92,
      "learning_rate": 4.1226494190921066e-07,
      "loss": 1.491,
      "step": 74500
    },
    {
      "epoch": 8.98,
      "learning_rate": 3.948975925260511e-07,
      "loss": 1.4907,
      "step": 75000
    },
    {
      "epoch": 9.0,
      "eval_loss": 1.5016120672225952,
      "eval_runtime": 100.2408,
      "eval_samples_per_second": 299.279,
      "eval_steps_per_second": 1.875,
      "step": 75145
    },
    {
      "epoch": 9.04,
      "learning_rate": 3.775302431428916e-07,
      "loss": 1.4908,
      "step": 75500
    },
    {
      "epoch": 9.1,
      "learning_rate": 3.601628937597316e-07,
      "loss": 1.4906,
      "step": 76000
    },
    {
      "epoch": 9.16,
      "learning_rate": 3.4279554437657206e-07,
      "loss": 1.4906,
      "step": 76500
    },
    {
      "epoch": 9.22,
      "learning_rate": 3.254281949934125e-07,
      "loss": 1.4903,
      "step": 77000
    },
    {
      "epoch": 9.28,
      "learning_rate": 3.0806084561025263e-07,
      "loss": 1.4905,
      "step": 77500
    },
    {
      "epoch": 9.34,
      "learning_rate": 2.90693496227093e-07,
      "loss": 1.4907,
      "step": 78000
    },
    {
      "epoch": 9.4,
      "learning_rate": 2.7332614684393346e-07,
      "loss": 1.4905,
      "step": 78500
    },
    {
      "epoch": 9.46,
      "learning_rate": 2.5595879746077385e-07,
      "loss": 1.4904,
      "step": 79000
    },
    {
      "epoch": 9.52,
      "learning_rate": 2.38591448077614e-07,
      "loss": 1.4905,
      "step": 79500
    },
    {
      "epoch": 9.58,
      "learning_rate": 2.2122409869445442e-07,
      "loss": 1.4906,
      "step": 80000
    },
    {
      "epoch": 9.64,
      "learning_rate": 2.0385674931129484e-07,
      "loss": 1.4904,
      "step": 80500
    },
    {
      "epoch": 9.7,
      "learning_rate": 1.8648939992813496e-07,
      "loss": 1.4903,
      "step": 81000
    },
    {
      "epoch": 9.76,
      "learning_rate": 1.6912205054497538e-07,
      "loss": 1.4906,
      "step": 81500
    },
    {
      "epoch": 9.82,
      "learning_rate": 1.5175470116181582e-07,
      "loss": 1.4908,
      "step": 82000
    },
    {
      "epoch": 9.88,
      "learning_rate": 1.3438735177865624e-07,
      "loss": 1.49,
      "step": 82500
    },
    {
      "epoch": 9.94,
      "learning_rate": 1.1702000239549635e-07,
      "loss": 1.4905,
      "step": 83000
    },
    {
      "epoch": 10.0,
      "eval_loss": 1.5012842416763306,
      "eval_runtime": 100.3947,
      "eval_samples_per_second": 298.821,
      "eval_steps_per_second": 1.873,
      "step": 83490
    },
    {
      "epoch": 10.0,
      "step": 83490,
      "total_flos": 1.463646111228887e+18,
      "train_loss": 1.4961074900378695,
      "train_runtime": 199226.4751,
      "train_samples_per_second": 268.221,
      "train_steps_per_second": 0.419
    }
  ],
  "max_steps": 83490,
  "num_train_epochs": 10,
  "total_flos": 1.463646111228887e+18,
  "trial_name": null,
  "trial_params": null
}