{
  "best_metric": 2.9959683418273926,
  "best_model_checkpoint": "./model_tweets_2020_Q2_75/checkpoint-192000",
  "epoch": 20.214101019969846,
  "eval_steps": 8000,
  "global_step": 2400000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.07,
      "eval_loss": 3.1519558429718018,
      "eval_runtime": 113.5135,
      "eval_samples_per_second": 880.794,
      "eval_steps_per_second": 55.051,
      "step": 8000
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.939131159843243e-06,
      "loss": 3.3704,
      "step": 16000
    },
    {
      "epoch": 0.13,
      "eval_loss": 3.1285579204559326,
      "eval_runtime": 113.7953,
      "eval_samples_per_second": 878.613,
      "eval_steps_per_second": 54.914,
      "step": 16000
    },
    {
      "epoch": 0.2,
      "eval_loss": 3.107924699783325,
      "eval_runtime": 113.2981,
      "eval_samples_per_second": 882.468,
      "eval_steps_per_second": 55.155,
      "step": 24000
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.872425581589261e-06,
      "loss": 3.2908,
      "step": 32000
    },
    {
      "epoch": 0.27,
      "eval_loss": 3.084472179412842,
      "eval_runtime": 112.7537,
      "eval_samples_per_second": 886.73,
      "eval_steps_per_second": 55.422,
      "step": 32000
    },
    {
      "epoch": 0.34,
      "eval_loss": 3.0868372917175293,
      "eval_runtime": 113.2037,
      "eval_samples_per_second": 883.204,
      "eval_steps_per_second": 55.201,
      "step": 40000
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.80572000333528e-06,
      "loss": 3.2742,
      "step": 48000
    },
    {
      "epoch": 0.4,
      "eval_loss": 3.076770067214966,
      "eval_runtime": 112.4863,
      "eval_samples_per_second": 888.837,
      "eval_steps_per_second": 55.553,
      "step": 48000
    },
    {
      "epoch": 0.47,
      "eval_loss": 3.0705816745758057,
      "eval_runtime": 113.9358,
      "eval_samples_per_second": 877.529,
      "eval_steps_per_second": 54.847,
      "step": 56000
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.739014425081299e-06,
      "loss": 3.2579,
      "step": 64000
    },
    {
      "epoch": 0.54,
      "eval_loss": 3.0620689392089844,
      "eval_runtime": 112.5706,
      "eval_samples_per_second": 888.172,
      "eval_steps_per_second": 55.512,
      "step": 64000
    },
    {
      "epoch": 0.61,
      "eval_loss": 3.0659217834472656,
      "eval_runtime": 112.8069,
      "eval_samples_per_second": 886.311,
      "eval_steps_per_second": 55.396,
      "step": 72000
    },
    {
      "epoch": 0.67,
      "learning_rate": 9.672308846827316e-06,
      "loss": 3.2448,
      "step": 80000
    },
    {
      "epoch": 0.67,
      "eval_loss": 3.0456507205963135,
      "eval_runtime": 114.4983,
      "eval_samples_per_second": 873.218,
      "eval_steps_per_second": 54.577,
      "step": 80000
    },
    {
      "epoch": 0.74,
      "eval_loss": 3.055419683456421,
      "eval_runtime": 113.1402,
      "eval_samples_per_second": 883.7,
      "eval_steps_per_second": 55.232,
      "step": 88000
    },
    {
      "epoch": 0.81,
      "learning_rate": 9.605603268573334e-06,
      "loss": 3.2416,
      "step": 96000
    },
    {
      "epoch": 0.81,
      "eval_loss": 3.0335144996643066,
      "eval_runtime": 115.0306,
      "eval_samples_per_second": 869.178,
      "eval_steps_per_second": 54.325,
      "step": 96000
    },
    {
      "epoch": 0.88,
      "eval_loss": 3.0320653915405273,
      "eval_runtime": 115.4623,
      "eval_samples_per_second": 865.928,
      "eval_steps_per_second": 54.122,
      "step": 104000
    },
    {
      "epoch": 0.94,
      "learning_rate": 9.538897690319354e-06,
      "loss": 3.23,
      "step": 112000
    },
    {
      "epoch": 0.94,
      "eval_loss": 3.0136971473693848,
      "eval_runtime": 114.4021,
      "eval_samples_per_second": 873.952,
      "eval_steps_per_second": 54.623,
      "step": 112000
    },
    {
      "epoch": 1.01,
      "eval_loss": 3.0060510635375977,
      "eval_runtime": 114.4587,
      "eval_samples_per_second": 873.52,
      "eval_steps_per_second": 54.596,
      "step": 120000
    },
    {
      "epoch": 1.08,
      "learning_rate": 9.472192112065373e-06,
      "loss": 3.2084,
      "step": 128000
    },
    {
      "epoch": 1.08,
      "eval_loss": 3.025130033493042,
      "eval_runtime": 113.0671,
      "eval_samples_per_second": 884.272,
      "eval_steps_per_second": 55.268,
      "step": 128000
    },
    {
      "epoch": 1.15,
      "eval_loss": 3.009242296218872,
      "eval_runtime": 112.796,
      "eval_samples_per_second": 886.397,
      "eval_steps_per_second": 55.401,
      "step": 136000
    },
    {
      "epoch": 1.21,
      "learning_rate": 9.405486533811392e-06,
      "loss": 3.2055,
      "step": 144000
    },
    {
      "epoch": 1.21,
      "eval_loss": 3.0042872428894043,
      "eval_runtime": 112.6772,
      "eval_samples_per_second": 887.332,
      "eval_steps_per_second": 55.459,
      "step": 144000
    },
    {
      "epoch": 1.28,
      "eval_loss": 3.0054819583892822,
      "eval_runtime": 112.4817,
      "eval_samples_per_second": 888.873,
      "eval_steps_per_second": 55.556,
      "step": 152000
    },
    {
      "epoch": 1.35,
      "learning_rate": 9.338780955557409e-06,
      "loss": 3.2026,
      "step": 160000
    },
    {
      "epoch": 1.35,
      "eval_loss": 3.0065696239471436,
      "eval_runtime": 112.6067,
      "eval_samples_per_second": 887.887,
      "eval_steps_per_second": 55.494,
      "step": 160000
    },
    {
      "epoch": 1.41,
      "eval_loss": 3.01246976852417,
      "eval_runtime": 112.6363,
      "eval_samples_per_second": 887.654,
      "eval_steps_per_second": 55.479,
      "step": 168000
    },
    {
      "epoch": 1.48,
      "learning_rate": 9.272075377303427e-06,
      "loss": 3.2069,
      "step": 176000
    },
    {
      "epoch": 1.48,
      "eval_loss": 3.0032289028167725,
      "eval_runtime": 112.0908,
      "eval_samples_per_second": 891.974,
      "eval_steps_per_second": 55.749,
      "step": 176000
    },
    {
      "epoch": 1.55,
      "eval_loss": 2.995911121368408,
      "eval_runtime": 114.3666,
      "eval_samples_per_second": 874.224,
      "eval_steps_per_second": 54.64,
      "step": 184000
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.205369799049446e-06,
      "loss": 3.1904,
      "step": 192000
    },
    {
      "epoch": 1.62,
      "eval_loss": 2.9959683418273926,
      "eval_runtime": 113.8999,
      "eval_samples_per_second": 877.806,
      "eval_steps_per_second": 54.864,
      "step": 192000
    },
    {
      "epoch": 1.68,
      "eval_loss": 3.003798246383667,
      "eval_runtime": 114.6475,
      "eval_samples_per_second": 872.082,
      "eval_steps_per_second": 54.506,
      "step": 200000
    },
    {
      "epoch": 1.75,
      "learning_rate": 9.138664220795464e-06,
      "loss": 3.1989,
      "step": 208000
    },
    {
      "epoch": 1.75,
      "eval_loss": 3.0016160011291504,
      "eval_runtime": 113.2837,
      "eval_samples_per_second": 882.58,
      "eval_steps_per_second": 55.162,
      "step": 208000
    },
    {
      "epoch": 1.82,
      "eval_loss": 3.004883050918579,
      "eval_runtime": 113.3643,
      "eval_samples_per_second": 881.953,
      "eval_steps_per_second": 55.123,
      "step": 216000
    },
    {
      "epoch": 1.89,
      "learning_rate": 9.071958642541483e-06,
      "loss": 3.2113,
      "step": 224000
    },
    {
      "epoch": 1.89,
      "eval_loss": 3.0085606575012207,
      "eval_runtime": 114.3159,
      "eval_samples_per_second": 874.612,
      "eval_steps_per_second": 54.664,
      "step": 224000
    },
    {
      "epoch": 1.95,
      "eval_loss": 3.010385274887085,
      "eval_runtime": 113.1162,
      "eval_samples_per_second": 883.887,
      "eval_steps_per_second": 55.244,
      "step": 232000
    },
    {
      "epoch": 2.02,
      "learning_rate": 9.005253064287502e-06,
      "loss": 3.217,
      "step": 240000
    },
    {
      "epoch": 2.02,
      "eval_loss": 3.0165932178497314,
      "eval_runtime": 113.3739,
      "eval_samples_per_second": 881.878,
      "eval_steps_per_second": 55.118,
      "step": 240000
    },
    {
      "epoch": 2.09,
      "eval_loss": 3.0139265060424805,
      "eval_runtime": 112.9548,
      "eval_samples_per_second": 885.15,
      "eval_steps_per_second": 55.323,
      "step": 248000
    },
    {
      "epoch": 2.16,
      "learning_rate": 8.93854748603352e-06,
      "loss": 3.2029,
      "step": 256000
    },
    {
      "epoch": 2.16,
      "eval_loss": 3.021667242050171,
      "eval_runtime": 113.782,
      "eval_samples_per_second": 878.716,
      "eval_steps_per_second": 54.921,
      "step": 256000
    },
    {
      "epoch": 2.22,
      "eval_loss": 3.0237627029418945,
      "eval_runtime": 112.8799,
      "eval_samples_per_second": 885.738,
      "eval_steps_per_second": 55.36,
      "step": 264000
    },
    {
      "epoch": 2.29,
      "learning_rate": 8.871841907779539e-06,
      "loss": 3.2226,
      "step": 272000
    },
    {
      "epoch": 2.29,
      "eval_loss": 3.0233664512634277,
      "eval_runtime": 112.7482,
      "eval_samples_per_second": 886.772,
      "eval_steps_per_second": 55.424,
      "step": 272000
    },
    {
      "epoch": 2.36,
      "eval_loss": 3.0216429233551025,
      "eval_runtime": 112.8816,
      "eval_samples_per_second": 885.725,
      "eval_steps_per_second": 55.359,
      "step": 280000
    },
    {
      "epoch": 2.43,
      "learning_rate": 8.805136329525557e-06,
      "loss": 3.2199,
      "step": 288000
    },
    {
      "epoch": 2.43,
      "eval_loss": 3.0175387859344482,
      "eval_runtime": 113.1399,
      "eval_samples_per_second": 883.703,
      "eval_steps_per_second": 55.233,
      "step": 288000
    },
    {
      "epoch": 2.49,
      "eval_loss": 3.036482095718384,
      "eval_runtime": 112.7456,
      "eval_samples_per_second": 886.793,
      "eval_steps_per_second": 55.426,
      "step": 296000
    },
    {
      "epoch": 2.56,
      "learning_rate": 8.738430751271576e-06,
      "loss": 3.2254,
      "step": 304000
    },
    {
      "epoch": 2.56,
      "eval_loss": 3.028167247772217,
      "eval_runtime": 113.0478,
      "eval_samples_per_second": 884.422,
      "eval_steps_per_second": 55.277,
      "step": 304000
    },
    {
      "epoch": 2.63,
      "eval_loss": 3.0228230953216553,
      "eval_runtime": 112.6735,
      "eval_samples_per_second": 887.36,
      "eval_steps_per_second": 55.461,
      "step": 312000
    },
    {
      "epoch": 2.7,
      "learning_rate": 8.671725173017595e-06,
      "loss": 3.2349,
      "step": 320000
    },
    {
      "epoch": 2.7,
      "eval_loss": 3.0204546451568604,
      "eval_runtime": 113.395,
      "eval_samples_per_second": 881.714,
      "eval_steps_per_second": 55.108,
      "step": 320000
    },
    {
      "epoch": 2.76,
      "eval_loss": 3.0406222343444824,
      "eval_runtime": 113.6641,
      "eval_samples_per_second": 879.627,
      "eval_steps_per_second": 54.978,
      "step": 328000
    },
    {
      "epoch": 2.83,
      "learning_rate": 8.605019594763613e-06,
      "loss": 3.2424,
      "step": 336000
    },
    {
      "epoch": 2.83,
      "eval_loss": 3.0306990146636963,
      "eval_runtime": 113.8559,
      "eval_samples_per_second": 878.145,
      "eval_steps_per_second": 54.885,
      "step": 336000
    },
    {
      "epoch": 2.9,
      "eval_loss": 3.041322708129883,
      "eval_runtime": 112.6936,
      "eval_samples_per_second": 887.202,
      "eval_steps_per_second": 55.451,
      "step": 344000
    },
    {
      "epoch": 2.96,
      "learning_rate": 8.538314016509632e-06,
      "loss": 3.2347,
      "step": 352000
    },
    {
      "epoch": 2.96,
      "eval_loss": 3.0401053428649902,
      "eval_runtime": 113.3728,
      "eval_samples_per_second": 881.887,
      "eval_steps_per_second": 55.119,
      "step": 352000
    },
    {
      "epoch": 3.03,
      "eval_loss": 3.051990270614624,
      "eval_runtime": 112.6317,
      "eval_samples_per_second": 887.69,
      "eval_steps_per_second": 55.482,
      "step": 360000
    },
    {
      "epoch": 3.1,
      "learning_rate": 8.471608438255649e-06,
      "loss": 3.2476,
      "step": 368000
    },
    {
      "epoch": 3.1,
      "eval_loss": 3.0489377975463867,
      "eval_runtime": 112.9,
      "eval_samples_per_second": 885.581,
      "eval_steps_per_second": 55.35,
      "step": 368000
    },
    {
      "epoch": 3.17,
      "eval_loss": 3.052086353302002,
      "eval_runtime": 114.3268,
      "eval_samples_per_second": 874.528,
      "eval_steps_per_second": 54.659,
      "step": 376000
    },
    {
      "epoch": 3.23,
      "learning_rate": 8.404902860001667e-06,
      "loss": 3.2506,
      "step": 384000
    },
    {
      "epoch": 3.23,
      "eval_loss": 3.068549633026123,
      "eval_runtime": 113.2619,
      "eval_samples_per_second": 882.751,
      "eval_steps_per_second": 55.173,
      "step": 384000
    },
    {
      "epoch": 3.3,
      "eval_loss": 3.0546491146087646,
      "eval_runtime": 113.505,
      "eval_samples_per_second": 880.86,
      "eval_steps_per_second": 55.055,
      "step": 392000
    },
    {
      "epoch": 3.37,
      "learning_rate": 8.338197281747686e-06,
      "loss": 3.2547,
      "step": 400000
    },
    {
      "epoch": 3.37,
      "eval_loss": 3.054222822189331,
      "eval_runtime": 113.4821,
      "eval_samples_per_second": 881.037,
      "eval_steps_per_second": 55.066,
      "step": 400000
    },
    {
      "epoch": 3.44,
      "eval_loss": 3.0536701679229736,
      "eval_runtime": 113.6042,
      "eval_samples_per_second": 880.091,
      "eval_steps_per_second": 55.007,
      "step": 408000
    },
    {
      "epoch": 3.5,
      "learning_rate": 8.271491703493705e-06,
      "loss": 3.2519,
      "step": 416000
    },
    {
      "epoch": 3.5,
      "eval_loss": 3.0587801933288574,
      "eval_runtime": 112.9459,
      "eval_samples_per_second": 885.22,
      "eval_steps_per_second": 55.327,
      "step": 416000
    },
    {
      "epoch": 3.57,
      "eval_loss": 3.0728721618652344,
      "eval_runtime": 113.366,
      "eval_samples_per_second": 881.94,
      "eval_steps_per_second": 55.122,
      "step": 424000
    },
    {
      "epoch": 3.64,
      "learning_rate": 8.204786125239725e-06,
      "loss": 3.2679,
      "step": 432000
    },
    {
      "epoch": 3.64,
      "eval_loss": 3.0841500759124756,
      "eval_runtime": 112.9942,
      "eval_samples_per_second": 884.842,
      "eval_steps_per_second": 55.304,
      "step": 432000
    },
    {
      "epoch": 3.71,
      "eval_loss": 3.0685195922851562,
      "eval_runtime": 113.8052,
      "eval_samples_per_second": 878.537,
      "eval_steps_per_second": 54.91,
      "step": 440000
    },
    {
      "epoch": 3.77,
      "learning_rate": 8.138080546985743e-06,
      "loss": 3.2656,
      "step": 448000
    },
    {
      "epoch": 3.77,
      "eval_loss": 3.094174861907959,
      "eval_runtime": 113.3221,
      "eval_samples_per_second": 882.281,
      "eval_steps_per_second": 55.144,
      "step": 448000
    },
    {
      "epoch": 3.84,
      "eval_loss": 3.094191074371338,
      "eval_runtime": 113.0432,
      "eval_samples_per_second": 884.458,
      "eval_steps_per_second": 55.28,
      "step": 456000
    },
    {
      "epoch": 3.91,
      "learning_rate": 8.07137496873176e-06,
      "loss": 3.2908,
      "step": 464000
    },
    {
      "epoch": 3.91,
      "eval_loss": 3.091813087463379,
      "eval_runtime": 113.945,
      "eval_samples_per_second": 877.459,
      "eval_steps_per_second": 54.842,
      "step": 464000
    },
    {
      "epoch": 3.98,
      "eval_loss": 3.0922000408172607,
      "eval_runtime": 113.1262,
      "eval_samples_per_second": 883.81,
      "eval_steps_per_second": 55.239,
      "step": 472000
    },
    {
      "epoch": 4.04,
      "learning_rate": 8.004669390477779e-06,
      "loss": 3.2944,
      "step": 480000
    },
    {
      "epoch": 4.04,
      "eval_loss": 3.109328508377075,
      "eval_runtime": 112.2933,
      "eval_samples_per_second": 890.365,
      "eval_steps_per_second": 55.649,
      "step": 480000
    },
    {
      "epoch": 4.11,
      "eval_loss": 3.11584734916687,
      "eval_runtime": 113.071,
      "eval_samples_per_second": 884.241,
      "eval_steps_per_second": 55.266,
      "step": 488000
    },
    {
      "epoch": 4.18,
      "learning_rate": 7.937963812223798e-06,
      "loss": 3.2917,
      "step": 496000
    },
    {
      "epoch": 4.18,
      "eval_loss": 3.0996689796447754,
      "eval_runtime": 113.0722,
      "eval_samples_per_second": 884.232,
      "eval_steps_per_second": 55.266,
      "step": 496000
    },
    {
      "epoch": 4.24,
      "eval_loss": 3.111070394515991,
      "eval_runtime": 112.4182,
      "eval_samples_per_second": 889.376,
      "eval_steps_per_second": 55.587,
      "step": 504000
    },
    {
      "epoch": 4.31,
      "learning_rate": 7.871258233969816e-06,
      "loss": 3.2916,
      "step": 512000
    },
    {
      "epoch": 4.31,
      "eval_loss": 3.1132729053497314,
      "eval_runtime": 113.7916,
      "eval_samples_per_second": 878.641,
      "eval_steps_per_second": 54.916,
      "step": 512000
    },
    {
      "epoch": 4.38,
      "eval_loss": 3.1128952503204346,
      "eval_runtime": 113.6553,
      "eval_samples_per_second": 879.695,
      "eval_steps_per_second": 54.982,
      "step": 520000
    },
    {
      "epoch": 4.45,
      "learning_rate": 7.804552655715835e-06,
      "loss": 3.2836,
      "step": 528000
    },
    {
      "epoch": 4.45,
      "eval_loss": 3.113443613052368,
      "eval_runtime": 113.5086,
      "eval_samples_per_second": 880.832,
      "eval_steps_per_second": 55.053,
      "step": 528000
    },
    {
      "epoch": 4.51,
      "eval_loss": 3.1057605743408203,
      "eval_runtime": 113.3609,
      "eval_samples_per_second": 881.98,
      "eval_steps_per_second": 55.125,
      "step": 536000
    },
    {
      "epoch": 4.58,
      "learning_rate": 7.737847077461853e-06,
      "loss": 3.3068,
      "step": 544000
    },
    {
      "epoch": 4.58,
      "eval_loss": 3.121134042739868,
      "eval_runtime": 114.162,
      "eval_samples_per_second": 875.79,
      "eval_steps_per_second": 54.738,
      "step": 544000
    },
    {
      "epoch": 4.65,
      "eval_loss": 3.094620704650879,
      "eval_runtime": 113.8018,
      "eval_samples_per_second": 878.563,
      "eval_steps_per_second": 54.911,
      "step": 552000
    },
    {
      "epoch": 4.72,
      "learning_rate": 7.671141499207872e-06,
      "loss": 3.3026,
      "step": 560000
    },
    {
      "epoch": 4.72,
      "eval_loss": 3.107940912246704,
      "eval_runtime": 112.9645,
      "eval_samples_per_second": 885.075,
      "eval_steps_per_second": 55.318,
      "step": 560000
    },
    {
      "epoch": 4.78,
      "eval_loss": 3.120192527770996,
      "eval_runtime": 112.8661,
      "eval_samples_per_second": 885.846,
      "eval_steps_per_second": 55.366,
      "step": 568000
    },
    {
      "epoch": 4.85,
      "learning_rate": 7.604435920953891e-06,
      "loss": 3.3078,
      "step": 576000
    },
    {
      "epoch": 4.85,
      "eval_loss": 3.1155478954315186,
      "eval_runtime": 114.0442,
      "eval_samples_per_second": 876.695,
      "eval_steps_per_second": 54.795,
      "step": 576000
    },
    {
      "epoch": 4.92,
      "eval_loss": 3.125351905822754,
      "eval_runtime": 112.7895,
      "eval_samples_per_second": 886.448,
      "eval_steps_per_second": 55.404,
      "step": 584000
    },
    {
      "epoch": 4.99,
      "learning_rate": 7.537730342699909e-06,
      "loss": 3.3168,
      "step": 592000
    },
    {
      "epoch": 4.99,
      "eval_loss": 3.127920627593994,
      "eval_runtime": 113.3145,
      "eval_samples_per_second": 882.34,
      "eval_steps_per_second": 55.147,
      "step": 592000
    },
    {
      "epoch": 5.05,
      "eval_loss": 3.1179165840148926,
      "eval_runtime": 113.6285,
      "eval_samples_per_second": 879.902,
      "eval_steps_per_second": 54.995,
      "step": 600000
    },
    {
      "epoch": 5.12,
      "learning_rate": 7.471024764445928e-06,
      "loss": 3.3113,
      "step": 608000
    },
    {
      "epoch": 5.12,
      "eval_loss": 3.1277198791503906,
      "eval_runtime": 114.2174,
      "eval_samples_per_second": 875.366,
      "eval_steps_per_second": 54.711,
      "step": 608000
    },
    {
      "epoch": 5.19,
      "eval_loss": 3.133394241333008,
      "eval_runtime": 113.6646,
      "eval_samples_per_second": 879.623,
      "eval_steps_per_second": 54.978,
      "step": 616000
    },
    {
      "epoch": 5.26,
      "learning_rate": 7.4043191861919465e-06,
      "loss": 3.3102,
      "step": 624000
    },
    {
      "epoch": 5.26,
      "eval_loss": 3.123286485671997,
      "eval_runtime": 113.9343,
      "eval_samples_per_second": 877.541,
      "eval_steps_per_second": 54.847,
      "step": 624000
    },
    {
      "epoch": 5.32,
      "eval_loss": 3.1273744106292725,
      "eval_runtime": 113.8778,
      "eval_samples_per_second": 877.976,
      "eval_steps_per_second": 54.875,
      "step": 632000
    },
    {
      "epoch": 5.39,
      "learning_rate": 7.337613607937964e-06,
      "loss": 3.3235,
      "step": 640000
    },
    {
      "epoch": 5.39,
      "eval_loss": 3.1433892250061035,
      "eval_runtime": 113.0915,
      "eval_samples_per_second": 884.08,
      "eval_steps_per_second": 55.256,
      "step": 640000
    },
    {
      "epoch": 5.46,
      "eval_loss": 3.1368374824523926,
      "eval_runtime": 113.2499,
      "eval_samples_per_second": 882.844,
      "eval_steps_per_second": 55.179,
      "step": 648000
    },
    {
      "epoch": 5.53,
      "learning_rate": 7.270908029683983e-06,
      "loss": 3.331,
      "step": 656000
    },
    {
      "epoch": 5.53,
      "eval_loss": 3.1590983867645264,
      "eval_runtime": 114.2316,
      "eval_samples_per_second": 875.257,
      "eval_steps_per_second": 54.705,
      "step": 656000
    },
    {
      "epoch": 5.59,
      "eval_loss": 3.154599189758301,
      "eval_runtime": 113.4705,
      "eval_samples_per_second": 881.128,
      "eval_steps_per_second": 55.072,
      "step": 664000
    },
    {
      "epoch": 5.66,
      "learning_rate": 7.2042024514300015e-06,
      "loss": 3.3308,
      "step": 672000
    },
    {
      "epoch": 5.66,
      "eval_loss": 3.1662509441375732,
      "eval_runtime": 113.7727,
      "eval_samples_per_second": 878.787,
      "eval_steps_per_second": 54.925,
      "step": 672000
    },
    {
      "epoch": 5.73,
      "eval_loss": 3.1535351276397705,
      "eval_runtime": 113.1426,
      "eval_samples_per_second": 883.681,
      "eval_steps_per_second": 55.231,
      "step": 680000
    },
    {
      "epoch": 5.79,
      "learning_rate": 7.13749687317602e-06,
      "loss": 3.3396,
      "step": 688000
    },
    {
      "epoch": 5.79,
      "eval_loss": 3.1558427810668945,
      "eval_runtime": 113.3017,
      "eval_samples_per_second": 882.44,
      "eval_steps_per_second": 55.154,
      "step": 688000
    },
    {
      "epoch": 5.86,
      "eval_loss": 3.169811725616455,
      "eval_runtime": 113.2795,
      "eval_samples_per_second": 882.614,
      "eval_steps_per_second": 55.164,
      "step": 696000
    },
    {
      "epoch": 5.93,
      "learning_rate": 7.070791294922038e-06,
      "loss": 3.3558,
      "step": 704000
    },
    {
      "epoch": 5.93,
      "eval_loss": 3.165127754211426,
      "eval_runtime": 113.7995,
      "eval_samples_per_second": 878.581,
      "eval_steps_per_second": 54.912,
      "step": 704000
    },
    {
      "epoch": 6.0,
      "eval_loss": 3.1705756187438965,
      "eval_runtime": 113.6919,
      "eval_samples_per_second": 879.412,
      "eval_steps_per_second": 54.964,
      "step": 712000
    },
    {
      "epoch": 6.06,
      "learning_rate": 7.0040857166680564e-06,
      "loss": 3.3474,
      "step": 720000
    },
    {
      "epoch": 6.06,
      "eval_loss": 3.194214344024658,
      "eval_runtime": 113.2694,
      "eval_samples_per_second": 882.692,
      "eval_steps_per_second": 55.169,
      "step": 720000
    },
    {
      "epoch": 6.13,
      "eval_loss": 3.170464515686035,
      "eval_runtime": 113.1666,
      "eval_samples_per_second": 883.494,
      "eval_steps_per_second": 55.219,
      "step": 728000
    },
    {
      "epoch": 6.2,
      "learning_rate": 6.937380138414076e-06,
      "loss": 3.3513,
      "step": 736000
    },
    {
      "epoch": 6.2,
      "eval_loss": 3.1834402084350586,
      "eval_runtime": 113.2814,
      "eval_samples_per_second": 882.599,
      "eval_steps_per_second": 55.164,
      "step": 736000
    },
    {
      "epoch": 6.27,
      "eval_loss": 3.1810226440429688,
      "eval_runtime": 113.726,
      "eval_samples_per_second": 879.148,
      "eval_steps_per_second": 54.948,
      "step": 744000
    },
    {
      "epoch": 6.33,
      "learning_rate": 6.8706745601600945e-06,
      "loss": 3.362,
      "step": 752000
    },
    {
      "epoch": 6.33,
      "eval_loss": 3.172321319580078,
      "eval_runtime": 113.4364,
      "eval_samples_per_second": 881.392,
      "eval_steps_per_second": 55.088,
      "step": 752000
    },
    {
      "epoch": 6.4,
      "eval_loss": 3.1826891899108887,
      "eval_runtime": 113.3067,
      "eval_samples_per_second": 882.401,
      "eval_steps_per_second": 55.151,
      "step": 760000
    },
    {
      "epoch": 6.47,
      "learning_rate": 6.803968981906113e-06,
      "loss": 3.3694,
      "step": 768000
    },
    {
      "epoch": 6.47,
      "eval_loss": 3.1937167644500732,
      "eval_runtime": 113.4796,
      "eval_samples_per_second": 881.057,
      "eval_steps_per_second": 55.067,
      "step": 768000
    },
    {
      "epoch": 6.54,
      "eval_loss": 3.2004363536834717,
      "eval_runtime": 114.2163,
      "eval_samples_per_second": 875.374,
      "eval_steps_per_second": 54.712,
      "step": 776000
    },
    {
      "epoch": 6.6,
      "learning_rate": 6.737263403652131e-06,
      "loss": 3.378,
      "step": 784000
    },
    {
      "epoch": 6.6,
      "eval_loss": 3.2023050785064697,
      "eval_runtime": 113.4517,
      "eval_samples_per_second": 881.273,
      "eval_steps_per_second": 55.081,
      "step": 784000
    },
    {
      "epoch": 6.67,
      "eval_loss": 3.1935720443725586,
      "eval_runtime": 113.2158,
      "eval_samples_per_second": 883.11,
      "eval_steps_per_second": 55.195,
      "step": 792000
    },
    {
      "epoch": 6.74,
      "learning_rate": 6.6705578253981495e-06,
      "loss": 3.3703,
      "step": 800000
    },
    {
      "epoch": 6.74,
      "eval_loss": 3.19478702545166,
      "eval_runtime": 112.9761,
      "eval_samples_per_second": 884.984,
      "eval_steps_per_second": 55.313,
      "step": 800000
    },
    {
      "epoch": 6.81,
      "eval_loss": 3.2082369327545166,
      "eval_runtime": 113.4959,
      "eval_samples_per_second": 880.93,
      "eval_steps_per_second": 55.059,
      "step": 808000
    },
    {
      "epoch": 6.87,
      "learning_rate": 6.603852247144168e-06,
      "loss": 3.3838,
      "step": 816000
    },
    {
      "epoch": 6.87,
      "eval_loss": 3.1974425315856934,
      "eval_runtime": 113.5005,
      "eval_samples_per_second": 880.895,
      "eval_steps_per_second": 55.057,
      "step": 816000
    },
    {
      "epoch": 6.94,
      "eval_loss": 3.2029407024383545,
      "eval_runtime": 113.4958,
      "eval_samples_per_second": 880.932,
      "eval_steps_per_second": 55.059,
      "step": 824000
    },
    {
      "epoch": 7.01,
      "learning_rate": 6.537146668890187e-06,
      "loss": 3.3871,
      "step": 832000
    },
    {
      "epoch": 7.01,
      "eval_loss": 3.216017007827759,
      "eval_runtime": 113.9004,
      "eval_samples_per_second": 877.802,
      "eval_steps_per_second": 54.864,
      "step": 832000
    },
    {
      "epoch": 7.07,
      "eval_loss": 3.21976900100708,
      "eval_runtime": 113.1938,
      "eval_samples_per_second": 883.281,
      "eval_steps_per_second": 55.206,
      "step": 840000
    },
    {
      "epoch": 7.14,
      "learning_rate": 6.4704410906362044e-06,
      "loss": 3.3839,
      "step": 848000
    },
    {
      "epoch": 7.14,
      "eval_loss": 3.219007968902588,
      "eval_runtime": 113.9329,
      "eval_samples_per_second": 877.552,
      "eval_steps_per_second": 54.848,
      "step": 848000
    },
    {
      "epoch": 7.21,
      "eval_loss": 3.2204408645629883,
      "eval_runtime": 114.3981,
      "eval_samples_per_second": 873.983,
      "eval_steps_per_second": 54.625,
      "step": 856000
    },
    {
      "epoch": 7.28,
      "learning_rate": 6.403735512382223e-06,
      "loss": 3.389,
      "step": 864000
    },
    {
      "epoch": 7.28,
      "eval_loss": 3.218768835067749,
      "eval_runtime": 115.2564,
      "eval_samples_per_second": 867.475,
      "eval_steps_per_second": 54.218,
      "step": 864000
    },
    {
      "epoch": 7.34,
      "eval_loss": 3.2246193885803223,
      "eval_runtime": 113.3,
      "eval_samples_per_second": 882.454,
      "eval_steps_per_second": 55.154,
      "step": 872000
    },
    {
      "epoch": 7.41,
      "learning_rate": 6.337029934128242e-06,
      "loss": 3.398,
      "step": 880000
    },
    {
      "epoch": 7.41,
      "eval_loss": 3.233250617980957,
      "eval_runtime": 113.6851,
      "eval_samples_per_second": 879.464,
      "eval_steps_per_second": 54.968,
      "step": 880000
    },
    {
      "epoch": 7.48,
      "eval_loss": 3.216823101043701,
      "eval_runtime": 113.0594,
      "eval_samples_per_second": 884.332,
      "eval_steps_per_second": 55.272,
      "step": 888000
    },
    {
      "epoch": 7.55,
      "learning_rate": 6.270324355874261e-06,
      "loss": 3.4001,
      "step": 896000
    },
    {
      "epoch": 7.55,
      "eval_loss": 3.2311105728149414,
      "eval_runtime": 113.3088,
      "eval_samples_per_second": 882.385,
      "eval_steps_per_second": 55.15,
      "step": 896000
    },
    {
      "epoch": 7.61,
      "eval_loss": 3.2389867305755615,
      "eval_runtime": 113.4715,
      "eval_samples_per_second": 881.12,
      "eval_steps_per_second": 55.071,
      "step": 904000
    },
    {
      "epoch": 7.68,
      "learning_rate": 6.20361877762028e-06,
      "loss": 3.4255,
      "step": 912000
    },
    {
      "epoch": 7.68,
      "eval_loss": 3.2446951866149902,
      "eval_runtime": 114.3046,
      "eval_samples_per_second": 874.698,
      "eval_steps_per_second": 54.67,
      "step": 912000
    },
    {
      "epoch": 7.75,
      "eval_loss": 3.254612684249878,
      "eval_runtime": 113.4286,
      "eval_samples_per_second": 881.453,
      "eval_steps_per_second": 55.092,
      "step": 920000
    },
    {
      "epoch": 7.82,
      "learning_rate": 6.1369131993662975e-06,
      "loss": 3.4218,
      "step": 928000
    },
    {
      "epoch": 7.82,
      "eval_loss": 3.250980854034424,
      "eval_runtime": 113.6708,
      "eval_samples_per_second": 879.575,
      "eval_steps_per_second": 54.975,
      "step": 928000
    },
    {
      "epoch": 7.88,
      "eval_loss": 3.243265151977539,
      "eval_runtime": 114.4709,
      "eval_samples_per_second": 873.427,
      "eval_steps_per_second": 54.59,
      "step": 936000
    },
    {
      "epoch": 7.95,
      "learning_rate": 6.070207621112316e-06,
      "loss": 3.4326,
      "step": 944000
    },
    {
      "epoch": 7.95,
      "eval_loss": 3.2509450912475586,
      "eval_runtime": 113.5188,
      "eval_samples_per_second": 880.753,
      "eval_steps_per_second": 55.048,
      "step": 944000
    },
    {
      "epoch": 8.02,
      "eval_loss": 3.257272481918335,
      "eval_runtime": 113.6074,
      "eval_samples_per_second": 880.066,
      "eval_steps_per_second": 55.005,
      "step": 952000
    },
    {
      "epoch": 8.09,
      "learning_rate": 6.003502042858335e-06,
      "loss": 3.4268,
      "step": 960000
    },
    {
      "epoch": 8.09,
      "eval_loss": 3.249929428100586,
      "eval_runtime": 113.664,
      "eval_samples_per_second": 879.628,
      "eval_steps_per_second": 54.978,
      "step": 960000
    },
    {
      "epoch": 8.15,
      "eval_loss": 3.2704036235809326,
      "eval_runtime": 113.9356,
      "eval_samples_per_second": 877.53,
      "eval_steps_per_second": 54.847,
      "step": 968000
    },
    {
      "epoch": 8.22,
      "learning_rate": 5.936796464604353e-06,
      "loss": 3.4165,
      "step": 976000
    },
    {
      "epoch": 8.22,
      "eval_loss": 3.25793194770813,
      "eval_runtime": 114.3815,
      "eval_samples_per_second": 874.11,
      "eval_steps_per_second": 54.633,
      "step": 976000
    },
    {
      "epoch": 8.29,
      "eval_loss": 3.266918420791626,
      "eval_runtime": 113.5329,
      "eval_samples_per_second": 880.643,
      "eval_steps_per_second": 55.041,
      "step": 984000
    },
    {
      "epoch": 8.36,
      "learning_rate": 5.870090886350371e-06,
      "loss": 3.4425,
      "step": 992000
    },
    {
      "epoch": 8.36,
      "eval_loss": 3.272322416305542,
      "eval_runtime": 113.6369,
      "eval_samples_per_second": 879.838,
      "eval_steps_per_second": 54.991,
      "step": 992000
    },
    {
      "epoch": 8.42,
      "eval_loss": 3.2718310356140137,
      "eval_runtime": 113.6845,
      "eval_samples_per_second": 879.469,
      "eval_steps_per_second": 54.968,
      "step": 1000000
    },
    {
      "epoch": 8.49,
      "learning_rate": 5.80338530809639e-06,
      "loss": 3.4433,
      "step": 1008000
    },
    {
      "epoch": 8.49,
      "eval_loss": 3.2655293941497803,
      "eval_runtime": 114.0468,
      "eval_samples_per_second": 876.675,
      "eval_steps_per_second": 54.793,
      "step": 1008000
    },
    {
      "epoch": 8.56,
      "eval_loss": 3.2794032096862793,
      "eval_runtime": 115.3844,
      "eval_samples_per_second": 866.512,
      "eval_steps_per_second": 54.158,
      "step": 1016000
    },
    {
      "epoch": 8.62,
      "learning_rate": 5.736679729842408e-06,
      "loss": 3.4437,
      "step": 1024000
    },
    {
      "epoch": 8.62,
      "eval_loss": 3.2807533740997314,
      "eval_runtime": 114.2793,
      "eval_samples_per_second": 874.892,
      "eval_steps_per_second": 54.682,
      "step": 1024000
    },
    {
      "epoch": 8.69,
      "eval_loss": 3.2731096744537354,
      "eval_runtime": 113.8065,
      "eval_samples_per_second": 878.527,
      "eval_steps_per_second": 54.909,
      "step": 1032000
    },
    {
      "epoch": 8.76,
      "learning_rate": 5.669974151588427e-06,
      "loss": 3.4499,
      "step": 1040000
    },
    {
      "epoch": 8.76,
      "eval_loss": 3.27854323387146,
      "eval_runtime": 113.78,
      "eval_samples_per_second": 878.731,
      "eval_steps_per_second": 54.922,
      "step": 1040000
    },
    {
      "epoch": 8.83,
      "eval_loss": 3.28226900100708,
      "eval_runtime": 113.9563,
      "eval_samples_per_second": 877.371,
      "eval_steps_per_second": 54.837,
      "step": 1048000
    },
    {
      "epoch": 8.89,
      "learning_rate": 5.603268573334446e-06,
      "loss": 3.4593,
      "step": 1056000
    },
    {
      "epoch": 8.89,
      "eval_loss": 3.2843921184539795,
      "eval_runtime": 114.3227,
      "eval_samples_per_second": 874.559,
      "eval_steps_per_second": 54.661,
      "step": 1056000
    },
    {
      "epoch": 8.96,
      "eval_loss": 3.2876641750335693,
      "eval_runtime": 114.4486,
      "eval_samples_per_second": 873.597,
      "eval_steps_per_second": 54.601,
      "step": 1064000
    },
    {
      "epoch": 9.03,
      "learning_rate": 5.536562995080464e-06,
      "loss": 3.4481,
      "step": 1072000
    },
    {
      "epoch": 9.03,
      "eval_loss": 3.2969281673431396,
      "eval_runtime": 113.9166,
      "eval_samples_per_second": 877.677,
      "eval_steps_per_second": 54.856,
      "step": 1072000
    },
    {
      "epoch": 9.1,
      "eval_loss": 3.286954879760742,
      "eval_runtime": 114.297,
      "eval_samples_per_second": 874.756,
      "eval_steps_per_second": 54.673,
      "step": 1080000
    },
    {
      "epoch": 9.16,
      "learning_rate": 5.469857416826483e-06,
      "loss": 3.4542,
      "step": 1088000
    },
    {
      "epoch": 9.16,
      "eval_loss": 3.294614791870117,
      "eval_runtime": 114.0483,
      "eval_samples_per_second": 876.663,
      "eval_steps_per_second": 54.793,
      "step": 1088000
    },
    {
      "epoch": 9.23,
      "eval_loss": 3.2901484966278076,
      "eval_runtime": 114.4116,
      "eval_samples_per_second": 873.88,
      "eval_steps_per_second": 54.619,
      "step": 1096000
    },
    {
      "epoch": 9.3,
      "learning_rate": 5.403151838572501e-06,
      "loss": 3.4547,
      "step": 1104000
    },
    {
      "epoch": 9.3,
      "eval_loss": 3.2812633514404297,
      "eval_runtime": 114.838,
      "eval_samples_per_second": 870.635,
      "eval_steps_per_second": 54.416,
      "step": 1104000
    },
    {
      "epoch": 9.37,
      "eval_loss": 3.2909789085388184,
      "eval_runtime": 114.5859,
      "eval_samples_per_second": 872.551,
      "eval_steps_per_second": 54.536,
      "step": 1112000
    },
    {
      "epoch": 9.43,
      "learning_rate": 5.33644626031852e-06,
      "loss": 3.4618,
      "step": 1120000
    },
    {
      "epoch": 9.43,
      "eval_loss": 3.2978355884552,
      "eval_runtime": 113.8913,
      "eval_samples_per_second": 877.872,
      "eval_steps_per_second": 54.868,
      "step": 1120000
    },
    {
      "epoch": 9.5,
      "eval_loss": 3.3054616451263428,
      "eval_runtime": 114.4856,
      "eval_samples_per_second": 873.315,
      "eval_steps_per_second": 54.583,
      "step": 1128000
    },
    {
      "epoch": 9.57,
      "learning_rate": 5.269740682064538e-06,
      "loss": 3.46,
      "step": 1136000
    },
    {
      "epoch": 9.57,
      "eval_loss": 3.288513422012329,
      "eval_runtime": 114.5126,
      "eval_samples_per_second": 873.109,
      "eval_steps_per_second": 54.57,
      "step": 1136000
    },
    {
      "epoch": 9.64,
      "eval_loss": 3.2871181964874268,
      "eval_runtime": 114.4417,
      "eval_samples_per_second": 873.65,
      "eval_steps_per_second": 54.604,
      "step": 1144000
    },
    {
      "epoch": 9.7,
      "learning_rate": 5.203035103810556e-06,
      "loss": 3.4572,
      "step": 1152000
    },
    {
      "epoch": 9.7,
      "eval_loss": 3.2905023097991943,
      "eval_runtime": 114.8756,
      "eval_samples_per_second": 870.35,
      "eval_steps_per_second": 54.398,
      "step": 1152000
    },
    {
      "epoch": 9.77,
      "eval_loss": 3.300590991973877,
      "eval_runtime": 114.1062,
      "eval_samples_per_second": 876.219,
      "eval_steps_per_second": 54.765,
      "step": 1160000
    },
    {
      "epoch": 9.84,
      "learning_rate": 5.136329525556575e-06,
      "loss": 3.4597,
      "step": 1168000
    },
    {
      "epoch": 9.84,
      "eval_loss": 3.3080625534057617,
      "eval_runtime": 114.1639,
      "eval_samples_per_second": 875.776,
      "eval_steps_per_second": 54.737,
      "step": 1168000
    },
    {
      "epoch": 9.9,
      "eval_loss": 3.3031253814697266,
      "eval_runtime": 115.5682,
      "eval_samples_per_second": 865.134,
      "eval_steps_per_second": 54.072,
      "step": 1176000
    },
    {
      "epoch": 9.97,
      "learning_rate": 5.0696239473025935e-06,
      "loss": 3.4651,
      "step": 1184000
    },
    {
      "epoch": 9.97,
      "eval_loss": 3.288254737854004,
      "eval_runtime": 115.6327,
      "eval_samples_per_second": 864.652,
      "eval_steps_per_second": 54.042,
      "step": 1184000
    },
    {
      "epoch": 10.04,
      "eval_loss": 3.31886625289917,
      "eval_runtime": 115.302,
      "eval_samples_per_second": 867.131,
      "eval_steps_per_second": 54.197,
      "step": 1192000
    },
    {
      "epoch": 10.11,
      "learning_rate": 5.002918369048611e-06,
      "loss": 3.4571,
      "step": 1200000
    },
    {
      "epoch": 10.11,
      "eval_loss": 3.297788619995117,
      "eval_runtime": 116.3659,
      "eval_samples_per_second": 859.204,
      "eval_steps_per_second": 53.701,
      "step": 1200000
    },
    {
      "epoch": 10.17,
      "eval_loss": 3.309051990509033,
      "eval_runtime": 116.156,
      "eval_samples_per_second": 860.756,
      "eval_steps_per_second": 53.798,
      "step": 1208000
    },
    {
      "epoch": 10.24,
      "learning_rate": 4.936212790794631e-06,
      "loss": 3.4567,
      "step": 1216000
    },
    {
      "epoch": 10.24,
      "eval_loss": 3.275514602661133,
      "eval_runtime": 114.7773,
      "eval_samples_per_second": 871.096,
      "eval_steps_per_second": 54.445,
      "step": 1216000
    },
    {
      "epoch": 10.31,
      "eval_loss": 3.296752691268921,
      "eval_runtime": 116.1942,
      "eval_samples_per_second": 860.473,
      "eval_steps_per_second": 53.781,
      "step": 1224000
    },
    {
      "epoch": 10.38,
      "learning_rate": 4.869507212540649e-06,
      "loss": 3.4584,
      "step": 1232000
    },
    {
      "epoch": 10.38,
      "eval_loss": 3.29911732673645,
      "eval_runtime": 115.3134,
      "eval_samples_per_second": 867.046,
      "eval_steps_per_second": 54.191,
      "step": 1232000
    },
    {
      "epoch": 10.44,
      "eval_loss": 3.2818071842193604,
      "eval_runtime": 115.3726,
      "eval_samples_per_second": 866.601,
      "eval_steps_per_second": 54.164,
      "step": 1240000
    },
    {
      "epoch": 10.51,
      "learning_rate": 4.802801634286667e-06,
      "loss": 3.4459,
      "step": 1248000
    },
    {
      "epoch": 10.51,
      "eval_loss": 3.282339334487915,
      "eval_runtime": 115.2771,
      "eval_samples_per_second": 867.319,
      "eval_steps_per_second": 54.209,
      "step": 1248000
    },
    {
      "epoch": 10.58,
      "eval_loss": 3.2800145149230957,
      "eval_runtime": 114.9649,
      "eval_samples_per_second": 869.674,
      "eval_steps_per_second": 54.356,
      "step": 1256000
    },
    {
      "epoch": 10.65,
      "learning_rate": 4.7360960560326865e-06,
      "loss": 3.4474,
      "step": 1264000
    },
    {
      "epoch": 10.65,
      "eval_loss": 3.285576820373535,
      "eval_runtime": 114.3634,
      "eval_samples_per_second": 874.248,
      "eval_steps_per_second": 54.642,
      "step": 1264000
    },
    {
      "epoch": 10.71,
      "eval_loss": 3.2844762802124023,
      "eval_runtime": 114.9015,
      "eval_samples_per_second": 870.154,
      "eval_steps_per_second": 54.386,
      "step": 1272000
    },
    {
      "epoch": 10.78,
      "learning_rate": 4.669390477778704e-06,
      "loss": 3.4383,
      "step": 1280000
    },
    {
      "epoch": 10.78,
      "eval_loss": 3.2804107666015625,
      "eval_runtime": 115.558,
      "eval_samples_per_second": 865.211,
      "eval_steps_per_second": 54.077,
      "step": 1280000
    },
    {
      "epoch": 10.85,
      "eval_loss": 3.270656108856201,
      "eval_runtime": 114.5704,
      "eval_samples_per_second": 872.669,
      "eval_steps_per_second": 54.543,
      "step": 1288000
    },
    {
      "epoch": 10.92,
      "learning_rate": 4.602684899524723e-06,
      "loss": 3.4496,
      "step": 1296000
    },
    {
      "epoch": 10.92,
      "eval_loss": 3.28237247467041,
      "eval_runtime": 115.9515,
      "eval_samples_per_second": 862.274,
      "eval_steps_per_second": 53.893,
      "step": 1296000
    },
    {
      "epoch": 10.98,
      "eval_loss": 3.276503562927246,
      "eval_runtime": 114.4558,
      "eval_samples_per_second": 873.542,
      "eval_steps_per_second": 54.597,
      "step": 1304000
    },
    {
      "epoch": 11.05,
      "learning_rate": 4.5359793212707415e-06,
      "loss": 3.4411,
      "step": 1312000
    },
    {
      "epoch": 11.05,
      "eval_loss": 3.2838242053985596,
      "eval_runtime": 115.3787,
      "eval_samples_per_second": 866.555,
      "eval_steps_per_second": 54.161,
      "step": 1312000
    },
    {
      "epoch": 11.12,
      "eval_loss": 3.2839205265045166,
      "eval_runtime": 114.911,
      "eval_samples_per_second": 870.082,
      "eval_steps_per_second": 54.381,
      "step": 1320000
    },
    {
      "epoch": 11.19,
      "learning_rate": 4.46927374301676e-06,
      "loss": 3.4305,
      "step": 1328000
    },
    {
      "epoch": 11.19,
      "eval_loss": 3.274820327758789,
      "eval_runtime": 114.8509,
      "eval_samples_per_second": 870.538,
      "eval_steps_per_second": 54.41,
      "step": 1328000
    },
    {
      "epoch": 11.25,
      "eval_loss": 3.28206205368042,
      "eval_runtime": 115.1535,
      "eval_samples_per_second": 868.25,
      "eval_steps_per_second": 54.267,
      "step": 1336000
    },
    {
      "epoch": 11.32,
      "learning_rate": 4.402568164762779e-06,
      "loss": 3.4258,
      "step": 1344000
    },
    {
      "epoch": 11.32,
      "eval_loss": 3.274627447128296,
      "eval_runtime": 115.2346,
      "eval_samples_per_second": 867.639,
      "eval_steps_per_second": 54.229,
      "step": 1344000
    },
    {
      "epoch": 11.39,
      "eval_loss": 3.2860774993896484,
      "eval_runtime": 115.4118,
      "eval_samples_per_second": 866.307,
      "eval_steps_per_second": 54.145,
      "step": 1352000
    },
    {
      "epoch": 11.45,
      "learning_rate": 4.335862586508797e-06,
      "loss": 3.4227,
      "step": 1360000
    },
    {
      "epoch": 11.45,
      "eval_loss": 3.27103853225708,
      "eval_runtime": 115.4526,
      "eval_samples_per_second": 866.0,
      "eval_steps_per_second": 54.126,
      "step": 1360000
    },
    {
      "epoch": 11.52,
      "eval_loss": 3.2788245677948,
      "eval_runtime": 114.6345,
      "eval_samples_per_second": 872.181,
      "eval_steps_per_second": 54.512,
      "step": 1368000
    },
    {
      "epoch": 11.59,
      "learning_rate": 4.269157008254816e-06,
      "loss": 3.4319,
      "step": 1376000
    },
    {
      "epoch": 11.59,
      "eval_loss": 3.279372215270996,
      "eval_runtime": 114.9532,
      "eval_samples_per_second": 869.763,
      "eval_steps_per_second": 54.361,
      "step": 1376000
    },
    {
      "epoch": 11.66,
      "eval_loss": 3.27657413482666,
      "eval_runtime": 114.59,
      "eval_samples_per_second": 872.519,
      "eval_steps_per_second": 54.534,
      "step": 1384000
    },
    {
      "epoch": 11.72,
      "learning_rate": 4.202451430000834e-06,
      "loss": 3.436,
      "step": 1392000
    },
    {
      "epoch": 11.72,
      "eval_loss": 3.2923691272735596,
      "eval_runtime": 114.6828,
      "eval_samples_per_second": 871.813,
      "eval_steps_per_second": 54.489,
      "step": 1392000
    },
    {
      "epoch": 11.79,
      "eval_loss": 3.281205654144287,
      "eval_runtime": 114.5759,
      "eval_samples_per_second": 872.626,
      "eval_steps_per_second": 54.54,
      "step": 1400000
    },
    {
      "epoch": 11.86,
      "learning_rate": 4.135745851746852e-06,
      "loss": 3.4368,
      "step": 1408000
    },
    {
      "epoch": 11.86,
      "eval_loss": 3.2851309776306152,
      "eval_runtime": 115.7382,
      "eval_samples_per_second": 863.863,
      "eval_steps_per_second": 53.993,
      "step": 1408000
    },
    {
      "epoch": 11.93,
      "eval_loss": 3.282189130783081,
      "eval_runtime": 115.2625,
      "eval_samples_per_second": 867.429,
      "eval_steps_per_second": 54.215,
      "step": 1416000
    },
    {
      "epoch": 11.99,
      "learning_rate": 4.069040273492872e-06,
      "loss": 3.4346,
      "step": 1424000
    },
    {
      "epoch": 11.99,
      "eval_loss": 3.265688419342041,
      "eval_runtime": 114.9393,
      "eval_samples_per_second": 869.868,
      "eval_steps_per_second": 54.368,
      "step": 1424000
    },
    {
      "epoch": 12.06,
      "eval_loss": 3.2747557163238525,
      "eval_runtime": 115.4082,
      "eval_samples_per_second": 866.334,
      "eval_steps_per_second": 54.147,
      "step": 1432000
    },
    {
      "epoch": 12.13,
      "learning_rate": 4.0023346952388895e-06,
      "loss": 3.4265,
      "step": 1440000
    },
    {
      "epoch": 12.13,
      "eval_loss": 3.2684779167175293,
      "eval_runtime": 115.418,
      "eval_samples_per_second": 866.26,
      "eval_steps_per_second": 54.142,
      "step": 1440000
    },
    {
      "epoch": 12.2,
      "eval_loss": 3.294729709625244,
      "eval_runtime": 116.8415,
      "eval_samples_per_second": 855.706,
      "eval_steps_per_second": 53.483,
      "step": 1448000
    },
    {
      "epoch": 12.26,
      "learning_rate": 3.935629116984908e-06,
      "loss": 3.4306,
      "step": 1456000
    },
    {
      "epoch": 12.26,
      "eval_loss": 3.2840843200683594,
      "eval_runtime": 114.8072,
      "eval_samples_per_second": 870.868,
      "eval_steps_per_second": 54.43,
      "step": 1456000
    },
    {
      "epoch": 12.33,
      "eval_loss": 3.2748351097106934,
      "eval_runtime": 114.5932,
      "eval_samples_per_second": 872.495,
      "eval_steps_per_second": 54.532,
      "step": 1464000
    },
    {
      "epoch": 12.4,
      "learning_rate": 3.868923538730927e-06,
      "loss": 3.4254,
      "step": 1472000
    },
    {
      "epoch": 12.4,
      "eval_loss": 3.2793681621551514,
      "eval_runtime": 114.5871,
      "eval_samples_per_second": 872.542,
      "eval_steps_per_second": 54.535,
      "step": 1472000
    },
    {
      "epoch": 12.47,
      "eval_loss": 3.2773730754852295,
      "eval_runtime": 114.5544,
      "eval_samples_per_second": 872.79,
      "eval_steps_per_second": 54.55,
      "step": 1480000
    },
    {
      "epoch": 12.53,
      "learning_rate": 3.8022179604769453e-06,
      "loss": 3.4353,
      "step": 1488000
    },
    {
      "epoch": 12.53,
      "eval_loss": 3.272627592086792,
      "eval_runtime": 115.5185,
      "eval_samples_per_second": 865.506,
      "eval_steps_per_second": 54.095,
      "step": 1488000
    },
    {
      "epoch": 12.6,
      "eval_loss": 3.27628493309021,
      "eval_runtime": 115.6599,
      "eval_samples_per_second": 864.448,
      "eval_steps_per_second": 54.029,
      "step": 1496000
    },
    {
      "epoch": 12.67,
      "learning_rate": 3.735512382222964e-06,
      "loss": 3.4358,
      "step": 1504000
    },
    {
      "epoch": 12.67,
      "eval_loss": 3.265916109085083,
      "eval_runtime": 116.0704,
      "eval_samples_per_second": 861.391,
      "eval_steps_per_second": 53.838,
      "step": 1504000
    },
    {
      "epoch": 12.73,
      "eval_loss": 3.270989418029785,
      "eval_runtime": 114.6284,
      "eval_samples_per_second": 872.227,
      "eval_steps_per_second": 54.515,
      "step": 1512000
    },
    {
      "epoch": 12.8,
      "learning_rate": 3.668806803968982e-06,
      "loss": 3.4182,
      "step": 1520000
    },
    {
      "epoch": 12.8,
      "eval_loss": 3.27774977684021,
      "eval_runtime": 114.716,
      "eval_samples_per_second": 871.561,
      "eval_steps_per_second": 54.474,
      "step": 1520000
    },
    {
      "epoch": 12.87,
      "eval_loss": 3.2824196815490723,
      "eval_runtime": 114.6356,
      "eval_samples_per_second": 872.173,
      "eval_steps_per_second": 54.512,
      "step": 1528000
    },
    {
      "epoch": 12.94,
      "learning_rate": 3.6021012257150007e-06,
      "loss": 3.4384,
      "step": 1536000
    },
    {
      "epoch": 12.94,
      "eval_loss": 3.2886571884155273,
      "eval_runtime": 117.3403,
      "eval_samples_per_second": 852.069,
      "eval_steps_per_second": 53.255,
      "step": 1536000
    },
    {
      "epoch": 13.0,
      "eval_loss": 3.2667236328125,
      "eval_runtime": 115.4584,
      "eval_samples_per_second": 865.957,
      "eval_steps_per_second": 54.123,
      "step": 1544000
    },
    {
      "epoch": 13.07,
      "learning_rate": 3.535395647461019e-06,
      "loss": 3.4287,
      "step": 1552000
    },
    {
      "epoch": 13.07,
      "eval_loss": 3.271289110183716,
      "eval_runtime": 115.0142,
      "eval_samples_per_second": 869.302,
      "eval_steps_per_second": 54.332,
      "step": 1552000
    },
    {
      "epoch": 13.14,
      "eval_loss": 3.2639546394348145,
      "eval_runtime": 116.2578,
      "eval_samples_per_second": 860.003,
      "eval_steps_per_second": 53.751,
      "step": 1560000
    },
    {
      "epoch": 13.21,
      "learning_rate": 3.468690069207038e-06,
      "loss": 3.4181,
      "step": 1568000
    },
    {
      "epoch": 13.21,
      "eval_loss": 3.260748863220215,
      "eval_runtime": 115.3695,
      "eval_samples_per_second": 866.624,
      "eval_steps_per_second": 54.165,
      "step": 1568000
    },
    {
      "epoch": 13.27,
      "eval_loss": 3.264313220977783,
      "eval_runtime": 115.2949,
      "eval_samples_per_second": 867.185,
      "eval_steps_per_second": 54.2,
      "step": 1576000
    },
    {
      "epoch": 13.34,
      "learning_rate": 3.4019844909530565e-06,
      "loss": 3.4173,
      "step": 1584000
    },
    {
      "epoch": 13.34,
      "eval_loss": 3.262951612472534,
      "eval_runtime": 115.3025,
      "eval_samples_per_second": 867.128,
      "eval_steps_per_second": 54.197,
      "step": 1584000
    },
    {
      "epoch": 13.41,
      "eval_loss": 3.2571523189544678,
      "eval_runtime": 115.8833,
      "eval_samples_per_second": 862.781,
      "eval_steps_per_second": 53.925,
      "step": 1592000
    },
    {
      "epoch": 13.48,
      "learning_rate": 3.3352789126990747e-06,
      "loss": 3.4214,
      "step": 1600000
    },
    {
      "epoch": 13.48,
      "eval_loss": 3.2727572917938232,
      "eval_runtime": 114.8401,
      "eval_samples_per_second": 870.62,
      "eval_steps_per_second": 54.415,
      "step": 1600000
    },
    {
      "epoch": 13.54,
      "eval_loss": 3.2822391986846924,
      "eval_runtime": 115.763,
      "eval_samples_per_second": 863.678,
      "eval_steps_per_second": 53.981,
      "step": 1608000
    },
    {
      "epoch": 13.61,
      "learning_rate": 3.2685733344450933e-06,
      "loss": 3.4223,
      "step": 1616000
    },
    {
      "epoch": 13.61,
      "eval_loss": 3.270418882369995,
      "eval_runtime": 114.8559,
      "eval_samples_per_second": 870.499,
      "eval_steps_per_second": 54.407,
      "step": 1616000
    },
    {
      "epoch": 13.68,
      "eval_loss": 3.263397216796875,
      "eval_runtime": 115.6612,
      "eval_samples_per_second": 864.439,
      "eval_steps_per_second": 54.029,
      "step": 1624000
    },
    {
      "epoch": 13.75,
      "learning_rate": 3.2018677561911115e-06,
      "loss": 3.417,
      "step": 1632000
    },
    {
      "epoch": 13.75,
      "eval_loss": 3.269148349761963,
      "eval_runtime": 116.0444,
      "eval_samples_per_second": 861.584,
      "eval_steps_per_second": 53.85,
      "step": 1632000
    },
    {
      "epoch": 13.81,
      "eval_loss": 3.255032539367676,
      "eval_runtime": 116.0315,
      "eval_samples_per_second": 861.68,
      "eval_steps_per_second": 53.856,
      "step": 1640000
    },
    {
      "epoch": 13.88,
      "learning_rate": 3.1351621779371306e-06,
      "loss": 3.4146,
      "step": 1648000
    },
    {
      "epoch": 13.88,
      "eval_loss": 3.2528505325317383,
      "eval_runtime": 115.6943,
      "eval_samples_per_second": 864.191,
      "eval_steps_per_second": 54.013,
      "step": 1648000
    },
    {
      "epoch": 13.95,
      "eval_loss": 3.271348237991333,
      "eval_runtime": 114.7472,
      "eval_samples_per_second": 871.324,
      "eval_steps_per_second": 54.459,
      "step": 1656000
    },
    {
      "epoch": 14.02,
      "learning_rate": 3.0684565996831487e-06,
      "loss": 3.4186,
      "step": 1664000
    },
    {
      "epoch": 14.02,
      "eval_loss": 3.2671616077423096,
      "eval_runtime": 115.2164,
      "eval_samples_per_second": 867.776,
      "eval_steps_per_second": 54.237,
      "step": 1664000
    },
    {
      "epoch": 14.08,
      "eval_loss": 3.2542309761047363,
      "eval_runtime": 115.2327,
      "eval_samples_per_second": 867.653,
      "eval_steps_per_second": 54.229,
      "step": 1672000
    },
    {
      "epoch": 14.15,
      "learning_rate": 3.0017510214291673e-06,
      "loss": 3.4082,
      "step": 1680000
    },
    {
      "epoch": 14.15,
      "eval_loss": 3.2575573921203613,
      "eval_runtime": 115.6638,
      "eval_samples_per_second": 864.419,
      "eval_steps_per_second": 54.027,
      "step": 1680000
    },
    {
      "epoch": 14.22,
      "eval_loss": 3.2680304050445557,
      "eval_runtime": 116.6886,
      "eval_samples_per_second": 856.827,
      "eval_steps_per_second": 53.553,
      "step": 1688000
    },
    {
      "epoch": 14.28,
      "learning_rate": 2.9350454431751855e-06,
      "loss": 3.4186,
      "step": 1696000
    },
    {
      "epoch": 14.28,
      "eval_loss": 3.266725778579712,
      "eval_runtime": 115.7616,
      "eval_samples_per_second": 863.689,
      "eval_steps_per_second": 53.982,
      "step": 1696000
    },
    {
      "epoch": 14.35,
      "eval_loss": 3.269421339035034,
      "eval_runtime": 115.2493,
      "eval_samples_per_second": 867.528,
      "eval_steps_per_second": 54.222,
      "step": 1704000
    },
    {
      "epoch": 14.42,
      "learning_rate": 2.868339864921204e-06,
      "loss": 3.4131,
      "step": 1712000
    },
    {
      "epoch": 14.42,
      "eval_loss": 3.2606043815612793,
      "eval_runtime": 116.1606,
      "eval_samples_per_second": 860.722,
      "eval_steps_per_second": 53.796,
      "step": 1712000
    },
    {
      "epoch": 14.49,
      "eval_loss": 3.2622175216674805,
      "eval_runtime": 115.2414,
      "eval_samples_per_second": 867.587,
      "eval_steps_per_second": 54.225,
      "step": 1720000
    },
    {
      "epoch": 14.55,
      "learning_rate": 2.801634286667223e-06,
      "loss": 3.4239,
      "step": 1728000
    },
    {
      "epoch": 14.55,
      "eval_loss": 3.2678098678588867,
      "eval_runtime": 116.298,
      "eval_samples_per_second": 859.705,
      "eval_steps_per_second": 53.733,
      "step": 1728000
    },
    {
      "epoch": 14.62,
      "eval_loss": 3.2707767486572266,
      "eval_runtime": 115.813,
      "eval_samples_per_second": 863.305,
      "eval_steps_per_second": 53.958,
      "step": 1736000
    },
    {
      "epoch": 14.69,
      "learning_rate": 2.7349287084132413e-06,
      "loss": 3.4197,
      "step": 1744000
    },
    {
      "epoch": 14.69,
      "eval_loss": 3.2622435092926025,
      "eval_runtime": 116.7137,
      "eval_samples_per_second": 856.643,
      "eval_steps_per_second": 53.541,
      "step": 1744000
    },
    {
      "epoch": 14.76,
      "eval_loss": 3.260528087615967,
      "eval_runtime": 115.8618,
      "eval_samples_per_second": 862.942,
      "eval_steps_per_second": 53.935,
      "step": 1752000
    },
    {
      "epoch": 14.82,
      "learning_rate": 2.66822313015926e-06,
      "loss": 3.4073,
      "step": 1760000
    },
    {
      "epoch": 14.82,
      "eval_loss": 3.2647342681884766,
      "eval_runtime": 116.0753,
      "eval_samples_per_second": 861.355,
      "eval_steps_per_second": 53.836,
      "step": 1760000
    },
    {
      "epoch": 14.89,
      "eval_loss": 3.2618629932403564,
      "eval_runtime": 115.7243,
      "eval_samples_per_second": 863.967,
      "eval_steps_per_second": 53.999,
      "step": 1768000
    },
    {
      "epoch": 14.96,
      "learning_rate": 2.601517551905278e-06,
      "loss": 3.4167,
      "step": 1776000
    },
    {
      "epoch": 14.96,
      "eval_loss": 3.2816412448883057,
      "eval_runtime": 116.0459,
      "eval_samples_per_second": 861.573,
      "eval_steps_per_second": 53.849,
      "step": 1776000
    },
    {
      "epoch": 15.03,
      "eval_loss": 3.260328531265259,
      "eval_runtime": 115.7751,
      "eval_samples_per_second": 863.588,
      "eval_steps_per_second": 53.975,
      "step": 1784000
    },
    {
      "epoch": 15.09,
      "learning_rate": 2.5348119736512967e-06,
      "loss": 3.413,
      "step": 1792000
    },
    {
      "epoch": 15.09,
      "eval_loss": 3.2661468982696533,
      "eval_runtime": 115.5087,
      "eval_samples_per_second": 865.58,
      "eval_steps_per_second": 54.1,
      "step": 1792000
    },
    {
      "epoch": 15.16,
      "eval_loss": 3.2589173316955566,
      "eval_runtime": 115.8968,
      "eval_samples_per_second": 862.681,
      "eval_steps_per_second": 53.919,
      "step": 1800000
    },
    {
      "epoch": 15.23,
      "learning_rate": 2.4681063953973154e-06,
      "loss": 3.4117,
      "step": 1808000
    },
    {
      "epoch": 15.23,
      "eval_loss": 3.2688403129577637,
      "eval_runtime": 115.5062,
      "eval_samples_per_second": 865.598,
      "eval_steps_per_second": 54.101,
      "step": 1808000
    },
    {
      "epoch": 15.3,
      "eval_loss": 3.2677767276763916,
      "eval_runtime": 115.467,
      "eval_samples_per_second": 865.892,
      "eval_steps_per_second": 54.119,
      "step": 1816000
    },
    {
      "epoch": 15.36,
      "learning_rate": 2.4014008171433335e-06,
      "loss": 3.4103,
      "step": 1824000
    },
    {
      "epoch": 15.36,
      "eval_loss": 3.266075372695923,
      "eval_runtime": 114.9844,
      "eval_samples_per_second": 869.527,
      "eval_steps_per_second": 54.347,
      "step": 1824000
    },
    {
      "epoch": 15.43,
      "eval_loss": 3.270479917526245,
      "eval_runtime": 116.2693,
      "eval_samples_per_second": 859.917,
      "eval_steps_per_second": 53.746,
      "step": 1832000
    },
    {
      "epoch": 15.5,
      "learning_rate": 2.334695238889352e-06,
      "loss": 3.4074,
      "step": 1840000
    },
    {
      "epoch": 15.5,
      "eval_loss": 3.267005443572998,
      "eval_runtime": 115.371,
      "eval_samples_per_second": 866.613,
      "eval_steps_per_second": 54.164,
      "step": 1840000
    },
    {
      "epoch": 15.56,
      "eval_loss": 3.2619106769561768,
      "eval_runtime": 115.9983,
      "eval_samples_per_second": 861.926,
      "eval_steps_per_second": 53.871,
      "step": 1848000
    },
    {
      "epoch": 15.63,
      "learning_rate": 2.2679896606353707e-06,
      "loss": 3.4167,
      "step": 1856000
    },
    {
      "epoch": 15.63,
      "eval_loss": 3.262441635131836,
      "eval_runtime": 115.7529,
      "eval_samples_per_second": 863.754,
      "eval_steps_per_second": 53.986,
      "step": 1856000
    },
    {
      "epoch": 15.7,
      "eval_loss": 3.255234479904175,
      "eval_runtime": 115.5701,
      "eval_samples_per_second": 865.12,
      "eval_steps_per_second": 54.071,
      "step": 1864000
    },
    {
      "epoch": 15.77,
      "learning_rate": 2.2012840823813894e-06,
      "loss": 3.4195,
      "step": 1872000
    },
    {
      "epoch": 15.77,
      "eval_loss": 3.2503316402435303,
      "eval_runtime": 115.6488,
      "eval_samples_per_second": 864.531,
      "eval_steps_per_second": 54.034,
      "step": 1872000
    },
    {
      "epoch": 15.83,
      "eval_loss": 3.2605812549591064,
      "eval_runtime": 115.723,
      "eval_samples_per_second": 863.977,
      "eval_steps_per_second": 54.0,
      "step": 1880000
    },
    {
      "epoch": 15.9,
      "learning_rate": 2.134578504127408e-06,
      "loss": 3.4091,
      "step": 1888000
    },
    {
      "epoch": 15.9,
      "eval_loss": 3.2811596393585205,
      "eval_runtime": 116.4883,
      "eval_samples_per_second": 858.301,
      "eval_steps_per_second": 53.645,
      "step": 1888000
    },
    {
      "epoch": 15.97,
      "eval_loss": 3.2836642265319824,
      "eval_runtime": 117.0403,
      "eval_samples_per_second": 854.253,
      "eval_steps_per_second": 53.392,
      "step": 1896000
    },
    {
      "epoch": 16.04,
      "learning_rate": 2.067872925873426e-06,
      "loss": 3.4116,
      "step": 1904000
    },
    {
      "epoch": 16.04,
      "eval_loss": 3.2657785415649414,
      "eval_runtime": 116.2602,
      "eval_samples_per_second": 859.985,
      "eval_steps_per_second": 53.75,
      "step": 1904000
    },
    {
      "epoch": 16.1,
      "eval_loss": 3.267613410949707,
      "eval_runtime": 115.5655,
      "eval_samples_per_second": 865.155,
      "eval_steps_per_second": 54.073,
      "step": 1912000
    },
    {
      "epoch": 16.17,
      "learning_rate": 2.0011673476194448e-06,
      "loss": 3.4183,
      "step": 1920000
    },
    {
      "epoch": 16.17,
      "eval_loss": 3.277005910873413,
      "eval_runtime": 116.4955,
      "eval_samples_per_second": 858.248,
      "eval_steps_per_second": 53.642,
      "step": 1920000
    },
    {
      "epoch": 16.24,
      "eval_loss": 3.2755773067474365,
      "eval_runtime": 116.1016,
      "eval_samples_per_second": 861.159,
      "eval_steps_per_second": 53.824,
      "step": 1928000
    },
    {
      "epoch": 16.31,
      "learning_rate": 1.9344617693654634e-06,
      "loss": 3.4177,
      "step": 1936000
    },
    {
      "epoch": 16.31,
      "eval_loss": 3.28764271736145,
      "eval_runtime": 115.8567,
      "eval_samples_per_second": 862.98,
      "eval_steps_per_second": 53.937,
      "step": 1936000
    },
    {
      "epoch": 16.37,
      "eval_loss": 3.261235475540161,
      "eval_runtime": 116.3096,
      "eval_samples_per_second": 859.619,
      "eval_steps_per_second": 53.727,
      "step": 1944000
    },
    {
      "epoch": 16.44,
      "learning_rate": 1.867756191111482e-06,
      "loss": 3.4226,
      "step": 1952000
    },
    {
      "epoch": 16.44,
      "eval_loss": 3.274751901626587,
      "eval_runtime": 116.724,
      "eval_samples_per_second": 856.568,
      "eval_steps_per_second": 53.537,
      "step": 1952000
    },
    {
      "epoch": 16.51,
      "eval_loss": 3.2679269313812256,
      "eval_runtime": 115.9032,
      "eval_samples_per_second": 862.634,
      "eval_steps_per_second": 53.916,
      "step": 1960000
    },
    {
      "epoch": 16.58,
      "learning_rate": 1.8010506128575004e-06,
      "loss": 3.4154,
      "step": 1968000
    },
    {
      "epoch": 16.58,
      "eval_loss": 3.2658944129943848,
      "eval_runtime": 116.7603,
      "eval_samples_per_second": 856.301,
      "eval_steps_per_second": 53.52,
      "step": 1968000
    },
    {
      "epoch": 16.64,
      "eval_loss": 3.268889904022217,
      "eval_runtime": 116.8782,
      "eval_samples_per_second": 855.438,
      "eval_steps_per_second": 53.466,
      "step": 1976000
    },
    {
      "epoch": 16.71,
      "learning_rate": 1.734345034603519e-06,
      "loss": 3.4199,
      "step": 1984000
    },
    {
      "epoch": 16.71,
      "eval_loss": 3.2701141834259033,
      "eval_runtime": 116.3032,
      "eval_samples_per_second": 859.667,
      "eval_steps_per_second": 53.73,
      "step": 1984000
    },
    {
      "epoch": 16.78,
      "eval_loss": 3.256370782852173,
      "eval_runtime": 116.1051,
      "eval_samples_per_second": 861.133,
      "eval_steps_per_second": 53.822,
      "step": 1992000
    },
    {
      "epoch": 16.85,
      "learning_rate": 1.6676394563495374e-06,
      "loss": 3.4166,
      "step": 2000000
    },
    {
      "epoch": 16.85,
      "eval_loss": 3.2714390754699707,
      "eval_runtime": 117.0225,
      "eval_samples_per_second": 854.383,
      "eval_steps_per_second": 53.4,
      "step": 2000000
    },
    {
      "epoch": 16.91,
      "eval_loss": 3.2737603187561035,
      "eval_runtime": 115.9388,
      "eval_samples_per_second": 862.369,
      "eval_steps_per_second": 53.899,
      "step": 2008000
    },
    {
      "epoch": 16.98,
      "learning_rate": 1.6009338780955558e-06,
      "loss": 3.4054,
      "step": 2016000
    },
    {
      "epoch": 16.98,
      "eval_loss": 3.2632555961608887,
      "eval_runtime": 116.0352,
      "eval_samples_per_second": 861.653,
      "eval_steps_per_second": 53.854,
      "step": 2016000
    },
    {
      "epoch": 17.05,
      "eval_loss": 3.2573704719543457,
      "eval_runtime": 116.0533,
      "eval_samples_per_second": 861.518,
      "eval_steps_per_second": 53.846,
      "step": 2024000
    },
    {
      "epoch": 17.11,
      "learning_rate": 1.5342282998415744e-06,
      "loss": 3.4022,
      "step": 2032000
    },
    {
      "epoch": 17.11,
      "eval_loss": 3.2636642456054688,
      "eval_runtime": 116.7961,
      "eval_samples_per_second": 856.039,
      "eval_steps_per_second": 53.503,
      "step": 2032000
    },
    {
      "epoch": 17.18,
      "eval_loss": 3.268772840499878,
      "eval_runtime": 116.7746,
      "eval_samples_per_second": 856.197,
      "eval_steps_per_second": 53.513,
      "step": 2040000
    },
    {
      "epoch": 17.25,
      "learning_rate": 1.4675227215875928e-06,
      "loss": 3.408,
      "step": 2048000
    },
    {
      "epoch": 17.25,
      "eval_loss": 3.2667033672332764,
      "eval_runtime": 116.5476,
      "eval_samples_per_second": 857.864,
      "eval_steps_per_second": 53.618,
      "step": 2048000
    },
    {
      "epoch": 17.32,
      "eval_loss": 3.2577567100524902,
      "eval_runtime": 116.0215,
      "eval_samples_per_second": 861.754,
      "eval_steps_per_second": 53.861,
      "step": 2056000
    },
    {
      "epoch": 17.38,
      "learning_rate": 1.4008171433336116e-06,
      "loss": 3.4065,
      "step": 2064000
    },
    {
      "epoch": 17.38,
      "eval_loss": 3.2604563236236572,
      "eval_runtime": 116.1892,
      "eval_samples_per_second": 860.51,
      "eval_steps_per_second": 53.783,
      "step": 2064000
    },
    {
      "epoch": 17.45,
      "eval_loss": 3.2768325805664062,
      "eval_runtime": 116.824,
      "eval_samples_per_second": 855.834,
      "eval_steps_per_second": 53.491,
      "step": 2072000
    },
    {
      "epoch": 17.52,
      "learning_rate": 1.33411156507963e-06,
      "loss": 3.4105,
      "step": 2080000
    },
    {
      "epoch": 17.52,
      "eval_loss": 3.256869316101074,
      "eval_runtime": 116.5621,
      "eval_samples_per_second": 857.758,
      "eval_steps_per_second": 53.611,
      "step": 2080000
    },
    {
      "epoch": 17.59,
      "eval_loss": 3.2519402503967285,
      "eval_runtime": 117.0733,
      "eval_samples_per_second": 854.012,
      "eval_steps_per_second": 53.377,
      "step": 2088000
    },
    {
      "epoch": 17.65,
      "learning_rate": 1.2674059868256484e-06,
      "loss": 3.4011,
      "step": 2096000
    },
    {
      "epoch": 17.65,
      "eval_loss": 3.2555432319641113,
      "eval_runtime": 116.8888,
      "eval_samples_per_second": 855.36,
      "eval_steps_per_second": 53.461,
      "step": 2096000
    },
    {
      "epoch": 17.72,
      "eval_loss": 3.248750686645508,
      "eval_runtime": 116.2807,
      "eval_samples_per_second": 859.833,
      "eval_steps_per_second": 53.741,
      "step": 2104000
    },
    {
      "epoch": 17.79,
      "learning_rate": 1.2007004085716668e-06,
      "loss": 3.4078,
      "step": 2112000
    },
    {
      "epoch": 17.79,
      "eval_loss": 3.2515714168548584,
      "eval_runtime": 115.6718,
      "eval_samples_per_second": 864.359,
      "eval_steps_per_second": 54.024,
      "step": 2112000
    },
    {
      "epoch": 17.86,
      "eval_loss": 3.252725124359131,
      "eval_runtime": 116.4663,
      "eval_samples_per_second": 858.463,
      "eval_steps_per_second": 53.655,
      "step": 2120000
    },
    {
      "epoch": 17.92,
      "learning_rate": 1.1339948303176854e-06,
      "loss": 3.4105,
      "step": 2128000
    },
    {
      "epoch": 17.92,
      "eval_loss": 3.256073236465454,
      "eval_runtime": 116.7382,
      "eval_samples_per_second": 856.464,
      "eval_steps_per_second": 53.53,
      "step": 2128000
    },
    {
      "epoch": 17.99,
      "eval_loss": 3.2580018043518066,
      "eval_runtime": 117.5932,
      "eval_samples_per_second": 850.236,
      "eval_steps_per_second": 53.141,
      "step": 2136000
    },
    {
      "epoch": 18.06,
      "learning_rate": 1.067289252063704e-06,
      "loss": 3.4054,
      "step": 2144000
    },
    {
      "epoch": 18.06,
      "eval_loss": 3.2453107833862305,
      "eval_runtime": 117.5514,
      "eval_samples_per_second": 850.538,
      "eval_steps_per_second": 53.16,
      "step": 2144000
    },
    {
      "epoch": 18.13,
      "eval_loss": 3.2425551414489746,
      "eval_runtime": 116.4473,
      "eval_samples_per_second": 858.603,
      "eval_steps_per_second": 53.664,
      "step": 2152000
    },
    {
      "epoch": 18.19,
      "learning_rate": 1.0005836738097224e-06,
      "loss": 3.3937,
      "step": 2160000
    },
    {
      "epoch": 18.19,
      "eval_loss": 3.251696825027466,
      "eval_runtime": 116.4315,
      "eval_samples_per_second": 858.719,
      "eval_steps_per_second": 53.671,
      "step": 2160000
    },
    {
      "epoch": 18.26,
      "eval_loss": 3.2446274757385254,
      "eval_runtime": 116.261,
      "eval_samples_per_second": 859.979,
      "eval_steps_per_second": 53.75,
      "step": 2168000
    },
    {
      "epoch": 18.33,
      "learning_rate": 9.33878095555741e-07,
      "loss": 3.4001,
      "step": 2176000
    },
    {
      "epoch": 18.33,
      "eval_loss": 3.2449288368225098,
      "eval_runtime": 116.9317,
      "eval_samples_per_second": 855.046,
      "eval_steps_per_second": 53.441,
      "step": 2176000
    },
    {
      "epoch": 18.39,
      "eval_loss": 3.252725601196289,
      "eval_runtime": 116.826,
      "eval_samples_per_second": 855.82,
      "eval_steps_per_second": 53.49,
      "step": 2184000
    },
    {
      "epoch": 18.46,
      "learning_rate": 8.671725173017595e-07,
      "loss": 3.413,
      "step": 2192000
    },
    {
      "epoch": 18.46,
      "eval_loss": 3.2557225227355957,
      "eval_runtime": 117.66,
      "eval_samples_per_second": 849.754,
      "eval_steps_per_second": 53.111,
      "step": 2192000
    },
    {
      "epoch": 18.53,
      "eval_loss": 3.2483036518096924,
      "eval_runtime": 116.1271,
      "eval_samples_per_second": 860.97,
      "eval_steps_per_second": 53.812,
      "step": 2200000
    },
    {
      "epoch": 18.6,
      "learning_rate": 8.004669390477779e-07,
      "loss": 3.3882,
      "step": 2208000
    },
    {
      "epoch": 18.6,
      "eval_loss": 3.25201416015625,
      "eval_runtime": 117.5264,
      "eval_samples_per_second": 850.72,
      "eval_steps_per_second": 53.171,
      "step": 2208000
    },
    {
      "epoch": 18.66,
      "eval_loss": 3.2354042530059814,
      "eval_runtime": 117.3232,
      "eval_samples_per_second": 852.193,
      "eval_steps_per_second": 53.263,
      "step": 2216000
    },
    {
      "epoch": 18.73,
      "learning_rate": 7.337613607937964e-07,
      "loss": 3.3974,
      "step": 2224000
    },
    {
      "epoch": 18.73,
      "eval_loss": 3.2540123462677,
      "eval_runtime": 116.4684,
      "eval_samples_per_second": 858.448,
      "eval_steps_per_second": 53.654,
      "step": 2224000
    },
    {
      "epoch": 18.8,
      "eval_loss": 3.242626190185547,
      "eval_runtime": 116.2833,
      "eval_samples_per_second": 859.814,
      "eval_steps_per_second": 53.739,
      "step": 2232000
    },
    {
      "epoch": 18.87,
      "learning_rate": 6.67055782539815e-07,
      "loss": 3.3864,
      "step": 2240000
    },
    {
      "epoch": 18.87,
      "eval_loss": 3.234111785888672,
      "eval_runtime": 117.5517,
      "eval_samples_per_second": 850.537,
      "eval_steps_per_second": 53.16,
      "step": 2240000
    },
    {
      "epoch": 18.93,
      "eval_loss": 3.240849018096924,
      "eval_runtime": 117.6624,
      "eval_samples_per_second": 849.736,
      "eval_steps_per_second": 53.11,
      "step": 2248000
    },
    {
      "epoch": 19.0,
      "learning_rate": 6.003502042858334e-07,
      "loss": 3.3896,
      "step": 2256000
    },
    {
      "epoch": 19.0,
      "eval_loss": 3.234224557876587,
      "eval_runtime": 117.5095,
      "eval_samples_per_second": 850.842,
      "eval_steps_per_second": 53.179,
      "step": 2256000
    },
    {
      "epoch": 19.07,
      "eval_loss": 3.2414724826812744,
      "eval_runtime": 117.6058,
      "eval_samples_per_second": 850.145,
      "eval_steps_per_second": 53.135,
      "step": 2264000
    },
    {
      "epoch": 19.14,
      "learning_rate": 5.33644626031852e-07,
      "loss": 3.3845,
      "step": 2272000
    },
    {
      "epoch": 19.14,
      "eval_loss": 3.2445499897003174,
      "eval_runtime": 117.4897,
      "eval_samples_per_second": 850.985,
      "eval_steps_per_second": 53.188,
      "step": 2272000
    },
    {
      "epoch": 19.2,
      "eval_loss": 3.2422473430633545,
      "eval_runtime": 117.2174,
      "eval_samples_per_second": 852.962,
      "eval_steps_per_second": 53.311,
      "step": 2280000
    },
    {
      "epoch": 19.27,
      "learning_rate": 4.669390477778705e-07,
      "loss": 3.3916,
      "step": 2288000
    },
    {
      "epoch": 19.27,
      "eval_loss": 3.2379391193389893,
      "eval_runtime": 116.6572,
      "eval_samples_per_second": 857.058,
      "eval_steps_per_second": 53.567,
      "step": 2288000
    },
    {
      "epoch": 19.34,
      "eval_loss": 3.241091012954712,
      "eval_runtime": 117.6824,
      "eval_samples_per_second": 849.592,
      "eval_steps_per_second": 53.101,
      "step": 2296000
    },
    {
      "epoch": 19.41,
      "learning_rate": 4.0023346952388894e-07,
      "loss": 3.3919,
      "step": 2304000
    },
    {
      "epoch": 19.41,
      "eval_loss": 3.24294114112854,
      "eval_runtime": 116.557,
      "eval_samples_per_second": 857.795,
      "eval_steps_per_second": 53.613,
      "step": 2304000
    },
    {
      "epoch": 19.47,
      "eval_loss": 3.2372183799743652,
      "eval_runtime": 117.1036,
      "eval_samples_per_second": 853.791,
      "eval_steps_per_second": 53.363,
      "step": 2312000
    },
    {
      "epoch": 19.54,
      "learning_rate": 3.335278912699075e-07,
      "loss": 3.39,
      "step": 2320000
    },
    {
      "epoch": 19.54,
      "eval_loss": 3.2379844188690186,
      "eval_runtime": 118.6177,
      "eval_samples_per_second": 842.892,
      "eval_steps_per_second": 52.682,
      "step": 2320000
    },
    {
      "epoch": 19.61,
      "eval_loss": 3.2353270053863525,
      "eval_runtime": 117.8042,
      "eval_samples_per_second": 848.713,
      "eval_steps_per_second": 53.046,
      "step": 2328000
    },
    {
      "epoch": 19.68,
      "learning_rate": 2.66822313015926e-07,
      "loss": 3.3905,
      "step": 2336000
    },
    {
      "epoch": 19.68,
      "eval_loss": 3.2327044010162354,
      "eval_runtime": 117.5491,
      "eval_samples_per_second": 850.555,
      "eval_steps_per_second": 53.161,
      "step": 2336000
    },
    {
      "epoch": 19.74,
      "eval_loss": 3.2494277954101562,
      "eval_runtime": 117.2833,
      "eval_samples_per_second": 852.483,
      "eval_steps_per_second": 53.281,
      "step": 2344000
    },
    {
      "epoch": 19.81,
      "learning_rate": 2.0011673476194447e-07,
      "loss": 3.3826,
      "step": 2352000
    },
    {
      "epoch": 19.81,
      "eval_loss": 3.2369370460510254,
      "eval_runtime": 117.3893,
      "eval_samples_per_second": 851.713,
      "eval_steps_per_second": 53.233,
      "step": 2352000
    },
    {
      "epoch": 19.88,
      "eval_loss": 3.2389721870422363,
      "eval_runtime": 117.5802,
      "eval_samples_per_second": 850.33,
      "eval_steps_per_second": 53.147,
      "step": 2360000
    },
    {
      "epoch": 19.94,
      "learning_rate": 1.33411156507963e-07,
      "loss": 3.3935,
      "step": 2368000
    },
    {
      "epoch": 19.94,
      "eval_loss": 3.241490125656128,
      "eval_runtime": 116.6908,
      "eval_samples_per_second": 856.811,
      "eval_steps_per_second": 53.552,
      "step": 2368000
    },
    {
      "epoch": 20.01,
      "eval_loss": 3.2485716342926025,
      "eval_runtime": 117.193,
      "eval_samples_per_second": 853.14,
      "eval_steps_per_second": 53.322,
      "step": 2376000
    },
    {
      "epoch": 20.08,
      "learning_rate": 6.67055782539815e-08,
      "loss": 3.3846,
      "step": 2384000
    },
    {
      "epoch": 20.08,
      "eval_loss": 3.2353618144989014,
      "eval_runtime": 116.9874,
      "eval_samples_per_second": 854.639,
      "eval_steps_per_second": 53.416,
      "step": 2384000
    },
    {
      "epoch": 20.15,
      "eval_loss": 3.2465925216674805,
      "eval_runtime": 117.8085,
      "eval_samples_per_second": 848.683,
      "eval_steps_per_second": 53.044,
      "step": 2392000
    },
    {
      "epoch": 20.21,
      "learning_rate": 0.0,
      "loss": 3.3875,
      "step": 2400000
    },
    {
      "epoch": 20.21,
      "eval_loss": 3.2425458431243896,
      "eval_runtime": 117.6341,
      "eval_samples_per_second": 849.941,
      "eval_steps_per_second": 53.122,
      "step": 2400000
    },
    {
      "epoch": 20.21,
      "step": 2400000,
      "total_flos": 7.857549112634404e+17,
      "train_loss": 3.3721608984375,
      "train_runtime": 188638.1929,
      "train_samples_per_second": 203.564,
      "train_steps_per_second": 12.723
    }
  ],
  "logging_steps": 16000,
  "max_steps": 2400000,
  "num_train_epochs": 21,
  "save_steps": 32000,
  "total_flos": 7.857549112634404e+17,
  "trial_name": null,
  "trial_params": null
}