{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.0,
  "eval_steps": 500,
  "global_step": 239310,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 4.989553299068154e-05,
      "loss": 0.5821,
      "step": 500
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.979106598136309e-05,
      "loss": 0.497,
      "step": 1000
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.968659897204463e-05,
      "loss": 0.4743,
      "step": 1500
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9582131962726175e-05,
      "loss": 0.4581,
      "step": 2000
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.947766495340771e-05,
      "loss": 0.4468,
      "step": 2500
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.9373197944089264e-05,
      "loss": 0.4379,
      "step": 3000
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.92687309347708e-05,
      "loss": 0.4312,
      "step": 3500
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.916426392545234e-05,
      "loss": 0.4258,
      "step": 4000
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.9059796916133885e-05,
      "loss": 0.4214,
      "step": 4500
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.895532990681543e-05,
      "loss": 0.4177,
      "step": 5000
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.8850862897496974e-05,
      "loss": 0.414,
      "step": 5500
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.874639588817851e-05,
      "loss": 0.4112,
      "step": 6000
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.864192887886006e-05,
      "loss": 0.4084,
      "step": 6500
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.85374618695416e-05,
      "loss": 0.4065,
      "step": 7000
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.8432994860223146e-05,
      "loss": 0.4041,
      "step": 7500
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.8328527850904684e-05,
      "loss": 0.4022,
      "step": 8000
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.822406084158623e-05,
      "loss": 0.4003,
      "step": 8500
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.8119593832267774e-05,
      "loss": 0.399,
      "step": 9000
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.801512682294931e-05,
      "loss": 0.3971,
      "step": 9500
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.791065981363086e-05,
      "loss": 0.3958,
      "step": 10000
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.78061928043124e-05,
      "loss": 0.3945,
      "step": 10500
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.7701725794993946e-05,
      "loss": 0.3935,
      "step": 11000
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.7597258785675484e-05,
      "loss": 0.3924,
      "step": 11500
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.749279177635703e-05,
      "loss": 0.3909,
      "step": 12000
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.7388324767038574e-05,
      "loss": 0.3898,
      "step": 12500
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.728385775772011e-05,
      "loss": 0.3891,
      "step": 13000
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.7179390748401656e-05,
      "loss": 0.388,
      "step": 13500
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.70749237390832e-05,
      "loss": 0.387,
      "step": 14000
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.6970456729764746e-05,
      "loss": 0.3867,
      "step": 14500
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.6865989720446284e-05,
      "loss": 0.3856,
      "step": 15000
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.676152271112783e-05,
      "loss": 0.3846,
      "step": 15500
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.665705570180937e-05,
      "loss": 0.3838,
      "step": 16000
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.655258869249091e-05,
      "loss": 0.3833,
      "step": 16500
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.6448121683172456e-05,
      "loss": 0.3823,
      "step": 17000
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.6343654673854e-05,
      "loss": 0.3818,
      "step": 17500
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.6239187664535545e-05,
      "loss": 0.3809,
      "step": 18000
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.613472065521708e-05,
      "loss": 0.3806,
      "step": 18500
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.603025364589863e-05,
      "loss": 0.3795,
      "step": 19000
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.592578663658017e-05,
      "loss": 0.3796,
      "step": 19500
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.582131962726171e-05,
      "loss": 0.3787,
      "step": 20000
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.5716852617943256e-05,
      "loss": 0.3782,
      "step": 20500
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.5612385608624794e-05,
      "loss": 0.3777,
      "step": 21000
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.5507918599306345e-05,
      "loss": 0.3772,
      "step": 21500
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.540345158998788e-05,
      "loss": 0.3766,
      "step": 22000
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.529898458066943e-05,
      "loss": 0.3761,
      "step": 22500
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.519451757135097e-05,
      "loss": 0.3755,
      "step": 23000
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.509005056203251e-05,
      "loss": 0.375,
      "step": 23500
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.8853351016181491,
      "eval_loss": 0.36145269870758057,
      "eval_runtime": 1392.8737,
      "eval_samples_per_second": 274.586,
      "eval_steps_per_second": 1.907,
      "step": 23931
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.4985583552714055e-05,
      "loss": 0.3746,
      "step": 24000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.488111654339559e-05,
      "loss": 0.3734,
      "step": 24500
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.4776649534077145e-05,
      "loss": 0.3734,
      "step": 25000
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.467218252475868e-05,
      "loss": 0.3732,
      "step": 25500
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.456771551544023e-05,
      "loss": 0.3723,
      "step": 26000
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.4463248506121765e-05,
      "loss": 0.372,
      "step": 26500
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.435878149680332e-05,
      "loss": 0.3717,
      "step": 27000
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.4254314487484855e-05,
      "loss": 0.3716,
      "step": 27500
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.414984747816639e-05,
      "loss": 0.3709,
      "step": 28000
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.4045380468847944e-05,
      "loss": 0.3707,
      "step": 28500
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.394091345952948e-05,
      "loss": 0.3702,
      "step": 29000
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.383644645021103e-05,
      "loss": 0.37,
      "step": 29500
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.3731979440892565e-05,
      "loss": 0.37,
      "step": 30000
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.3627512431574117e-05,
      "loss": 0.3693,
      "step": 30500
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.3523045422255655e-05,
      "loss": 0.3691,
      "step": 31000
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.341857841293719e-05,
      "loss": 0.3685,
      "step": 31500
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.331411140361874e-05,
      "loss": 0.368,
      "step": 32000
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.320964439430028e-05,
      "loss": 0.368,
      "step": 32500
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.310517738498183e-05,
      "loss": 0.3678,
      "step": 33000
    },
    {
      "epoch": 1.4,
      "learning_rate": 4.3000710375663365e-05,
      "loss": 0.3675,
      "step": 33500
    },
    {
      "epoch": 1.42,
      "learning_rate": 4.289624336634491e-05,
      "loss": 0.3672,
      "step": 34000
    },
    {
      "epoch": 1.44,
      "learning_rate": 4.2791776357026454e-05,
      "loss": 0.3667,
      "step": 34500
    },
    {
      "epoch": 1.46,
      "learning_rate": 4.268730934770799e-05,
      "loss": 0.3667,
      "step": 35000
    },
    {
      "epoch": 1.48,
      "learning_rate": 4.258284233838954e-05,
      "loss": 0.3667,
      "step": 35500
    },
    {
      "epoch": 1.5,
      "learning_rate": 4.247837532907108e-05,
      "loss": 0.3663,
      "step": 36000
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.2373908319752626e-05,
      "loss": 0.3661,
      "step": 36500
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.2269441310434164e-05,
      "loss": 0.3655,
      "step": 37000
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.216497430111571e-05,
      "loss": 0.3655,
      "step": 37500
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.2060507291797254e-05,
      "loss": 0.3654,
      "step": 38000
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.19560402824788e-05,
      "loss": 0.3645,
      "step": 38500
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.1851573273160337e-05,
      "loss": 0.3643,
      "step": 39000
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.174710626384188e-05,
      "loss": 0.3642,
      "step": 39500
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.1642639254523426e-05,
      "loss": 0.3642,
      "step": 40000
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.1538172245204964e-05,
      "loss": 0.364,
      "step": 40500
    },
    {
      "epoch": 1.71,
      "learning_rate": 4.143370523588651e-05,
      "loss": 0.3636,
      "step": 41000
    },
    {
      "epoch": 1.73,
      "learning_rate": 4.1329238226568054e-05,
      "loss": 0.363,
      "step": 41500
    },
    {
      "epoch": 1.76,
      "learning_rate": 4.12247712172496e-05,
      "loss": 0.3632,
      "step": 42000
    },
    {
      "epoch": 1.78,
      "learning_rate": 4.1120304207931136e-05,
      "loss": 0.363,
      "step": 42500
    },
    {
      "epoch": 1.8,
      "learning_rate": 4.101583719861268e-05,
      "loss": 0.3626,
      "step": 43000
    },
    {
      "epoch": 1.82,
      "learning_rate": 4.0911370189294226e-05,
      "loss": 0.3626,
      "step": 43500
    },
    {
      "epoch": 1.84,
      "learning_rate": 4.0806903179975764e-05,
      "loss": 0.3623,
      "step": 44000
    },
    {
      "epoch": 1.86,
      "learning_rate": 4.070243617065731e-05,
      "loss": 0.3621,
      "step": 44500
    },
    {
      "epoch": 1.88,
      "learning_rate": 4.059796916133885e-05,
      "loss": 0.3617,
      "step": 45000
    },
    {
      "epoch": 1.9,
      "learning_rate": 4.04935021520204e-05,
      "loss": 0.3619,
      "step": 45500
    },
    {
      "epoch": 1.92,
      "learning_rate": 4.0389035142701936e-05,
      "loss": 0.3615,
      "step": 46000
    },
    {
      "epoch": 1.94,
      "learning_rate": 4.0284568133383474e-05,
      "loss": 0.361,
      "step": 46500
    },
    {
      "epoch": 1.96,
      "learning_rate": 4.0180101124065025e-05,
      "loss": 0.3611,
      "step": 47000
    },
    {
      "epoch": 1.98,
      "learning_rate": 4.007563411474656e-05,
      "loss": 0.3609,
      "step": 47500
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.8886839759560093,
      "eval_loss": 0.34936127066612244,
      "eval_runtime": 1403.4535,
      "eval_samples_per_second": 272.516,
      "eval_steps_per_second": 1.892,
      "step": 47862
    },
    {
      "epoch": 2.01,
      "learning_rate": 3.997116710542811e-05,
      "loss": 0.3604,
      "step": 48000
    },
    {
      "epoch": 2.03,
      "learning_rate": 3.9866700096109646e-05,
      "loss": 0.3599,
      "step": 48500
    },
    {
      "epoch": 2.05,
      "learning_rate": 3.97622330867912e-05,
      "loss": 0.36,
      "step": 49000
    },
    {
      "epoch": 2.07,
      "learning_rate": 3.9657766077472736e-05,
      "loss": 0.3593,
      "step": 49500
    },
    {
      "epoch": 2.09,
      "learning_rate": 3.955329906815428e-05,
      "loss": 0.3594,
      "step": 50000
    },
    {
      "epoch": 2.11,
      "learning_rate": 3.9448832058835825e-05,
      "loss": 0.3591,
      "step": 50500
    },
    {
      "epoch": 2.13,
      "learning_rate": 3.934436504951736e-05,
      "loss": 0.3591,
      "step": 51000
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.923989804019891e-05,
      "loss": 0.3587,
      "step": 51500
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.9135431030880446e-05,
      "loss": 0.3588,
      "step": 52000
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.9030964021562e-05,
      "loss": 0.3584,
      "step": 52500
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.8926497012243535e-05,
      "loss": 0.3586,
      "step": 53000
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.882203000292508e-05,
      "loss": 0.3582,
      "step": 53500
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.871756299360662e-05,
      "loss": 0.3581,
      "step": 54000
    },
    {
      "epoch": 2.28,
      "learning_rate": 3.861309598428816e-05,
      "loss": 0.3582,
      "step": 54500
    },
    {
      "epoch": 2.3,
      "learning_rate": 3.850862897496971e-05,
      "loss": 0.3579,
      "step": 55000
    },
    {
      "epoch": 2.32,
      "learning_rate": 3.8404161965651245e-05,
      "loss": 0.3577,
      "step": 55500
    },
    {
      "epoch": 2.34,
      "learning_rate": 3.829969495633279e-05,
      "loss": 0.3575,
      "step": 56000
    },
    {
      "epoch": 2.36,
      "learning_rate": 3.8195227947014335e-05,
      "loss": 0.357,
      "step": 56500
    },
    {
      "epoch": 2.38,
      "learning_rate": 3.809076093769588e-05,
      "loss": 0.357,
      "step": 57000
    },
    {
      "epoch": 2.4,
      "learning_rate": 3.798629392837742e-05,
      "loss": 0.357,
      "step": 57500
    },
    {
      "epoch": 2.42,
      "learning_rate": 3.788182691905896e-05,
      "loss": 0.3571,
      "step": 58000
    },
    {
      "epoch": 2.44,
      "learning_rate": 3.777735990974051e-05,
      "loss": 0.3568,
      "step": 58500
    },
    {
      "epoch": 2.47,
      "learning_rate": 3.7672892900422045e-05,
      "loss": 0.3565,
      "step": 59000
    },
    {
      "epoch": 2.49,
      "learning_rate": 3.756842589110359e-05,
      "loss": 0.3564,
      "step": 59500
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.7463958881785134e-05,
      "loss": 0.3567,
      "step": 60000
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.735949187246668e-05,
      "loss": 0.3561,
      "step": 60500
    },
    {
      "epoch": 2.55,
      "learning_rate": 3.725502486314822e-05,
      "loss": 0.3562,
      "step": 61000
    },
    {
      "epoch": 2.57,
      "learning_rate": 3.715055785382976e-05,
      "loss": 0.3558,
      "step": 61500
    },
    {
      "epoch": 2.59,
      "learning_rate": 3.704609084451131e-05,
      "loss": 0.3559,
      "step": 62000
    },
    {
      "epoch": 2.61,
      "learning_rate": 3.6941623835192845e-05,
      "loss": 0.3559,
      "step": 62500
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.683715682587439e-05,
      "loss": 0.3558,
      "step": 63000
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.6732689816555934e-05,
      "loss": 0.3553,
      "step": 63500
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.662822280723748e-05,
      "loss": 0.3554,
      "step": 64000
    },
    {
      "epoch": 2.7,
      "learning_rate": 3.652375579791902e-05,
      "loss": 0.3553,
      "step": 64500
    },
    {
      "epoch": 2.72,
      "learning_rate": 3.641928878860056e-05,
      "loss": 0.3547,
      "step": 65000
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.6314821779282106e-05,
      "loss": 0.3547,
      "step": 65500
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.6210354769963644e-05,
      "loss": 0.3551,
      "step": 66000
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.610588776064519e-05,
      "loss": 0.3547,
      "step": 66500
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.6001420751326734e-05,
      "loss": 0.3546,
      "step": 67000
    },
    {
      "epoch": 2.82,
      "learning_rate": 3.589695374200828e-05,
      "loss": 0.3545,
      "step": 67500
    },
    {
      "epoch": 2.84,
      "learning_rate": 3.5792486732689817e-05,
      "loss": 0.3544,
      "step": 68000
    },
    {
      "epoch": 2.86,
      "learning_rate": 3.568801972337136e-05,
      "loss": 0.3544,
      "step": 68500
    },
    {
      "epoch": 2.88,
      "learning_rate": 3.5583552714052906e-05,
      "loss": 0.354,
      "step": 69000
    },
    {
      "epoch": 2.9,
      "learning_rate": 3.5479085704734444e-05,
      "loss": 0.354,
      "step": 69500
    },
    {
      "epoch": 2.93,
      "learning_rate": 3.537461869541599e-05,
      "loss": 0.3539,
      "step": 70000
    },
    {
      "epoch": 2.95,
      "learning_rate": 3.527015168609753e-05,
      "loss": 0.3539,
      "step": 70500
    },
    {
      "epoch": 2.97,
      "learning_rate": 3.516568467677908e-05,
      "loss": 0.3536,
      "step": 71000
    },
    {
      "epoch": 2.99,
      "learning_rate": 3.5061217667460616e-05,
      "loss": 0.3533,
      "step": 71500
    },
    {
      "epoch": 3.0,
      "eval_accuracy": 0.8903601451222372,
      "eval_loss": 0.3432493805885315,
      "eval_runtime": 1415.4457,
      "eval_samples_per_second": 270.207,
      "eval_steps_per_second": 1.876,
      "step": 71793
    },
    {
      "epoch": 3.01,
      "learning_rate": 3.495675065814216e-05,
      "loss": 0.3533,
      "step": 72000
    },
    {
      "epoch": 3.03,
      "learning_rate": 3.4852283648823706e-05,
      "loss": 0.3527,
      "step": 72500
    },
    {
      "epoch": 3.05,
      "learning_rate": 3.474781663950525e-05,
      "loss": 0.3527,
      "step": 73000
    },
    {
      "epoch": 3.07,
      "learning_rate": 3.464334963018679e-05,
      "loss": 0.3526,
      "step": 73500
    },
    {
      "epoch": 3.09,
      "learning_rate": 3.4538882620868326e-05,
      "loss": 0.3524,
      "step": 74000
    },
    {
      "epoch": 3.11,
      "learning_rate": 3.443441561154988e-05,
      "loss": 0.3526,
      "step": 74500
    },
    {
      "epoch": 3.13,
      "learning_rate": 3.4329948602231416e-05,
      "loss": 0.3524,
      "step": 75000
    },
    {
      "epoch": 3.15,
      "learning_rate": 3.422548159291296e-05,
      "loss": 0.3522,
      "step": 75500
    },
    {
      "epoch": 3.18,
      "learning_rate": 3.41210145835945e-05,
      "loss": 0.3521,
      "step": 76000
    },
    {
      "epoch": 3.2,
      "learning_rate": 3.401654757427605e-05,
      "loss": 0.3519,
      "step": 76500
    },
    {
      "epoch": 3.22,
      "learning_rate": 3.391208056495759e-05,
      "loss": 0.3518,
      "step": 77000
    },
    {
      "epoch": 3.24,
      "learning_rate": 3.3807613555639126e-05,
      "loss": 0.3519,
      "step": 77500
    },
    {
      "epoch": 3.26,
      "learning_rate": 3.370314654632068e-05,
      "loss": 0.3518,
      "step": 78000
    },
    {
      "epoch": 3.28,
      "learning_rate": 3.3598679537002215e-05,
      "loss": 0.3516,
      "step": 78500
    },
    {
      "epoch": 3.3,
      "learning_rate": 3.349421252768376e-05,
      "loss": 0.3513,
      "step": 79000
    },
    {
      "epoch": 3.32,
      "learning_rate": 3.33897455183653e-05,
      "loss": 0.3514,
      "step": 79500
    },
    {
      "epoch": 3.34,
      "learning_rate": 3.328527850904685e-05,
      "loss": 0.3512,
      "step": 80000
    },
    {
      "epoch": 3.36,
      "learning_rate": 3.318081149972839e-05,
      "loss": 0.3512,
      "step": 80500
    },
    {
      "epoch": 3.38,
      "learning_rate": 3.307634449040993e-05,
      "loss": 0.3512,
      "step": 81000
    },
    {
      "epoch": 3.41,
      "learning_rate": 3.297187748109147e-05,
      "loss": 0.3515,
      "step": 81500
    },
    {
      "epoch": 3.43,
      "learning_rate": 3.2867410471773015e-05,
      "loss": 0.351,
      "step": 82000
    },
    {
      "epoch": 3.45,
      "learning_rate": 3.276294346245456e-05,
      "loss": 0.3512,
      "step": 82500
    },
    {
      "epoch": 3.47,
      "learning_rate": 3.26584764531361e-05,
      "loss": 0.3506,
      "step": 83000
    },
    {
      "epoch": 3.49,
      "learning_rate": 3.255400944381764e-05,
      "loss": 0.3508,
      "step": 83500
    },
    {
      "epoch": 3.51,
      "learning_rate": 3.244954243449919e-05,
      "loss": 0.3509,
      "step": 84000
    },
    {
      "epoch": 3.53,
      "learning_rate": 3.234507542518073e-05,
      "loss": 0.3505,
      "step": 84500
    },
    {
      "epoch": 3.55,
      "learning_rate": 3.224060841586227e-05,
      "loss": 0.3505,
      "step": 85000
    },
    {
      "epoch": 3.57,
      "learning_rate": 3.2136141406543815e-05,
      "loss": 0.3502,
      "step": 85500
    },
    {
      "epoch": 3.59,
      "learning_rate": 3.203167439722536e-05,
      "loss": 0.3502,
      "step": 86000
    },
    {
      "epoch": 3.61,
      "learning_rate": 3.19272073879069e-05,
      "loss": 0.3499,
      "step": 86500
    },
    {
      "epoch": 3.64,
      "learning_rate": 3.182274037858844e-05,
      "loss": 0.3502,
      "step": 87000
    },
    {
      "epoch": 3.66,
      "learning_rate": 3.171827336926999e-05,
      "loss": 0.3499,
      "step": 87500
    },
    {
      "epoch": 3.68,
      "learning_rate": 3.161380635995153e-05,
      "loss": 0.3496,
      "step": 88000
    },
    {
      "epoch": 3.7,
      "learning_rate": 3.150933935063307e-05,
      "loss": 0.3499,
      "step": 88500
    },
    {
      "epoch": 3.72,
      "learning_rate": 3.1404872341314614e-05,
      "loss": 0.3499,
      "step": 89000
    },
    {
      "epoch": 3.74,
      "learning_rate": 3.130040533199616e-05,
      "loss": 0.3497,
      "step": 89500
    },
    {
      "epoch": 3.76,
      "learning_rate": 3.11959383226777e-05,
      "loss": 0.3496,
      "step": 90000
    },
    {
      "epoch": 3.78,
      "learning_rate": 3.109147131335924e-05,
      "loss": 0.3494,
      "step": 90500
    },
    {
      "epoch": 3.8,
      "learning_rate": 3.098700430404079e-05,
      "loss": 0.3496,
      "step": 91000
    },
    {
      "epoch": 3.82,
      "learning_rate": 3.088253729472233e-05,
      "loss": 0.3494,
      "step": 91500
    },
    {
      "epoch": 3.84,
      "learning_rate": 3.077807028540387e-05,
      "loss": 0.3492,
      "step": 92000
    },
    {
      "epoch": 3.87,
      "learning_rate": 3.0673603276085414e-05,
      "loss": 0.3494,
      "step": 92500
    },
    {
      "epoch": 3.89,
      "learning_rate": 3.056913626676696e-05,
      "loss": 0.3494,
      "step": 93000
    },
    {
      "epoch": 3.91,
      "learning_rate": 3.0464669257448497e-05,
      "loss": 0.349,
      "step": 93500
    },
    {
      "epoch": 3.93,
      "learning_rate": 3.036020224813004e-05,
      "loss": 0.3491,
      "step": 94000
    },
    {
      "epoch": 3.95,
      "learning_rate": 3.0255735238811583e-05,
      "loss": 0.3486,
      "step": 94500
    },
    {
      "epoch": 3.97,
      "learning_rate": 3.0151268229493128e-05,
      "loss": 0.3489,
      "step": 95000
    },
    {
      "epoch": 3.99,
      "learning_rate": 3.004680122017467e-05,
      "loss": 0.3486,
      "step": 95500
    },
    {
      "epoch": 4.0,
      "eval_accuracy": 0.8913953473712201,
      "eval_loss": 0.3394201993942261,
      "eval_runtime": 1409.3785,
      "eval_samples_per_second": 271.371,
      "eval_steps_per_second": 1.885,
      "step": 95724
    },
    {
      "epoch": 4.01,
      "learning_rate": 2.9942334210856217e-05,
      "loss": 0.3483,
      "step": 96000
    },
    {
      "epoch": 4.03,
      "learning_rate": 2.9837867201537755e-05,
      "loss": 0.3479,
      "step": 96500
    },
    {
      "epoch": 4.05,
      "learning_rate": 2.9733400192219296e-05,
      "loss": 0.348,
      "step": 97000
    },
    {
      "epoch": 4.07,
      "learning_rate": 2.962893318290084e-05,
      "loss": 0.3479,
      "step": 97500
    },
    {
      "epoch": 4.1,
      "learning_rate": 2.9524466173582383e-05,
      "loss": 0.348,
      "step": 98000
    },
    {
      "epoch": 4.12,
      "learning_rate": 2.9419999164263927e-05,
      "loss": 0.3478,
      "step": 98500
    },
    {
      "epoch": 4.14,
      "learning_rate": 2.931553215494547e-05,
      "loss": 0.3477,
      "step": 99000
    },
    {
      "epoch": 4.16,
      "learning_rate": 2.9211065145627013e-05,
      "loss": 0.3477,
      "step": 99500
    },
    {
      "epoch": 4.18,
      "learning_rate": 2.9106598136308555e-05,
      "loss": 0.3479,
      "step": 100000
    },
    {
      "epoch": 4.2,
      "learning_rate": 2.9002131126990096e-05,
      "loss": 0.3479,
      "step": 100500
    },
    {
      "epoch": 4.22,
      "learning_rate": 2.889766411767164e-05,
      "loss": 0.3478,
      "step": 101000
    },
    {
      "epoch": 4.24,
      "learning_rate": 2.8793197108353182e-05,
      "loss": 0.3475,
      "step": 101500
    },
    {
      "epoch": 4.26,
      "learning_rate": 2.8688730099034727e-05,
      "loss": 0.3476,
      "step": 102000
    },
    {
      "epoch": 4.28,
      "learning_rate": 2.858426308971627e-05,
      "loss": 0.3472,
      "step": 102500
    },
    {
      "epoch": 4.3,
      "learning_rate": 2.8479796080397813e-05,
      "loss": 0.3473,
      "step": 103000
    },
    {
      "epoch": 4.32,
      "learning_rate": 2.8375329071079354e-05,
      "loss": 0.3472,
      "step": 103500
    },
    {
      "epoch": 4.35,
      "learning_rate": 2.82708620617609e-05,
      "loss": 0.347,
      "step": 104000
    },
    {
      "epoch": 4.37,
      "learning_rate": 2.816639505244244e-05,
      "loss": 0.3471,
      "step": 104500
    },
    {
      "epoch": 4.39,
      "learning_rate": 2.8061928043123982e-05,
      "loss": 0.3467,
      "step": 105000
    },
    {
      "epoch": 4.41,
      "learning_rate": 2.7957461033805527e-05,
      "loss": 0.3469,
      "step": 105500
    },
    {
      "epoch": 4.43,
      "learning_rate": 2.7852994024487068e-05,
      "loss": 0.3469,
      "step": 106000
    },
    {
      "epoch": 4.45,
      "learning_rate": 2.7748527015168613e-05,
      "loss": 0.3469,
      "step": 106500
    },
    {
      "epoch": 4.47,
      "learning_rate": 2.7644060005850154e-05,
      "loss": 0.3469,
      "step": 107000
    },
    {
      "epoch": 4.49,
      "learning_rate": 2.75395929965317e-05,
      "loss": 0.3468,
      "step": 107500
    },
    {
      "epoch": 4.51,
      "learning_rate": 2.743512598721324e-05,
      "loss": 0.3466,
      "step": 108000
    },
    {
      "epoch": 4.53,
      "learning_rate": 2.7330658977894778e-05,
      "loss": 0.3468,
      "step": 108500
    },
    {
      "epoch": 4.55,
      "learning_rate": 2.7226191968576326e-05,
      "loss": 0.3464,
      "step": 109000
    },
    {
      "epoch": 4.58,
      "learning_rate": 2.7121724959257868e-05,
      "loss": 0.3466,
      "step": 109500
    },
    {
      "epoch": 4.6,
      "learning_rate": 2.7017257949939412e-05,
      "loss": 0.3463,
      "step": 110000
    },
    {
      "epoch": 4.62,
      "learning_rate": 2.6912790940620954e-05,
      "loss": 0.3466,
      "step": 110500
    },
    {
      "epoch": 4.64,
      "learning_rate": 2.68083239313025e-05,
      "loss": 0.3462,
      "step": 111000
    },
    {
      "epoch": 4.66,
      "learning_rate": 2.670385692198404e-05,
      "loss": 0.3462,
      "step": 111500
    },
    {
      "epoch": 4.68,
      "learning_rate": 2.6599389912665578e-05,
      "loss": 0.3462,
      "step": 112000
    },
    {
      "epoch": 4.7,
      "learning_rate": 2.6494922903347126e-05,
      "loss": 0.3463,
      "step": 112500
    },
    {
      "epoch": 4.72,
      "learning_rate": 2.6390455894028664e-05,
      "loss": 0.3463,
      "step": 113000
    },
    {
      "epoch": 4.74,
      "learning_rate": 2.6285988884710212e-05,
      "loss": 0.3461,
      "step": 113500
    },
    {
      "epoch": 4.76,
      "learning_rate": 2.618152187539175e-05,
      "loss": 0.346,
      "step": 114000
    },
    {
      "epoch": 4.78,
      "learning_rate": 2.6077054866073298e-05,
      "loss": 0.3461,
      "step": 114500
    },
    {
      "epoch": 4.81,
      "learning_rate": 2.5972587856754836e-05,
      "loss": 0.3463,
      "step": 115000
    },
    {
      "epoch": 4.83,
      "learning_rate": 2.5868120847436384e-05,
      "loss": 0.3458,
      "step": 115500
    },
    {
      "epoch": 4.85,
      "learning_rate": 2.5763653838117922e-05,
      "loss": 0.3459,
      "step": 116000
    },
    {
      "epoch": 4.87,
      "learning_rate": 2.5659186828799464e-05,
      "loss": 0.3459,
      "step": 116500
    },
    {
      "epoch": 4.89,
      "learning_rate": 2.555471981948101e-05,
      "loss": 0.3455,
      "step": 117000
    },
    {
      "epoch": 4.91,
      "learning_rate": 2.545025281016255e-05,
      "loss": 0.3459,
      "step": 117500
    },
    {
      "epoch": 4.93,
      "learning_rate": 2.5345785800844098e-05,
      "loss": 0.3456,
      "step": 118000
    },
    {
      "epoch": 4.95,
      "learning_rate": 2.5241318791525636e-05,
      "loss": 0.3455,
      "step": 118500
    },
    {
      "epoch": 4.97,
      "learning_rate": 2.5136851782207184e-05,
      "loss": 0.3454,
      "step": 119000
    },
    {
      "epoch": 4.99,
      "learning_rate": 2.5032384772888722e-05,
      "loss": 0.3456,
      "step": 119500
    },
    {
      "epoch": 5.0,
      "eval_accuracy": 0.8921225304021042,
      "eval_loss": 0.33668240904808044,
      "eval_runtime": 1399.0502,
      "eval_samples_per_second": 273.374,
      "eval_steps_per_second": 1.898,
      "step": 119655
    },
    {
      "epoch": 5.01,
      "learning_rate": 2.4927917763570267e-05,
      "loss": 0.3447,
      "step": 120000
    },
    {
      "epoch": 5.04,
      "learning_rate": 2.4823450754251808e-05,
      "loss": 0.345,
      "step": 120500
    },
    {
      "epoch": 5.06,
      "learning_rate": 2.4718983744933353e-05,
      "loss": 0.345,
      "step": 121000
    },
    {
      "epoch": 5.08,
      "learning_rate": 2.4614516735614894e-05,
      "loss": 0.3447,
      "step": 121500
    },
    {
      "epoch": 5.1,
      "learning_rate": 2.4510049726296435e-05,
      "loss": 0.3448,
      "step": 122000
    },
    {
      "epoch": 5.12,
      "learning_rate": 2.440558271697798e-05,
      "loss": 0.3448,
      "step": 122500
    },
    {
      "epoch": 5.14,
      "learning_rate": 2.430111570765952e-05,
      "loss": 0.3446,
      "step": 123000
    },
    {
      "epoch": 5.16,
      "learning_rate": 2.4196648698341066e-05,
      "loss": 0.3446,
      "step": 123500
    },
    {
      "epoch": 5.18,
      "learning_rate": 2.4092181689022608e-05,
      "loss": 0.3443,
      "step": 124000
    },
    {
      "epoch": 5.2,
      "learning_rate": 2.3987714679704152e-05,
      "loss": 0.3442,
      "step": 124500
    },
    {
      "epoch": 5.22,
      "learning_rate": 2.3883247670385694e-05,
      "loss": 0.3444,
      "step": 125000
    },
    {
      "epoch": 5.24,
      "learning_rate": 2.3778780661067235e-05,
      "loss": 0.3446,
      "step": 125500
    },
    {
      "epoch": 5.27,
      "learning_rate": 2.3674313651748776e-05,
      "loss": 0.3444,
      "step": 126000
    },
    {
      "epoch": 5.29,
      "learning_rate": 2.356984664243032e-05,
      "loss": 0.3443,
      "step": 126500
    },
    {
      "epoch": 5.31,
      "learning_rate": 2.3465379633111863e-05,
      "loss": 0.3443,
      "step": 127000
    },
    {
      "epoch": 5.33,
      "learning_rate": 2.3360912623793407e-05,
      "loss": 0.3442,
      "step": 127500
    },
    {
      "epoch": 5.35,
      "learning_rate": 2.3256445614474952e-05,
      "loss": 0.3442,
      "step": 128000
    },
    {
      "epoch": 5.37,
      "learning_rate": 2.3151978605156493e-05,
      "loss": 0.3442,
      "step": 128500
    },
    {
      "epoch": 5.39,
      "learning_rate": 2.3047511595838038e-05,
      "loss": 0.3444,
      "step": 129000
    },
    {
      "epoch": 5.41,
      "learning_rate": 2.2943044586519576e-05,
      "loss": 0.3441,
      "step": 129500
    },
    {
      "epoch": 5.43,
      "learning_rate": 2.283857757720112e-05,
      "loss": 0.3442,
      "step": 130000
    },
    {
      "epoch": 5.45,
      "learning_rate": 2.2734110567882662e-05,
      "loss": 0.3438,
      "step": 130500
    },
    {
      "epoch": 5.47,
      "learning_rate": 2.2629643558564207e-05,
      "loss": 0.344,
      "step": 131000
    },
    {
      "epoch": 5.49,
      "learning_rate": 2.2525176549245748e-05,
      "loss": 0.3439,
      "step": 131500
    },
    {
      "epoch": 5.52,
      "learning_rate": 2.2420709539927293e-05,
      "loss": 0.3437,
      "step": 132000
    },
    {
      "epoch": 5.54,
      "learning_rate": 2.2316242530608834e-05,
      "loss": 0.3435,
      "step": 132500
    },
    {
      "epoch": 5.56,
      "learning_rate": 2.221177552129038e-05,
      "loss": 0.3435,
      "step": 133000
    },
    {
      "epoch": 5.58,
      "learning_rate": 2.210730851197192e-05,
      "loss": 0.3438,
      "step": 133500
    },
    {
      "epoch": 5.6,
      "learning_rate": 2.2002841502653462e-05,
      "loss": 0.3436,
      "step": 134000
    },
    {
      "epoch": 5.62,
      "learning_rate": 2.1898374493335007e-05,
      "loss": 0.3437,
      "step": 134500
    },
    {
      "epoch": 5.64,
      "learning_rate": 2.1793907484016548e-05,
      "loss": 0.3436,
      "step": 135000
    },
    {
      "epoch": 5.66,
      "learning_rate": 2.1689440474698093e-05,
      "loss": 0.3436,
      "step": 135500
    },
    {
      "epoch": 5.68,
      "learning_rate": 2.1584973465379634e-05,
      "loss": 0.3434,
      "step": 136000
    },
    {
      "epoch": 5.7,
      "learning_rate": 2.148050645606118e-05,
      "loss": 0.3435,
      "step": 136500
    },
    {
      "epoch": 5.72,
      "learning_rate": 2.1376039446742717e-05,
      "loss": 0.3435,
      "step": 137000
    },
    {
      "epoch": 5.75,
      "learning_rate": 2.127157243742426e-05,
      "loss": 0.3437,
      "step": 137500
    },
    {
      "epoch": 5.77,
      "learning_rate": 2.1167105428105806e-05,
      "loss": 0.3434,
      "step": 138000
    },
    {
      "epoch": 5.79,
      "learning_rate": 2.1062638418787348e-05,
      "loss": 0.3434,
      "step": 138500
    },
    {
      "epoch": 5.81,
      "learning_rate": 2.0958171409468892e-05,
      "loss": 0.3433,
      "step": 139000
    },
    {
      "epoch": 5.83,
      "learning_rate": 2.0853704400150434e-05,
      "loss": 0.3434,
      "step": 139500
    },
    {
      "epoch": 5.85,
      "learning_rate": 2.074923739083198e-05,
      "loss": 0.3434,
      "step": 140000
    },
    {
      "epoch": 5.87,
      "learning_rate": 2.064477038151352e-05,
      "loss": 0.3431,
      "step": 140500
    },
    {
      "epoch": 5.89,
      "learning_rate": 2.054030337219506e-05,
      "loss": 0.3432,
      "step": 141000
    },
    {
      "epoch": 5.91,
      "learning_rate": 2.0435836362876602e-05,
      "loss": 0.3432,
      "step": 141500
    },
    {
      "epoch": 5.93,
      "learning_rate": 2.0331369353558147e-05,
      "loss": 0.3432,
      "step": 142000
    },
    {
      "epoch": 5.95,
      "learning_rate": 2.022690234423969e-05,
      "loss": 0.3433,
      "step": 142500
    },
    {
      "epoch": 5.98,
      "learning_rate": 2.0122435334921233e-05,
      "loss": 0.3431,
      "step": 143000
    },
    {
      "epoch": 6.0,
      "learning_rate": 2.0017968325602775e-05,
      "loss": 0.3432,
      "step": 143500
    },
    {
      "epoch": 6.0,
      "eval_accuracy": 0.8926879801904946,
      "eval_loss": 0.33457934856414795,
      "eval_runtime": 1401.0979,
      "eval_samples_per_second": 272.975,
      "eval_steps_per_second": 1.896,
      "step": 143586
    },
    {
      "epoch": 6.02,
      "learning_rate": 1.991350131628432e-05,
      "loss": 0.3427,
      "step": 144000
    },
    {
      "epoch": 6.04,
      "learning_rate": 1.980903430696586e-05,
      "loss": 0.3424,
      "step": 144500
    },
    {
      "epoch": 6.06,
      "learning_rate": 1.9704567297647402e-05,
      "loss": 0.3425,
      "step": 145000
    },
    {
      "epoch": 6.08,
      "learning_rate": 1.9600100288328947e-05,
      "loss": 0.3421,
      "step": 145500
    },
    {
      "epoch": 6.1,
      "learning_rate": 1.9495633279010488e-05,
      "loss": 0.3423,
      "step": 146000
    },
    {
      "epoch": 6.12,
      "learning_rate": 1.9391166269692033e-05,
      "loss": 0.3425,
      "step": 146500
    },
    {
      "epoch": 6.14,
      "learning_rate": 1.9286699260373574e-05,
      "loss": 0.3425,
      "step": 147000
    },
    {
      "epoch": 6.16,
      "learning_rate": 1.918223225105512e-05,
      "loss": 0.342,
      "step": 147500
    },
    {
      "epoch": 6.18,
      "learning_rate": 1.907776524173666e-05,
      "loss": 0.3423,
      "step": 148000
    },
    {
      "epoch": 6.21,
      "learning_rate": 1.8973298232418205e-05,
      "loss": 0.3424,
      "step": 148500
    },
    {
      "epoch": 6.23,
      "learning_rate": 1.8868831223099747e-05,
      "loss": 0.3421,
      "step": 149000
    },
    {
      "epoch": 6.25,
      "learning_rate": 1.8764364213781288e-05,
      "loss": 0.342,
      "step": 149500
    },
    {
      "epoch": 6.27,
      "learning_rate": 1.8659897204462833e-05,
      "loss": 0.3423,
      "step": 150000
    },
    {
      "epoch": 6.29,
      "learning_rate": 1.8555430195144374e-05,
      "loss": 0.3423,
      "step": 150500
    },
    {
      "epoch": 6.31,
      "learning_rate": 1.845096318582592e-05,
      "loss": 0.342,
      "step": 151000
    },
    {
      "epoch": 6.33,
      "learning_rate": 1.834649617650746e-05,
      "loss": 0.3421,
      "step": 151500
    },
    {
      "epoch": 6.35,
      "learning_rate": 1.8242029167189005e-05,
      "loss": 0.3421,
      "step": 152000
    },
    {
      "epoch": 6.37,
      "learning_rate": 1.8137562157870543e-05,
      "loss": 0.342,
      "step": 152500
    },
    {
      "epoch": 6.39,
      "learning_rate": 1.8033095148552088e-05,
      "loss": 0.3421,
      "step": 153000
    },
    {
      "epoch": 6.41,
      "learning_rate": 1.792862813923363e-05,
      "loss": 0.342,
      "step": 153500
    },
    {
      "epoch": 6.44,
      "learning_rate": 1.7824161129915174e-05,
      "loss": 0.3418,
      "step": 154000
    },
    {
      "epoch": 6.46,
      "learning_rate": 1.7719694120596715e-05,
      "loss": 0.342,
      "step": 154500
    },
    {
      "epoch": 6.48,
      "learning_rate": 1.761522711127826e-05,
      "loss": 0.3419,
      "step": 155000
    },
    {
      "epoch": 6.5,
      "learning_rate": 1.75107601019598e-05,
      "loss": 0.3419,
      "step": 155500
    },
    {
      "epoch": 6.52,
      "learning_rate": 1.7406293092641346e-05,
      "loss": 0.3415,
      "step": 156000
    },
    {
      "epoch": 6.54,
      "learning_rate": 1.7301826083322887e-05,
      "loss": 0.3417,
      "step": 156500
    },
    {
      "epoch": 6.56,
      "learning_rate": 1.719735907400443e-05,
      "loss": 0.3417,
      "step": 157000
    },
    {
      "epoch": 6.58,
      "learning_rate": 1.7092892064685973e-05,
      "loss": 0.3421,
      "step": 157500
    },
    {
      "epoch": 6.6,
      "learning_rate": 1.6988425055367515e-05,
      "loss": 0.3415,
      "step": 158000
    },
    {
      "epoch": 6.62,
      "learning_rate": 1.688395804604906e-05,
      "loss": 0.3415,
      "step": 158500
    },
    {
      "epoch": 6.64,
      "learning_rate": 1.67794910367306e-05,
      "loss": 0.3416,
      "step": 159000
    },
    {
      "epoch": 6.66,
      "learning_rate": 1.6675024027412146e-05,
      "loss": 0.3414,
      "step": 159500
    },
    {
      "epoch": 6.69,
      "learning_rate": 1.6570557018093687e-05,
      "loss": 0.3415,
      "step": 160000
    },
    {
      "epoch": 6.71,
      "learning_rate": 1.6466090008775228e-05,
      "loss": 0.3416,
      "step": 160500
    },
    {
      "epoch": 6.73,
      "learning_rate": 1.6361622999456773e-05,
      "loss": 0.3415,
      "step": 161000
    },
    {
      "epoch": 6.75,
      "learning_rate": 1.6257155990138314e-05,
      "loss": 0.3414,
      "step": 161500
    },
    {
      "epoch": 6.77,
      "learning_rate": 1.615268898081986e-05,
      "loss": 0.3414,
      "step": 162000
    },
    {
      "epoch": 6.79,
      "learning_rate": 1.60482219715014e-05,
      "loss": 0.3414,
      "step": 162500
    },
    {
      "epoch": 6.81,
      "learning_rate": 1.5943754962182945e-05,
      "loss": 0.3413,
      "step": 163000
    },
    {
      "epoch": 6.83,
      "learning_rate": 1.5839287952864487e-05,
      "loss": 0.3417,
      "step": 163500
    },
    {
      "epoch": 6.85,
      "learning_rate": 1.5734820943546028e-05,
      "loss": 0.3414,
      "step": 164000
    },
    {
      "epoch": 6.87,
      "learning_rate": 1.563035393422757e-05,
      "loss": 0.3413,
      "step": 164500
    },
    {
      "epoch": 6.89,
      "learning_rate": 1.5525886924909114e-05,
      "loss": 0.3412,
      "step": 165000
    },
    {
      "epoch": 6.92,
      "learning_rate": 1.5421419915590655e-05,
      "loss": 0.3413,
      "step": 165500
    },
    {
      "epoch": 6.94,
      "learning_rate": 1.53169529062722e-05,
      "loss": 0.3412,
      "step": 166000
    },
    {
      "epoch": 6.96,
      "learning_rate": 1.5212485896953743e-05,
      "loss": 0.3412,
      "step": 166500
    },
    {
      "epoch": 6.98,
      "learning_rate": 1.5108018887635286e-05,
      "loss": 0.3412,
      "step": 167000
    },
    {
      "epoch": 7.0,
      "learning_rate": 1.500355187831683e-05,
      "loss": 0.3412,
      "step": 167500
    },
    {
      "epoch": 7.0,
      "eval_accuracy": 0.8930347020010898,
      "eval_loss": 0.33325716853141785,
      "eval_runtime": 1406.6403,
      "eval_samples_per_second": 271.899,
      "eval_steps_per_second": 1.888,
      "step": 167517
    },
    {
      "epoch": 7.02,
      "learning_rate": 1.489908486899837e-05,
      "loss": 0.3407,
      "step": 168000
    },
    {
      "epoch": 7.04,
      "learning_rate": 1.4794617859679914e-05,
      "loss": 0.3407,
      "step": 168500
    },
    {
      "epoch": 7.06,
      "learning_rate": 1.4690150850361457e-05,
      "loss": 0.3406,
      "step": 169000
    },
    {
      "epoch": 7.08,
      "learning_rate": 1.4585683841043e-05,
      "loss": 0.3407,
      "step": 169500
    },
    {
      "epoch": 7.1,
      "learning_rate": 1.4481216831724543e-05,
      "loss": 0.3407,
      "step": 170000
    },
    {
      "epoch": 7.12,
      "learning_rate": 1.4376749822406086e-05,
      "loss": 0.3406,
      "step": 170500
    },
    {
      "epoch": 7.15,
      "learning_rate": 1.4272282813087629e-05,
      "loss": 0.3406,
      "step": 171000
    },
    {
      "epoch": 7.17,
      "learning_rate": 1.4167815803769172e-05,
      "loss": 0.3406,
      "step": 171500
    },
    {
      "epoch": 7.19,
      "learning_rate": 1.4063348794450712e-05,
      "loss": 0.3407,
      "step": 172000
    },
    {
      "epoch": 7.21,
      "learning_rate": 1.3958881785132255e-05,
      "loss": 0.3406,
      "step": 172500
    },
    {
      "epoch": 7.23,
      "learning_rate": 1.3854414775813798e-05,
      "loss": 0.3404,
      "step": 173000
    },
    {
      "epoch": 7.25,
      "learning_rate": 1.374994776649534e-05,
      "loss": 0.3406,
      "step": 173500
    },
    {
      "epoch": 7.27,
      "learning_rate": 1.3645480757176884e-05,
      "loss": 0.3405,
      "step": 174000
    },
    {
      "epoch": 7.29,
      "learning_rate": 1.3541013747858427e-05,
      "loss": 0.3405,
      "step": 174500
    },
    {
      "epoch": 7.31,
      "learning_rate": 1.343654673853997e-05,
      "loss": 0.3402,
      "step": 175000
    },
    {
      "epoch": 7.33,
      "learning_rate": 1.3332079729221515e-05,
      "loss": 0.3404,
      "step": 175500
    },
    {
      "epoch": 7.35,
      "learning_rate": 1.3227612719903054e-05,
      "loss": 0.3404,
      "step": 176000
    },
    {
      "epoch": 7.38,
      "learning_rate": 1.3123145710584597e-05,
      "loss": 0.3403,
      "step": 176500
    },
    {
      "epoch": 7.4,
      "learning_rate": 1.301867870126614e-05,
      "loss": 0.3405,
      "step": 177000
    },
    {
      "epoch": 7.42,
      "learning_rate": 1.2914211691947683e-05,
      "loss": 0.3404,
      "step": 177500
    },
    {
      "epoch": 7.44,
      "learning_rate": 1.2809744682629226e-05,
      "loss": 0.3405,
      "step": 178000
    },
    {
      "epoch": 7.46,
      "learning_rate": 1.270527767331077e-05,
      "loss": 0.3403,
      "step": 178500
    },
    {
      "epoch": 7.48,
      "learning_rate": 1.2600810663992313e-05,
      "loss": 0.3401,
      "step": 179000
    },
    {
      "epoch": 7.5,
      "learning_rate": 1.2496343654673854e-05,
      "loss": 0.3404,
      "step": 179500
    },
    {
      "epoch": 7.52,
      "learning_rate": 1.2391876645355397e-05,
      "loss": 0.34,
      "step": 180000
    },
    {
      "epoch": 7.54,
      "learning_rate": 1.228740963603694e-05,
      "loss": 0.3402,
      "step": 180500
    },
    {
      "epoch": 7.56,
      "learning_rate": 1.2182942626718483e-05,
      "loss": 0.3402,
      "step": 181000
    },
    {
      "epoch": 7.58,
      "learning_rate": 1.2078475617400026e-05,
      "loss": 0.3401,
      "step": 181500
    },
    {
      "epoch": 7.61,
      "learning_rate": 1.197400860808157e-05,
      "loss": 0.3402,
      "step": 182000
    },
    {
      "epoch": 7.63,
      "learning_rate": 1.186954159876311e-05,
      "loss": 0.3403,
      "step": 182500
    },
    {
      "epoch": 7.65,
      "learning_rate": 1.1765074589444654e-05,
      "loss": 0.3401,
      "step": 183000
    },
    {
      "epoch": 7.67,
      "learning_rate": 1.1660607580126197e-05,
      "loss": 0.3401,
      "step": 183500
    },
    {
      "epoch": 7.69,
      "learning_rate": 1.155614057080774e-05,
      "loss": 0.3399,
      "step": 184000
    },
    {
      "epoch": 7.71,
      "learning_rate": 1.1451673561489281e-05,
      "loss": 0.34,
      "step": 184500
    },
    {
      "epoch": 7.73,
      "learning_rate": 1.1347206552170824e-05,
      "loss": 0.3402,
      "step": 185000
    },
    {
      "epoch": 7.75,
      "learning_rate": 1.1242739542852367e-05,
      "loss": 0.34,
      "step": 185500
    },
    {
      "epoch": 7.77,
      "learning_rate": 1.1138272533533912e-05,
      "loss": 0.34,
      "step": 186000
    },
    {
      "epoch": 7.79,
      "learning_rate": 1.1033805524215453e-05,
      "loss": 0.34,
      "step": 186500
    },
    {
      "epoch": 7.81,
      "learning_rate": 1.0929338514896996e-05,
      "loss": 0.3402,
      "step": 187000
    },
    {
      "epoch": 7.84,
      "learning_rate": 1.082487150557854e-05,
      "loss": 0.34,
      "step": 187500
    },
    {
      "epoch": 7.86,
      "learning_rate": 1.0720404496260082e-05,
      "loss": 0.3397,
      "step": 188000
    },
    {
      "epoch": 7.88,
      "learning_rate": 1.0615937486941624e-05,
      "loss": 0.3398,
      "step": 188500
    },
    {
      "epoch": 7.9,
      "learning_rate": 1.0511470477623167e-05,
      "loss": 0.3399,
      "step": 189000
    },
    {
      "epoch": 7.92,
      "learning_rate": 1.040700346830471e-05,
      "loss": 0.3398,
      "step": 189500
    },
    {
      "epoch": 7.94,
      "learning_rate": 1.0302536458986253e-05,
      "loss": 0.3399,
      "step": 190000
    },
    {
      "epoch": 7.96,
      "learning_rate": 1.0198069449667794e-05,
      "loss": 0.3399,
      "step": 190500
    },
    {
      "epoch": 7.98,
      "learning_rate": 1.0093602440349337e-05,
      "loss": 0.3397,
      "step": 191000
    },
    {
      "epoch": 8.0,
      "eval_accuracy": 0.8933395795756411,
      "eval_loss": 0.33218324184417725,
      "eval_runtime": 1414.7217,
      "eval_samples_per_second": 270.346,
      "eval_steps_per_second": 1.877,
      "step": 191448
    },
    {
      "epoch": 8.0,
      "learning_rate": 9.989135431030882e-06,
      "loss": 0.3397,
      "step": 191500
    },
    {
      "epoch": 8.02,
      "learning_rate": 9.884668421712423e-06,
      "loss": 0.3393,
      "step": 192000
    },
    {
      "epoch": 8.04,
      "learning_rate": 9.780201412393966e-06,
      "loss": 0.3395,
      "step": 192500
    },
    {
      "epoch": 8.06,
      "learning_rate": 9.67573440307551e-06,
      "loss": 0.3394,
      "step": 193000
    },
    {
      "epoch": 8.09,
      "learning_rate": 9.571267393757053e-06,
      "loss": 0.3394,
      "step": 193500
    },
    {
      "epoch": 8.11,
      "learning_rate": 9.466800384438594e-06,
      "loss": 0.3396,
      "step": 194000
    },
    {
      "epoch": 8.13,
      "learning_rate": 9.362333375120137e-06,
      "loss": 0.3391,
      "step": 194500
    },
    {
      "epoch": 8.15,
      "learning_rate": 9.25786636580168e-06,
      "loss": 0.3393,
      "step": 195000
    },
    {
      "epoch": 8.17,
      "learning_rate": 9.153399356483223e-06,
      "loss": 0.3393,
      "step": 195500
    },
    {
      "epoch": 8.19,
      "learning_rate": 9.048932347164764e-06,
      "loss": 0.3392,
      "step": 196000
    },
    {
      "epoch": 8.21,
      "learning_rate": 8.94446533784631e-06,
      "loss": 0.3393,
      "step": 196500
    },
    {
      "epoch": 8.23,
      "learning_rate": 8.839998328527852e-06,
      "loss": 0.3395,
      "step": 197000
    },
    {
      "epoch": 8.25,
      "learning_rate": 8.735531319209395e-06,
      "loss": 0.3393,
      "step": 197500
    },
    {
      "epoch": 8.27,
      "learning_rate": 8.631064309890937e-06,
      "loss": 0.3394,
      "step": 198000
    },
    {
      "epoch": 8.29,
      "learning_rate": 8.52659730057248e-06,
      "loss": 0.3391,
      "step": 198500
    },
    {
      "epoch": 8.32,
      "learning_rate": 8.422130291254023e-06,
      "loss": 0.3392,
      "step": 199000
    },
    {
      "epoch": 8.34,
      "learning_rate": 8.317663281935566e-06,
      "loss": 0.3394,
      "step": 199500
    },
    {
      "epoch": 8.36,
      "learning_rate": 8.213196272617107e-06,
      "loss": 0.3389,
      "step": 200000
    },
    {
      "epoch": 8.38,
      "learning_rate": 8.10872926329865e-06,
      "loss": 0.3392,
      "step": 200500
    },
    {
      "epoch": 8.4,
      "learning_rate": 8.004262253980193e-06,
      "loss": 0.3394,
      "step": 201000
    },
    {
      "epoch": 8.42,
      "learning_rate": 7.899795244661736e-06,
      "loss": 0.3389,
      "step": 201500
    },
    {
      "epoch": 8.44,
      "learning_rate": 7.79532823534328e-06,
      "loss": 0.3389,
      "step": 202000
    },
    {
      "epoch": 8.46,
      "learning_rate": 7.690861226024822e-06,
      "loss": 0.339,
      "step": 202500
    },
    {
      "epoch": 8.48,
      "learning_rate": 7.586394216706365e-06,
      "loss": 0.3392,
      "step": 203000
    },
    {
      "epoch": 8.5,
      "learning_rate": 7.481927207387908e-06,
      "loss": 0.339,
      "step": 203500
    },
    {
      "epoch": 8.52,
      "learning_rate": 7.37746019806945e-06,
      "loss": 0.3391,
      "step": 204000
    },
    {
      "epoch": 8.55,
      "learning_rate": 7.272993188750993e-06,
      "loss": 0.339,
      "step": 204500
    },
    {
      "epoch": 8.57,
      "learning_rate": 7.168526179432536e-06,
      "loss": 0.3391,
      "step": 205000
    },
    {
      "epoch": 8.59,
      "learning_rate": 7.064059170114077e-06,
      "loss": 0.339,
      "step": 205500
    },
    {
      "epoch": 8.61,
      "learning_rate": 6.95959216079562e-06,
      "loss": 0.339,
      "step": 206000
    },
    {
      "epoch": 8.63,
      "learning_rate": 6.855125151477164e-06,
      "loss": 0.3391,
      "step": 206500
    },
    {
      "epoch": 8.65,
      "learning_rate": 6.750658142158707e-06,
      "loss": 0.339,
      "step": 207000
    },
    {
      "epoch": 8.67,
      "learning_rate": 6.646191132840249e-06,
      "loss": 0.3389,
      "step": 207500
    },
    {
      "epoch": 8.69,
      "learning_rate": 6.541724123521792e-06,
      "loss": 0.3391,
      "step": 208000
    },
    {
      "epoch": 8.71,
      "learning_rate": 6.437257114203335e-06,
      "loss": 0.339,
      "step": 208500
    },
    {
      "epoch": 8.73,
      "learning_rate": 6.332790104884878e-06,
      "loss": 0.339,
      "step": 209000
    },
    {
      "epoch": 8.75,
      "learning_rate": 6.228323095566421e-06,
      "loss": 0.3391,
      "step": 209500
    },
    {
      "epoch": 8.78,
      "learning_rate": 6.123856086247963e-06,
      "loss": 0.3388,
      "step": 210000
    },
    {
      "epoch": 8.8,
      "learning_rate": 6.019389076929506e-06,
      "loss": 0.3393,
      "step": 210500
    },
    {
      "epoch": 8.82,
      "learning_rate": 5.914922067611048e-06,
      "loss": 0.3389,
      "step": 211000
    },
    {
      "epoch": 8.84,
      "learning_rate": 5.810455058292591e-06,
      "loss": 0.3392,
      "step": 211500
    },
    {
      "epoch": 8.86,
      "learning_rate": 5.705988048974134e-06,
      "loss": 0.3391,
      "step": 212000
    },
    {
      "epoch": 8.88,
      "learning_rate": 5.6015210396556775e-06,
      "loss": 0.339,
      "step": 212500
    },
    {
      "epoch": 8.9,
      "learning_rate": 5.49705403033722e-06,
      "loss": 0.3389,
      "step": 213000
    },
    {
      "epoch": 8.92,
      "learning_rate": 5.392587021018763e-06,
      "loss": 0.3389,
      "step": 213500
    },
    {
      "epoch": 8.94,
      "learning_rate": 5.288120011700305e-06,
      "loss": 0.339,
      "step": 214000
    },
    {
      "epoch": 8.96,
      "learning_rate": 5.183653002381848e-06,
      "loss": 0.3386,
      "step": 214500
    },
    {
      "epoch": 8.98,
      "learning_rate": 5.079185993063391e-06,
      "loss": 0.339,
      "step": 215000
    },
    {
      "epoch": 9.0,
      "eval_accuracy": 0.8935473987846363,
      "eval_loss": 0.33142516016960144,
      "eval_runtime": 1417.4586,
      "eval_samples_per_second": 269.824,
      "eval_steps_per_second": 1.874,
      "step": 215379
    },
    {
      "epoch": 9.01,
      "learning_rate": 4.974718983744934e-06,
      "loss": 0.3389,
      "step": 215500
    },
    {
      "epoch": 9.03,
      "learning_rate": 4.870251974426476e-06,
      "loss": 0.3384,
      "step": 216000
    },
    {
      "epoch": 9.05,
      "learning_rate": 4.765784965108019e-06,
      "loss": 0.3383,
      "step": 216500
    },
    {
      "epoch": 9.07,
      "learning_rate": 4.6613179557895615e-06,
      "loss": 0.3385,
      "step": 217000
    },
    {
      "epoch": 9.09,
      "learning_rate": 4.5568509464711046e-06,
      "loss": 0.3383,
      "step": 217500
    },
    {
      "epoch": 9.11,
      "learning_rate": 4.452383937152648e-06,
      "loss": 0.3387,
      "step": 218000
    },
    {
      "epoch": 9.13,
      "learning_rate": 4.34791692783419e-06,
      "loss": 0.3386,
      "step": 218500
    },
    {
      "epoch": 9.15,
      "learning_rate": 4.243449918515733e-06,
      "loss": 0.3382,
      "step": 219000
    },
    {
      "epoch": 9.17,
      "learning_rate": 4.138982909197275e-06,
      "loss": 0.3385,
      "step": 219500
    },
    {
      "epoch": 9.19,
      "learning_rate": 4.034515899878818e-06,
      "loss": 0.3384,
      "step": 220000
    },
    {
      "epoch": 9.21,
      "learning_rate": 3.930048890560361e-06,
      "loss": 0.3386,
      "step": 220500
    },
    {
      "epoch": 9.23,
      "learning_rate": 3.825581881241904e-06,
      "loss": 0.3387,
      "step": 221000
    },
    {
      "epoch": 9.26,
      "learning_rate": 3.7211148719234464e-06,
      "loss": 0.3383,
      "step": 221500
    },
    {
      "epoch": 9.28,
      "learning_rate": 3.6166478626049895e-06,
      "loss": 0.3383,
      "step": 222000
    },
    {
      "epoch": 9.3,
      "learning_rate": 3.512180853286532e-06,
      "loss": 0.3384,
      "step": 222500
    },
    {
      "epoch": 9.32,
      "learning_rate": 3.407713843968075e-06,
      "loss": 0.3385,
      "step": 223000
    },
    {
      "epoch": 9.34,
      "learning_rate": 3.3032468346496178e-06,
      "loss": 0.3386,
      "step": 223500
    },
    {
      "epoch": 9.36,
      "learning_rate": 3.198779825331161e-06,
      "loss": 0.3385,
      "step": 224000
    },
    {
      "epoch": 9.38,
      "learning_rate": 3.0943128160127035e-06,
      "loss": 0.3387,
      "step": 224500
    },
    {
      "epoch": 9.4,
      "learning_rate": 2.989845806694246e-06,
      "loss": 0.3384,
      "step": 225000
    },
    {
      "epoch": 9.42,
      "learning_rate": 2.8853787973757887e-06,
      "loss": 0.3381,
      "step": 225500
    },
    {
      "epoch": 9.44,
      "learning_rate": 2.7809117880573313e-06,
      "loss": 0.3384,
      "step": 226000
    },
    {
      "epoch": 9.46,
      "learning_rate": 2.6764447787388744e-06,
      "loss": 0.3386,
      "step": 226500
    },
    {
      "epoch": 9.49,
      "learning_rate": 2.571977769420417e-06,
      "loss": 0.3384,
      "step": 227000
    },
    {
      "epoch": 9.51,
      "learning_rate": 2.4675107601019596e-06,
      "loss": 0.3381,
      "step": 227500
    },
    {
      "epoch": 9.53,
      "learning_rate": 2.3630437507835027e-06,
      "loss": 0.3383,
      "step": 228000
    },
    {
      "epoch": 9.55,
      "learning_rate": 2.2585767414650453e-06,
      "loss": 0.3385,
      "step": 228500
    },
    {
      "epoch": 9.57,
      "learning_rate": 2.154109732146588e-06,
      "loss": 0.3383,
      "step": 229000
    },
    {
      "epoch": 9.59,
      "learning_rate": 2.049642722828131e-06,
      "loss": 0.3385,
      "step": 229500
    },
    {
      "epoch": 9.61,
      "learning_rate": 1.9451757135096736e-06,
      "loss": 0.3383,
      "step": 230000
    },
    {
      "epoch": 9.63,
      "learning_rate": 1.8407087041912165e-06,
      "loss": 0.3384,
      "step": 230500
    },
    {
      "epoch": 9.65,
      "learning_rate": 1.7362416948727593e-06,
      "loss": 0.3384,
      "step": 231000
    },
    {
      "epoch": 9.67,
      "learning_rate": 1.631774685554302e-06,
      "loss": 0.3383,
      "step": 231500
    },
    {
      "epoch": 9.69,
      "learning_rate": 1.5273076762358448e-06,
      "loss": 0.3384,
      "step": 232000
    },
    {
      "epoch": 9.72,
      "learning_rate": 1.4228406669173876e-06,
      "loss": 0.3384,
      "step": 232500
    },
    {
      "epoch": 9.74,
      "learning_rate": 1.3183736575989304e-06,
      "loss": 0.3385,
      "step": 233000
    },
    {
      "epoch": 9.76,
      "learning_rate": 1.213906648280473e-06,
      "loss": 0.3383,
      "step": 233500
    },
    {
      "epoch": 9.78,
      "learning_rate": 1.1094396389620159e-06,
      "loss": 0.3388,
      "step": 234000
    },
    {
      "epoch": 9.8,
      "learning_rate": 1.0049726296435587e-06,
      "loss": 0.3383,
      "step": 234500
    },
    {
      "epoch": 9.82,
      "learning_rate": 9.005056203251015e-07,
      "loss": 0.3383,
      "step": 235000
    },
    {
      "epoch": 9.84,
      "learning_rate": 7.960386110066441e-07,
      "loss": 0.3382,
      "step": 235500
    },
    {
      "epoch": 9.86,
      "learning_rate": 6.915716016881869e-07,
      "loss": 0.338,
      "step": 236000
    },
    {
      "epoch": 9.88,
      "learning_rate": 5.871045923697297e-07,
      "loss": 0.3382,
      "step": 236500
    },
    {
      "epoch": 9.9,
      "learning_rate": 4.826375830512725e-07,
      "loss": 0.3383,
      "step": 237000
    },
    {
      "epoch": 9.92,
      "learning_rate": 3.7817057373281523e-07,
      "loss": 0.3383,
      "step": 237500
    },
    {
      "epoch": 9.95,
      "learning_rate": 2.7370356441435796e-07,
      "loss": 0.3386,
      "step": 238000
    },
    {
      "epoch": 9.97,
      "learning_rate": 1.6923655509590072e-07,
      "loss": 0.3382,
      "step": 238500
    },
    {
      "epoch": 9.99,
      "learning_rate": 6.476954577744348e-08,
      "loss": 0.3383,
      "step": 239000
    },
    {
      "epoch": 10.0,
      "eval_accuracy": 0.8936249984035948,
      "eval_loss": 0.3311329483985901,
      "eval_runtime": 1436.3126,
      "eval_samples_per_second": 266.282,
      "eval_steps_per_second": 1.849,
      "step": 239310
    },
    {
      "epoch": 10.0,
      "step": 239310,
      "total_flos": 9.004290600882668e+18,
      "train_loss": 0.35271068150736834,
      "train_runtime": 202889.1723,
      "train_samples_per_second": 169.846,
      "train_steps_per_second": 1.18
    }
  ],
  "logging_steps": 500,
  "max_steps": 239310,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 9.004290600882668e+18,
  "train_batch_size": 24,
  "trial_name": null,
  "trial_params": null
}