{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.979843490633152,
  "global_step": 168000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 4.9925895186151295e-05,
      "loss": 1.6378,
      "step": 500
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.985179037230259e-05,
      "loss": 1.6723,
      "step": 1000
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.977768555845388e-05,
      "loss": 1.6694,
      "step": 1500
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.970358074460517e-05,
      "loss": 1.6685,
      "step": 2000
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.9629475930756463e-05,
      "loss": 1.6812,
      "step": 2500
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.9555371116907756e-05,
      "loss": 1.6792,
      "step": 3000
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.948126630305905e-05,
      "loss": 1.6873,
      "step": 3500
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.940716148921034e-05,
      "loss": 1.7036,
      "step": 4000
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.866611335072326e-05,
      "loss": 1.7003,
      "step": 4500
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.8517903723025847e-05,
      "loss": 1.6828,
      "step": 5000
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.836969409532843e-05,
      "loss": 1.6904,
      "step": 5500
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.8221484467631015e-05,
      "loss": 1.6975,
      "step": 6000
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.80732748399336e-05,
      "loss": 1.6904,
      "step": 6500
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.7925065212236184e-05,
      "loss": 1.6979,
      "step": 7000
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.7776855584538775e-05,
      "loss": 1.6774,
      "step": 7500
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.762864595684136e-05,
      "loss": 1.6976,
      "step": 8000
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.7480436329143944e-05,
      "loss": 1.715,
      "step": 8500
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.733222670144653e-05,
      "loss": 1.7125,
      "step": 9000
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.718401707374911e-05,
      "loss": 1.7065,
      "step": 9500
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.70358074460517e-05,
      "loss": 1.7062,
      "step": 10000
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.688759781835428e-05,
      "loss": 1.7038,
      "step": 10500
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.6739388190656866e-05,
      "loss": 1.7092,
      "step": 11000
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.659117856295945e-05,
      "loss": 1.7377,
      "step": 11500
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.6442968935262035e-05,
      "loss": 1.7306,
      "step": 12000
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.629475930756462e-05,
      "loss": 1.7128,
      "step": 12500
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.6146549679867204e-05,
      "loss": 1.7324,
      "step": 13000
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.599834005216979e-05,
      "loss": 1.7483,
      "step": 13500
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.585013042447238e-05,
      "loss": 1.7351,
      "step": 14000
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.5701920796774964e-05,
      "loss": 1.7238,
      "step": 14500
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.555371116907755e-05,
      "loss": 1.7315,
      "step": 15000
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.540550154138013e-05,
      "loss": 1.7391,
      "step": 15500
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.525729191368272e-05,
      "loss": 1.7384,
      "step": 16000
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.51090822859853e-05,
      "loss": 1.7613,
      "step": 16500
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.4960872658287886e-05,
      "loss": 1.7558,
      "step": 17000
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.481266303059047e-05,
      "loss": 1.7353,
      "step": 17500
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.4664453402893055e-05,
      "loss": 1.7513,
      "step": 18000
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.451624377519564e-05,
      "loss": 1.7365,
      "step": 18500
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.4368034147498224e-05,
      "loss": 1.7428,
      "step": 19000
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.421982451980081e-05,
      "loss": 1.7352,
      "step": 19500
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.407161489210339e-05,
      "loss": 1.7561,
      "step": 20000
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.392340526440598e-05,
      "loss": 1.7478,
      "step": 20500
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.377519563670856e-05,
      "loss": 1.7504,
      "step": 21000
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.3626986009011146e-05,
      "loss": 1.7484,
      "step": 21500
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.347877638131374e-05,
      "loss": 1.7685,
      "step": 22000
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.333056675361632e-05,
      "loss": 1.743,
      "step": 22500
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.3182357125918906e-05,
      "loss": 1.7624,
      "step": 23000
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.303414749822149e-05,
      "loss": 1.7769,
      "step": 23500
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.2885937870524075e-05,
      "loss": 1.7504,
      "step": 24000
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.273772824282666e-05,
      "loss": 1.7516,
      "step": 24500
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.258951861512924e-05,
      "loss": 1.7763,
      "step": 25000
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.244130898743183e-05,
      "loss": 1.7626,
      "step": 25500
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.229309935973441e-05,
      "loss": 1.7935,
      "step": 26000
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.2144889732036997e-05,
      "loss": 1.7821,
      "step": 26500
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.199668010433958e-05,
      "loss": 1.7785,
      "step": 27000
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.1848470476642165e-05,
      "loss": 1.7732,
      "step": 27500
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.170026084894475e-05,
      "loss": 1.7696,
      "step": 28000
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.1552051221247334e-05,
      "loss": 1.7821,
      "step": 28500
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.140384159354992e-05,
      "loss": 1.7855,
      "step": 29000
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.12556319658525e-05,
      "loss": 1.7815,
      "step": 29500
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.110742233815509e-05,
      "loss": 1.7776,
      "step": 30000
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.095921271045767e-05,
      "loss": 1.7864,
      "step": 30500
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.0811003082760256e-05,
      "loss": 1.7894,
      "step": 31000
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.066279345506284e-05,
      "loss": 1.7879,
      "step": 31500
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.0514583827365425e-05,
      "loss": 1.7928,
      "step": 32000
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.036637419966801e-05,
      "loss": 1.7965,
      "step": 32500
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.0218164571970594e-05,
      "loss": 1.7906,
      "step": 33000
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.006995494427318e-05,
      "loss": 1.8053,
      "step": 33500
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.992174531657576e-05,
      "loss": 1.6997,
      "step": 34000
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.977353568887835e-05,
      "loss": 1.627,
      "step": 34500
    },
    {
      "epoch": 1.04,
      "learning_rate": 3.962532606118093e-05,
      "loss": 1.6205,
      "step": 35000
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.9477116433483516e-05,
      "loss": 1.624,
      "step": 35500
    },
    {
      "epoch": 1.07,
      "learning_rate": 3.93289068057861e-05,
      "loss": 1.6387,
      "step": 36000
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.9180697178088685e-05,
      "loss": 1.632,
      "step": 36500
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.9032487550391276e-05,
      "loss": 1.6339,
      "step": 37000
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.888427792269386e-05,
      "loss": 1.6487,
      "step": 37500
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.8736068294996445e-05,
      "loss": 1.6449,
      "step": 38000
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.858785866729903e-05,
      "loss": 1.6598,
      "step": 38500
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.8439649039601614e-05,
      "loss": 1.6621,
      "step": 39000
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.82914394119042e-05,
      "loss": 1.6735,
      "step": 39500
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.814322978420678e-05,
      "loss": 1.665,
      "step": 40000
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.799502015650937e-05,
      "loss": 1.6542,
      "step": 40500
    },
    {
      "epoch": 1.22,
      "learning_rate": 3.784681052881195e-05,
      "loss": 1.6632,
      "step": 41000
    },
    {
      "epoch": 1.23,
      "learning_rate": 3.7698600901114536e-05,
      "loss": 1.6096,
      "step": 41500
    },
    {
      "epoch": 1.24,
      "learning_rate": 3.755039127341713e-05,
      "loss": 1.5457,
      "step": 42000
    },
    {
      "epoch": 1.26,
      "learning_rate": 3.740218164571971e-05,
      "loss": 1.538,
      "step": 42500
    },
    {
      "epoch": 1.27,
      "learning_rate": 3.7253972018022296e-05,
      "loss": 1.5538,
      "step": 43000
    },
    {
      "epoch": 1.29,
      "learning_rate": 3.710576239032488e-05,
      "loss": 1.5505,
      "step": 43500
    },
    {
      "epoch": 1.3,
      "learning_rate": 3.6957552762627465e-05,
      "loss": 1.5386,
      "step": 44000
    },
    {
      "epoch": 1.32,
      "learning_rate": 3.680934313493005e-05,
      "loss": 1.5629,
      "step": 44500
    },
    {
      "epoch": 1.33,
      "learning_rate": 3.6661133507232633e-05,
      "loss": 1.5639,
      "step": 45000
    },
    {
      "epoch": 1.35,
      "learning_rate": 3.651292387953522e-05,
      "loss": 1.6376,
      "step": 45500
    },
    {
      "epoch": 1.36,
      "learning_rate": 3.63647142518378e-05,
      "loss": 1.6273,
      "step": 46000
    },
    {
      "epoch": 1.38,
      "learning_rate": 3.621650462414039e-05,
      "loss": 1.6433,
      "step": 46500
    },
    {
      "epoch": 1.39,
      "learning_rate": 3.606829499644297e-05,
      "loss": 1.6421,
      "step": 47000
    },
    {
      "epoch": 1.41,
      "learning_rate": 3.5920085368745556e-05,
      "loss": 1.6529,
      "step": 47500
    },
    {
      "epoch": 1.42,
      "learning_rate": 3.577187574104814e-05,
      "loss": 1.6501,
      "step": 48000
    },
    {
      "epoch": 1.44,
      "learning_rate": 3.5623666113350724e-05,
      "loss": 1.6568,
      "step": 48500
    },
    {
      "epoch": 1.45,
      "learning_rate": 3.547545648565331e-05,
      "loss": 1.6454,
      "step": 49000
    },
    {
      "epoch": 1.47,
      "learning_rate": 3.532724685795589e-05,
      "loss": 1.6698,
      "step": 49500
    },
    {
      "epoch": 1.48,
      "learning_rate": 3.517903723025848e-05,
      "loss": 1.672,
      "step": 50000
    },
    {
      "epoch": 1.5,
      "learning_rate": 3.503082760256106e-05,
      "loss": 1.6761,
      "step": 50500
    },
    {
      "epoch": 1.51,
      "learning_rate": 3.4882617974863646e-05,
      "loss": 1.6686,
      "step": 51000
    },
    {
      "epoch": 1.53,
      "learning_rate": 3.473440834716624e-05,
      "loss": 1.6736,
      "step": 51500
    },
    {
      "epoch": 1.54,
      "learning_rate": 3.458619871946882e-05,
      "loss": 1.6939,
      "step": 52000
    },
    {
      "epoch": 1.56,
      "learning_rate": 3.4437989091771406e-05,
      "loss": 1.6905,
      "step": 52500
    },
    {
      "epoch": 1.57,
      "learning_rate": 3.428977946407399e-05,
      "loss": 1.7088,
      "step": 53000
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.4141569836376575e-05,
      "loss": 1.7015,
      "step": 53500
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.399336020867916e-05,
      "loss": 1.7107,
      "step": 54000
    },
    {
      "epoch": 1.62,
      "learning_rate": 3.3845150580981744e-05,
      "loss": 1.705,
      "step": 54500
    },
    {
      "epoch": 1.63,
      "learning_rate": 3.369694095328433e-05,
      "loss": 1.71,
      "step": 55000
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.354873132558691e-05,
      "loss": 1.7015,
      "step": 55500
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.34005216978895e-05,
      "loss": 1.6962,
      "step": 56000
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.325231207019208e-05,
      "loss": 1.7304,
      "step": 56500
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.3104102442494666e-05,
      "loss": 1.7234,
      "step": 57000
    },
    {
      "epoch": 1.7,
      "learning_rate": 3.295589281479725e-05,
      "loss": 1.7124,
      "step": 57500
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.2807683187099835e-05,
      "loss": 1.7161,
      "step": 58000
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.265947355940242e-05,
      "loss": 1.7298,
      "step": 58500
    },
    {
      "epoch": 1.75,
      "learning_rate": 3.2511263931705004e-05,
      "loss": 1.7463,
      "step": 59000
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.236305430400759e-05,
      "loss": 1.7358,
      "step": 59500
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.221484467631017e-05,
      "loss": 1.7545,
      "step": 60000
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.206663504861276e-05,
      "loss": 1.7504,
      "step": 60500
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.191842542091534e-05,
      "loss": 1.7291,
      "step": 61000
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.1770215793217926e-05,
      "loss": 1.7584,
      "step": 61500
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.162200616552051e-05,
      "loss": 1.7563,
      "step": 62000
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.1473796537823095e-05,
      "loss": 1.7621,
      "step": 62500
    },
    {
      "epoch": 1.87,
      "learning_rate": 3.132558691012568e-05,
      "loss": 1.7567,
      "step": 63000
    },
    {
      "epoch": 1.88,
      "learning_rate": 3.1177377282428263e-05,
      "loss": 1.7759,
      "step": 63500
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.102916765473085e-05,
      "loss": 1.7544,
      "step": 64000
    },
    {
      "epoch": 1.91,
      "learning_rate": 3.088095802703343e-05,
      "loss": 1.7663,
      "step": 64500
    },
    {
      "epoch": 1.93,
      "learning_rate": 3.073274839933602e-05,
      "loss": 1.7943,
      "step": 65000
    },
    {
      "epoch": 1.94,
      "learning_rate": 3.05845387716386e-05,
      "loss": 1.7706,
      "step": 65500
    },
    {
      "epoch": 1.96,
      "learning_rate": 3.0436329143941196e-05,
      "loss": 1.8033,
      "step": 66000
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.028811951624378e-05,
      "loss": 1.7962,
      "step": 66500
    },
    {
      "epoch": 1.99,
      "learning_rate": 3.0139909888546365e-05,
      "loss": 1.8109,
      "step": 67000
    },
    {
      "epoch": 2.0,
      "learning_rate": 2.999170026084895e-05,
      "loss": 1.7925,
      "step": 67500
    },
    {
      "epoch": 2.02,
      "learning_rate": 2.9843490633151533e-05,
      "loss": 1.5786,
      "step": 68000
    },
    {
      "epoch": 2.03,
      "learning_rate": 2.9695281005454118e-05,
      "loss": 1.5676,
      "step": 68500
    },
    {
      "epoch": 2.05,
      "learning_rate": 2.9547071377756702e-05,
      "loss": 1.5703,
      "step": 69000
    },
    {
      "epoch": 2.06,
      "learning_rate": 2.9398861750059287e-05,
      "loss": 1.566,
      "step": 69500
    },
    {
      "epoch": 2.07,
      "learning_rate": 2.925065212236187e-05,
      "loss": 1.5811,
      "step": 70000
    },
    {
      "epoch": 2.09,
      "learning_rate": 2.9102442494664455e-05,
      "loss": 1.5771,
      "step": 70500
    },
    {
      "epoch": 2.1,
      "learning_rate": 2.895423286696704e-05,
      "loss": 1.5829,
      "step": 71000
    },
    {
      "epoch": 2.12,
      "learning_rate": 2.8806023239269624e-05,
      "loss": 1.5809,
      "step": 71500
    },
    {
      "epoch": 2.13,
      "learning_rate": 2.865781361157221e-05,
      "loss": 1.5957,
      "step": 72000
    },
    {
      "epoch": 2.15,
      "learning_rate": 2.8509603983874793e-05,
      "loss": 1.5873,
      "step": 72500
    },
    {
      "epoch": 2.16,
      "learning_rate": 2.8361394356177377e-05,
      "loss": 1.6051,
      "step": 73000
    },
    {
      "epoch": 2.18,
      "learning_rate": 2.8213184728479962e-05,
      "loss": 1.6021,
      "step": 73500
    },
    {
      "epoch": 2.19,
      "learning_rate": 2.8064975100782546e-05,
      "loss": 1.6071,
      "step": 74000
    },
    {
      "epoch": 2.21,
      "learning_rate": 2.7916765473085134e-05,
      "loss": 1.6173,
      "step": 74500
    },
    {
      "epoch": 2.22,
      "learning_rate": 2.776855584538772e-05,
      "loss": 1.6023,
      "step": 75000
    },
    {
      "epoch": 2.24,
      "learning_rate": 2.7620346217690303e-05,
      "loss": 1.5913,
      "step": 75500
    },
    {
      "epoch": 2.25,
      "learning_rate": 2.7472136589992887e-05,
      "loss": 1.6075,
      "step": 76000
    },
    {
      "epoch": 2.27,
      "learning_rate": 2.7323926962295472e-05,
      "loss": 1.6212,
      "step": 76500
    },
    {
      "epoch": 2.28,
      "learning_rate": 2.7175717334598056e-05,
      "loss": 1.6085,
      "step": 77000
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.702750770690064e-05,
      "loss": 1.6209,
      "step": 77500
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.6879298079203225e-05,
      "loss": 1.6171,
      "step": 78000
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.673108845150581e-05,
      "loss": 1.6225,
      "step": 78500
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.6582878823808394e-05,
      "loss": 1.6236,
      "step": 79000
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.6434669196110978e-05,
      "loss": 1.6109,
      "step": 79500
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.6286459568413563e-05,
      "loss": 1.6405,
      "step": 80000
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.6138249940716147e-05,
      "loss": 1.6367,
      "step": 80500
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.5990040313018738e-05,
      "loss": 1.6158,
      "step": 81000
    },
    {
      "epoch": 2.42,
      "learning_rate": 2.5841830685321323e-05,
      "loss": 1.6349,
      "step": 81500
    },
    {
      "epoch": 2.43,
      "learning_rate": 2.5693621057623907e-05,
      "loss": 1.6462,
      "step": 82000
    },
    {
      "epoch": 2.45,
      "learning_rate": 2.554541142992649e-05,
      "loss": 1.6367,
      "step": 82500
    },
    {
      "epoch": 2.46,
      "learning_rate": 2.5397201802229076e-05,
      "loss": 1.6175,
      "step": 83000
    },
    {
      "epoch": 2.48,
      "learning_rate": 2.524899217453166e-05,
      "loss": 1.6294,
      "step": 83500
    },
    {
      "epoch": 2.49,
      "learning_rate": 2.5100782546834245e-05,
      "loss": 1.6447,
      "step": 84000
    },
    {
      "epoch": 2.5,
      "learning_rate": 2.495257291913683e-05,
      "loss": 1.6362,
      "step": 84500
    },
    {
      "epoch": 2.52,
      "learning_rate": 2.4804363291439414e-05,
      "loss": 1.6515,
      "step": 85000
    },
    {
      "epoch": 2.53,
      "learning_rate": 2.4656153663741998e-05,
      "loss": 1.6596,
      "step": 85500
    },
    {
      "epoch": 2.55,
      "learning_rate": 2.4507944036044582e-05,
      "loss": 1.6572,
      "step": 86000
    },
    {
      "epoch": 2.56,
      "learning_rate": 2.4359734408347167e-05,
      "loss": 1.6417,
      "step": 86500
    },
    {
      "epoch": 2.58,
      "learning_rate": 2.421152478064975e-05,
      "loss": 1.656,
      "step": 87000
    },
    {
      "epoch": 2.59,
      "learning_rate": 2.4063315152952336e-05,
      "loss": 1.6642,
      "step": 87500
    },
    {
      "epoch": 2.61,
      "learning_rate": 2.391510552525492e-05,
      "loss": 1.6481,
      "step": 88000
    },
    {
      "epoch": 2.62,
      "learning_rate": 2.3766895897557508e-05,
      "loss": 1.6528,
      "step": 88500
    },
    {
      "epoch": 2.64,
      "learning_rate": 2.3618686269860092e-05,
      "loss": 1.64,
      "step": 89000
    },
    {
      "epoch": 2.65,
      "learning_rate": 2.3470476642162677e-05,
      "loss": 1.6621,
      "step": 89500
    },
    {
      "epoch": 2.67,
      "learning_rate": 2.332226701446526e-05,
      "loss": 1.6731,
      "step": 90000
    },
    {
      "epoch": 2.68,
      "learning_rate": 2.3174057386767846e-05,
      "loss": 1.6672,
      "step": 90500
    },
    {
      "epoch": 2.7,
      "learning_rate": 2.302584775907043e-05,
      "loss": 1.6476,
      "step": 91000
    },
    {
      "epoch": 2.71,
      "learning_rate": 2.2877638131373014e-05,
      "loss": 1.6773,
      "step": 91500
    },
    {
      "epoch": 2.73,
      "learning_rate": 2.2729428503675602e-05,
      "loss": 1.6671,
      "step": 92000
    },
    {
      "epoch": 2.74,
      "learning_rate": 2.2581218875978187e-05,
      "loss": 1.6623,
      "step": 92500
    },
    {
      "epoch": 2.76,
      "learning_rate": 2.243300924828077e-05,
      "loss": 1.6602,
      "step": 93000
    },
    {
      "epoch": 2.77,
      "learning_rate": 2.2284799620583355e-05,
      "loss": 1.6765,
      "step": 93500
    },
    {
      "epoch": 2.79,
      "learning_rate": 2.213658999288594e-05,
      "loss": 1.6874,
      "step": 94000
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.1988380365188524e-05,
      "loss": 1.6777,
      "step": 94500
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.184017073749111e-05,
      "loss": 1.6582,
      "step": 95000
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.1691961109793693e-05,
      "loss": 1.6663,
      "step": 95500
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.1543751482096277e-05,
      "loss": 1.6773,
      "step": 96000
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.1395541854398862e-05,
      "loss": 1.6781,
      "step": 96500
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.1247332226701446e-05,
      "loss": 1.6983,
      "step": 97000
    },
    {
      "epoch": 2.89,
      "learning_rate": 2.109912259900403e-05,
      "loss": 1.6733,
      "step": 97500
    },
    {
      "epoch": 2.9,
      "learning_rate": 2.0950912971306615e-05,
      "loss": 1.6931,
      "step": 98000
    },
    {
      "epoch": 2.92,
      "learning_rate": 2.08027033436092e-05,
      "loss": 1.6948,
      "step": 98500
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.0654493715911787e-05,
      "loss": 1.6866,
      "step": 99000
    },
    {
      "epoch": 2.95,
      "learning_rate": 2.0506284088214372e-05,
      "loss": 1.7013,
      "step": 99500
    },
    {
      "epoch": 2.96,
      "learning_rate": 2.0358074460516956e-05,
      "loss": 1.6963,
      "step": 100000
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.020986483281954e-05,
      "loss": 1.6904,
      "step": 100500
    },
    {
      "epoch": 2.99,
      "learning_rate": 2.0061655205122125e-05,
      "loss": 1.7072,
      "step": 101000
    },
    {
      "epoch": 3.01,
      "learning_rate": 1.991344557742471e-05,
      "loss": 1.6192,
      "step": 101500
    },
    {
      "epoch": 3.02,
      "learning_rate": 1.9765235949727294e-05,
      "loss": 1.5539,
      "step": 102000
    },
    {
      "epoch": 3.04,
      "learning_rate": 1.961702632202988e-05,
      "loss": 1.5617,
      "step": 102500
    },
    {
      "epoch": 3.05,
      "learning_rate": 1.9468816694332466e-05,
      "loss": 1.5699,
      "step": 103000
    },
    {
      "epoch": 3.07,
      "learning_rate": 1.932060706663505e-05,
      "loss": 1.5772,
      "step": 103500
    },
    {
      "epoch": 3.08,
      "learning_rate": 1.9172397438937635e-05,
      "loss": 1.5697,
      "step": 104000
    },
    {
      "epoch": 3.1,
      "learning_rate": 1.902418781124022e-05,
      "loss": 1.5681,
      "step": 104500
    },
    {
      "epoch": 3.11,
      "learning_rate": 1.8875978183542804e-05,
      "loss": 1.5799,
      "step": 105000
    },
    {
      "epoch": 3.13,
      "learning_rate": 1.8727768555845388e-05,
      "loss": 1.5845,
      "step": 105500
    },
    {
      "epoch": 3.14,
      "learning_rate": 1.8579558928147972e-05,
      "loss": 1.5768,
      "step": 106000
    },
    {
      "epoch": 3.16,
      "learning_rate": 1.843134930045056e-05,
      "loss": 1.5776,
      "step": 106500
    },
    {
      "epoch": 3.17,
      "learning_rate": 1.8283139672753145e-05,
      "loss": 1.5728,
      "step": 107000
    },
    {
      "epoch": 3.19,
      "learning_rate": 1.813493004505573e-05,
      "loss": 1.5831,
      "step": 107500
    },
    {
      "epoch": 3.2,
      "learning_rate": 1.7986720417358314e-05,
      "loss": 1.5911,
      "step": 108000
    },
    {
      "epoch": 3.22,
      "learning_rate": 1.7838510789660898e-05,
      "loss": 1.5932,
      "step": 108500
    },
    {
      "epoch": 3.23,
      "learning_rate": 1.7690301161963482e-05,
      "loss": 1.5702,
      "step": 109000
    },
    {
      "epoch": 3.25,
      "learning_rate": 1.7542091534266067e-05,
      "loss": 1.5901,
      "step": 109500
    },
    {
      "epoch": 3.26,
      "learning_rate": 1.739388190656865e-05,
      "loss": 1.594,
      "step": 110000
    },
    {
      "epoch": 3.28,
      "learning_rate": 1.7245672278871236e-05,
      "loss": 1.5922,
      "step": 110500
    },
    {
      "epoch": 3.29,
      "learning_rate": 1.709746265117382e-05,
      "loss": 1.5669,
      "step": 111000
    },
    {
      "epoch": 3.31,
      "learning_rate": 1.6949253023476404e-05,
      "loss": 1.5945,
      "step": 111500
    },
    {
      "epoch": 3.32,
      "learning_rate": 1.680104339577899e-05,
      "loss": 1.5999,
      "step": 112000
    },
    {
      "epoch": 3.33,
      "learning_rate": 1.6652833768081573e-05,
      "loss": 1.5991,
      "step": 112500
    },
    {
      "epoch": 3.35,
      "learning_rate": 1.6504624140384158e-05,
      "loss": 1.5865,
      "step": 113000
    },
    {
      "epoch": 3.36,
      "learning_rate": 1.6356414512686745e-05,
      "loss": 1.598,
      "step": 113500
    },
    {
      "epoch": 3.38,
      "learning_rate": 1.620820488498933e-05,
      "loss": 1.5958,
      "step": 114000
    },
    {
      "epoch": 3.39,
      "learning_rate": 1.6059995257291914e-05,
      "loss": 1.5898,
      "step": 114500
    },
    {
      "epoch": 3.41,
      "learning_rate": 1.59117856295945e-05,
      "loss": 1.602,
      "step": 115000
    },
    {
      "epoch": 3.42,
      "learning_rate": 1.5763576001897083e-05,
      "loss": 1.5987,
      "step": 115500
    },
    {
      "epoch": 3.44,
      "learning_rate": 1.5615366374199667e-05,
      "loss": 1.5998,
      "step": 116000
    },
    {
      "epoch": 3.45,
      "learning_rate": 1.5467156746502255e-05,
      "loss": 1.6131,
      "step": 116500
    },
    {
      "epoch": 3.47,
      "learning_rate": 1.531894711880484e-05,
      "loss": 1.6055,
      "step": 117000
    },
    {
      "epoch": 3.48,
      "learning_rate": 1.5170737491107422e-05,
      "loss": 1.6107,
      "step": 117500
    },
    {
      "epoch": 3.5,
      "learning_rate": 1.5022527863410007e-05,
      "loss": 1.5989,
      "step": 118000
    },
    {
      "epoch": 3.51,
      "learning_rate": 1.4874318235712591e-05,
      "loss": 1.6009,
      "step": 118500
    },
    {
      "epoch": 3.53,
      "learning_rate": 1.4726108608015177e-05,
      "loss": 1.601,
      "step": 119000
    },
    {
      "epoch": 3.54,
      "learning_rate": 1.4577898980317762e-05,
      "loss": 1.6077,
      "step": 119500
    },
    {
      "epoch": 3.56,
      "learning_rate": 1.4429689352620346e-05,
      "loss": 1.5987,
      "step": 120000
    },
    {
      "epoch": 3.57,
      "learning_rate": 1.428147972492293e-05,
      "loss": 1.6148,
      "step": 120500
    },
    {
      "epoch": 3.59,
      "learning_rate": 1.4133270097225518e-05,
      "loss": 1.6126,
      "step": 121000
    },
    {
      "epoch": 3.6,
      "learning_rate": 1.3985060469528103e-05,
      "loss": 1.6129,
      "step": 121500
    },
    {
      "epoch": 3.62,
      "learning_rate": 1.3836850841830687e-05,
      "loss": 1.6093,
      "step": 122000
    },
    {
      "epoch": 3.63,
      "learning_rate": 1.3688641214133272e-05,
      "loss": 1.6098,
      "step": 122500
    },
    {
      "epoch": 3.65,
      "learning_rate": 1.3540431586435856e-05,
      "loss": 1.6178,
      "step": 123000
    },
    {
      "epoch": 3.66,
      "learning_rate": 1.339222195873844e-05,
      "loss": 1.6175,
      "step": 123500
    },
    {
      "epoch": 3.68,
      "learning_rate": 1.3244012331041025e-05,
      "loss": 1.6113,
      "step": 124000
    },
    {
      "epoch": 3.69,
      "learning_rate": 1.309580270334361e-05,
      "loss": 1.6271,
      "step": 124500
    },
    {
      "epoch": 3.71,
      "learning_rate": 1.2947593075646194e-05,
      "loss": 1.6193,
      "step": 125000
    },
    {
      "epoch": 3.72,
      "learning_rate": 1.279938344794878e-05,
      "loss": 1.621,
      "step": 125500
    },
    {
      "epoch": 3.73,
      "learning_rate": 1.2651173820251364e-05,
      "loss": 1.6174,
      "step": 126000
    },
    {
      "epoch": 3.75,
      "learning_rate": 1.2502964192553949e-05,
      "loss": 1.6252,
      "step": 126500
    },
    {
      "epoch": 3.76,
      "learning_rate": 1.2354754564856535e-05,
      "loss": 1.6591,
      "step": 127000
    },
    {
      "epoch": 3.78,
      "learning_rate": 1.220654493715912e-05,
      "loss": 1.6134,
      "step": 127500
    },
    {
      "epoch": 3.79,
      "learning_rate": 1.2058335309461704e-05,
      "loss": 1.6309,
      "step": 128000
    },
    {
      "epoch": 3.81,
      "learning_rate": 1.1910125681764288e-05,
      "loss": 1.6346,
      "step": 128500
    },
    {
      "epoch": 3.82,
      "learning_rate": 1.1761916054066872e-05,
      "loss": 1.6497,
      "step": 129000
    },
    {
      "epoch": 3.84,
      "learning_rate": 1.1613706426369457e-05,
      "loss": 1.6228,
      "step": 129500
    },
    {
      "epoch": 3.85,
      "learning_rate": 1.1465496798672043e-05,
      "loss": 1.6137,
      "step": 130000
    },
    {
      "epoch": 3.87,
      "learning_rate": 1.1317287170974627e-05,
      "loss": 1.6424,
      "step": 130500
    },
    {
      "epoch": 3.88,
      "learning_rate": 1.1169077543277212e-05,
      "loss": 1.6277,
      "step": 131000
    },
    {
      "epoch": 3.9,
      "learning_rate": 1.1020867915579796e-05,
      "loss": 1.6309,
      "step": 131500
    },
    {
      "epoch": 3.91,
      "learning_rate": 1.087265828788238e-05,
      "loss": 1.6329,
      "step": 132000
    },
    {
      "epoch": 3.93,
      "learning_rate": 1.0724448660184967e-05,
      "loss": 1.6469,
      "step": 132500
    },
    {
      "epoch": 3.94,
      "learning_rate": 1.0576239032487551e-05,
      "loss": 1.628,
      "step": 133000
    },
    {
      "epoch": 3.96,
      "learning_rate": 1.0428029404790136e-05,
      "loss": 1.6392,
      "step": 133500
    },
    {
      "epoch": 3.97,
      "learning_rate": 1.0279819777092722e-05,
      "loss": 1.6337,
      "step": 134000
    },
    {
      "epoch": 3.99,
      "learning_rate": 1.0131610149395306e-05,
      "loss": 1.6455,
      "step": 134500
    },
    {
      "epoch": 4.0,
      "learning_rate": 9.98340052169789e-06,
      "loss": 1.6251,
      "step": 135000
    },
    {
      "epoch": 4.02,
      "learning_rate": 9.835190894000475e-06,
      "loss": 1.5574,
      "step": 135500
    },
    {
      "epoch": 4.03,
      "learning_rate": 9.68698126630306e-06,
      "loss": 1.5637,
      "step": 136000
    },
    {
      "epoch": 4.05,
      "learning_rate": 9.538771638605644e-06,
      "loss": 1.5522,
      "step": 136500
    },
    {
      "epoch": 4.06,
      "learning_rate": 9.390562010908228e-06,
      "loss": 1.5643,
      "step": 137000
    },
    {
      "epoch": 4.08,
      "learning_rate": 9.242352383210814e-06,
      "loss": 1.5713,
      "step": 137500
    },
    {
      "epoch": 4.09,
      "learning_rate": 9.094142755513399e-06,
      "loss": 1.5562,
      "step": 138000
    },
    {
      "epoch": 4.11,
      "learning_rate": 8.945933127815983e-06,
      "loss": 1.561,
      "step": 138500
    },
    {
      "epoch": 4.12,
      "learning_rate": 8.797723500118567e-06,
      "loss": 1.57,
      "step": 139000
    },
    {
      "epoch": 4.14,
      "learning_rate": 8.649513872421154e-06,
      "loss": 1.5747,
      "step": 139500
    },
    {
      "epoch": 4.15,
      "learning_rate": 8.501304244723738e-06,
      "loss": 1.5745,
      "step": 140000
    },
    {
      "epoch": 4.16,
      "learning_rate": 8.353094617026322e-06,
      "loss": 1.5642,
      "step": 140500
    },
    {
      "epoch": 4.18,
      "learning_rate": 8.204884989328908e-06,
      "loss": 1.5595,
      "step": 141000
    },
    {
      "epoch": 4.19,
      "learning_rate": 8.056675361631493e-06,
      "loss": 1.563,
      "step": 141500
    },
    {
      "epoch": 4.21,
      "learning_rate": 7.908465733934077e-06,
      "loss": 1.573,
      "step": 142000
    },
    {
      "epoch": 4.22,
      "learning_rate": 7.760256106236662e-06,
      "loss": 1.5599,
      "step": 142500
    },
    {
      "epoch": 4.24,
      "learning_rate": 7.612046478539246e-06,
      "loss": 1.5728,
      "step": 143000
    },
    {
      "epoch": 4.25,
      "learning_rate": 7.4638368508418305e-06,
      "loss": 1.5663,
      "step": 143500
    },
    {
      "epoch": 4.27,
      "learning_rate": 7.315627223144415e-06,
      "loss": 1.5639,
      "step": 144000
    },
    {
      "epoch": 4.28,
      "learning_rate": 7.167417595447e-06,
      "loss": 1.5638,
      "step": 144500
    },
    {
      "epoch": 4.3,
      "learning_rate": 7.0192079677495855e-06,
      "loss": 1.5848,
      "step": 145000
    },
    {
      "epoch": 4.31,
      "learning_rate": 6.870998340052171e-06,
      "loss": 1.5749,
      "step": 145500
    },
    {
      "epoch": 4.33,
      "learning_rate": 6.722788712354755e-06,
      "loss": 1.5706,
      "step": 146000
    },
    {
      "epoch": 4.34,
      "learning_rate": 6.5745790846573396e-06,
      "loss": 1.5832,
      "step": 146500
    },
    {
      "epoch": 4.36,
      "learning_rate": 6.426369456959924e-06,
      "loss": 1.5897,
      "step": 147000
    },
    {
      "epoch": 4.37,
      "learning_rate": 6.278159829262508e-06,
      "loss": 1.5883,
      "step": 147500
    },
    {
      "epoch": 4.39,
      "learning_rate": 6.129950201565094e-06,
      "loss": 1.6014,
      "step": 148000
    },
    {
      "epoch": 4.4,
      "learning_rate": 5.981740573867679e-06,
      "loss": 1.5756,
      "step": 148500
    },
    {
      "epoch": 4.42,
      "learning_rate": 5.833530946170263e-06,
      "loss": 1.5797,
      "step": 149000
    },
    {
      "epoch": 4.43,
      "learning_rate": 5.6853213184728486e-06,
      "loss": 1.577,
      "step": 149500
    },
    {
      "epoch": 4.45,
      "learning_rate": 5.537111690775433e-06,
      "loss": 1.583,
      "step": 150000
    },
    {
      "epoch": 4.46,
      "learning_rate": 5.388902063078017e-06,
      "loss": 1.5979,
      "step": 150500
    },
    {
      "epoch": 4.48,
      "learning_rate": 5.240692435380603e-06,
      "loss": 1.5847,
      "step": 151000
    },
    {
      "epoch": 4.49,
      "learning_rate": 5.092482807683188e-06,
      "loss": 1.5839,
      "step": 151500
    },
    {
      "epoch": 4.51,
      "learning_rate": 4.944273179985772e-06,
      "loss": 1.5707,
      "step": 152000
    },
    {
      "epoch": 4.52,
      "learning_rate": 4.796063552288357e-06,
      "loss": 1.5825,
      "step": 152500
    },
    {
      "epoch": 4.54,
      "learning_rate": 4.647853924590942e-06,
      "loss": 1.5799,
      "step": 153000
    },
    {
      "epoch": 4.55,
      "learning_rate": 4.499644296893526e-06,
      "loss": 1.5804,
      "step": 153500
    },
    {
      "epoch": 4.56,
      "learning_rate": 4.351434669196111e-06,
      "loss": 1.5841,
      "step": 154000
    },
    {
      "epoch": 4.58,
      "learning_rate": 4.203225041498696e-06,
      "loss": 1.582,
      "step": 154500
    },
    {
      "epoch": 4.59,
      "learning_rate": 4.055015413801281e-06,
      "loss": 1.5959,
      "step": 155000
    },
    {
      "epoch": 4.61,
      "learning_rate": 3.906805786103866e-06,
      "loss": 1.5959,
      "step": 155500
    },
    {
      "epoch": 4.62,
      "learning_rate": 3.75859615840645e-06,
      "loss": 1.6,
      "step": 156000
    },
    {
      "epoch": 4.64,
      "learning_rate": 3.6103865307090354e-06,
      "loss": 1.5994,
      "step": 156500
    },
    {
      "epoch": 4.65,
      "learning_rate": 3.46217690301162e-06,
      "loss": 1.5941,
      "step": 157000
    },
    {
      "epoch": 4.67,
      "learning_rate": 3.3139672753142047e-06,
      "loss": 1.582,
      "step": 157500
    },
    {
      "epoch": 4.68,
      "learning_rate": 3.165757647616789e-06,
      "loss": 1.5842,
      "step": 158000
    },
    {
      "epoch": 4.7,
      "learning_rate": 3.017548019919374e-06,
      "loss": 1.5866,
      "step": 158500
    },
    {
      "epoch": 4.71,
      "learning_rate": 2.8693383922219588e-06,
      "loss": 1.5969,
      "step": 159000
    },
    {
      "epoch": 4.73,
      "learning_rate": 2.7211287645245436e-06,
      "loss": 1.6188,
      "step": 159500
    },
    {
      "epoch": 4.74,
      "learning_rate": 2.5729191368271284e-06,
      "loss": 1.5876,
      "step": 160000
    },
    {
      "epoch": 4.76,
      "learning_rate": 2.424709509129713e-06,
      "loss": 1.6092,
      "step": 160500
    },
    {
      "epoch": 4.77,
      "learning_rate": 2.276499881432298e-06,
      "loss": 1.5818,
      "step": 161000
    },
    {
      "epoch": 4.79,
      "learning_rate": 2.1282902537348825e-06,
      "loss": 1.586,
      "step": 161500
    },
    {
      "epoch": 4.8,
      "learning_rate": 1.9800806260374674e-06,
      "loss": 1.5991,
      "step": 162000
    },
    {
      "epoch": 4.82,
      "learning_rate": 1.8318709983400524e-06,
      "loss": 1.5824,
      "step": 162500
    },
    {
      "epoch": 4.83,
      "learning_rate": 1.683661370642637e-06,
      "loss": 1.5907,
      "step": 163000
    },
    {
      "epoch": 4.85,
      "learning_rate": 1.5354517429452217e-06,
      "loss": 1.607,
      "step": 163500
    },
    {
      "epoch": 4.86,
      "learning_rate": 1.3872421152478065e-06,
      "loss": 1.591,
      "step": 164000
    },
    {
      "epoch": 4.88,
      "learning_rate": 1.2390324875503913e-06,
      "loss": 1.6106,
      "step": 164500
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.0908228598529762e-06,
      "loss": 1.6029,
      "step": 165000
    },
    {
      "epoch": 4.91,
      "learning_rate": 9.426132321555609e-07,
      "loss": 1.6054,
      "step": 165500
    },
    {
      "epoch": 4.92,
      "learning_rate": 7.944036044581456e-07,
      "loss": 1.6038,
      "step": 166000
    },
    {
      "epoch": 4.94,
      "learning_rate": 6.461939767607305e-07,
      "loss": 1.6129,
      "step": 166500
    },
    {
      "epoch": 4.95,
      "learning_rate": 4.979843490633152e-07,
      "loss": 1.603,
      "step": 167000
    },
    {
      "epoch": 4.97,
      "learning_rate": 3.4977472136589993e-07,
      "loss": 1.5922,
      "step": 167500
    },
    {
      "epoch": 4.98,
      "learning_rate": 2.0156509366848474e-07,
      "loss": 1.6177,
      "step": 168000
    },
    {
      "epoch": 4.98,
      "step": 168000,
      "total_flos": 6.240794371172598e+17,
      "train_loss": 0.0,
      "train_runtime": 0.0953,
      "train_samples_per_second": 4248275.036,
      "train_steps_per_second": 1062084.5
    }
  ],
  "max_steps": 101208,
  "num_train_epochs": 3,
  "total_flos": 6.240794371172598e+17,
  "trial_name": null,
  "trial_params": null
}