{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 6.079355858472596,
  "global_step": 900000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 2.2222222222222223e-05,
      "loss": 10.6946,
      "step": 2000
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.4444444444444447e-05,
      "loss": 10.5675,
      "step": 4000
    },
    {
      "epoch": 0.04,
      "learning_rate": 6.666666666666667e-05,
      "loss": 10.4211,
      "step": 6000
    },
    {
      "epoch": 0.05,
      "learning_rate": 8.888888888888889e-05,
      "loss": 10.2041,
      "step": 8000
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.988776655443322e-05,
      "loss": 9.8916,
      "step": 10000
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.966329966329967e-05,
      "loss": 9.4948,
      "step": 12000
    },
    {
      "epoch": 0.09,
      "learning_rate": 9.943883277216612e-05,
      "loss": 9.0658,
      "step": 14000
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.921436588103256e-05,
      "loss": 8.6284,
      "step": 16000
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.8989898989899e-05,
      "loss": 8.1946,
      "step": 18000
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.876543209876543e-05,
      "loss": 7.801,
      "step": 20000
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.854096520763188e-05,
      "loss": 7.4541,
      "step": 22000
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.831649831649832e-05,
      "loss": 7.283,
      "step": 24000
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.809203142536476e-05,
      "loss": 7.2067,
      "step": 26000
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.78675645342312e-05,
      "loss": 7.1146,
      "step": 28000
    },
    {
      "epoch": 0.2,
      "learning_rate": 9.764309764309764e-05,
      "loss": 7.0541,
      "step": 30000
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.74186307519641e-05,
      "loss": 6.9996,
      "step": 32000
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.719416386083054e-05,
      "loss": 6.967,
      "step": 34000
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.696969696969698e-05,
      "loss": 6.9272,
      "step": 36000
    },
    {
      "epoch": 0.26,
      "learning_rate": 9.674523007856342e-05,
      "loss": 6.9136,
      "step": 38000
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.652076318742987e-05,
      "loss": 6.8708,
      "step": 40000
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.62962962962963e-05,
      "loss": 6.8544,
      "step": 42000
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.607182940516274e-05,
      "loss": 6.8332,
      "step": 44000
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.584736251402918e-05,
      "loss": 6.8299,
      "step": 46000
    },
    {
      "epoch": 0.32,
      "learning_rate": 9.562289562289563e-05,
      "loss": 6.806,
      "step": 48000
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.539842873176207e-05,
      "loss": 6.7812,
      "step": 50000
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.517396184062851e-05,
      "loss": 6.7821,
      "step": 52000
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.494949494949495e-05,
      "loss": 6.756,
      "step": 54000
    },
    {
      "epoch": 0.38,
      "learning_rate": 9.472502805836139e-05,
      "loss": 6.7547,
      "step": 56000
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.450056116722785e-05,
      "loss": 6.7199,
      "step": 58000
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.427609427609429e-05,
      "loss": 6.6967,
      "step": 60000
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.405162738496073e-05,
      "loss": 6.6726,
      "step": 62000
    },
    {
      "epoch": 0.43,
      "learning_rate": 9.382716049382717e-05,
      "loss": 6.6519,
      "step": 64000
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.36026936026936e-05,
      "loss": 6.6244,
      "step": 66000
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.337822671156005e-05,
      "loss": 6.5836,
      "step": 68000
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.315375982042649e-05,
      "loss": 6.5503,
      "step": 70000
    },
    {
      "epoch": 0.49,
      "learning_rate": 9.292929292929293e-05,
      "loss": 6.5117,
      "step": 72000
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.270482603815938e-05,
      "loss": 6.4852,
      "step": 74000
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.248035914702582e-05,
      "loss": 6.4517,
      "step": 76000
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.225589225589226e-05,
      "loss": 6.405,
      "step": 78000
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.20314253647587e-05,
      "loss": 6.3787,
      "step": 80000
    },
    {
      "epoch": 0.55,
      "learning_rate": 9.180695847362514e-05,
      "loss": 6.3171,
      "step": 82000
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.15824915824916e-05,
      "loss": 6.2806,
      "step": 84000
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.135802469135802e-05,
      "loss": 6.3758,
      "step": 86000
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.113355780022446e-05,
      "loss": 6.3753,
      "step": 88000
    },
    {
      "epoch": 0.61,
      "learning_rate": 9.090909090909092e-05,
      "loss": 6.2428,
      "step": 90000
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.068462401795736e-05,
      "loss": 5.9605,
      "step": 92000
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.04601571268238e-05,
      "loss": 5.774,
      "step": 94000
    },
    {
      "epoch": 0.65,
      "learning_rate": 9.023569023569024e-05,
      "loss": 5.8005,
      "step": 96000
    },
    {
      "epoch": 0.66,
      "learning_rate": 9.001122334455668e-05,
      "loss": 5.8735,
      "step": 98000
    },
    {
      "epoch": 0.68,
      "learning_rate": 8.978675645342313e-05,
      "loss": 5.8452,
      "step": 100000
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.956228956228957e-05,
      "loss": 5.7882,
      "step": 102000
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.933782267115601e-05,
      "loss": 6.0639,
      "step": 104000
    },
    {
      "epoch": 0.72,
      "learning_rate": 8.911335578002245e-05,
      "loss": 6.0042,
      "step": 106000
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.888888888888889e-05,
      "loss": 5.9511,
      "step": 108000
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.866442199775533e-05,
      "loss": 5.8909,
      "step": 110000
    },
    {
      "epoch": 0.76,
      "learning_rate": 8.843995510662177e-05,
      "loss": 5.8925,
      "step": 112000
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.821548821548821e-05,
      "loss": 5.9042,
      "step": 114000
    },
    {
      "epoch": 0.78,
      "learning_rate": 8.799102132435467e-05,
      "loss": 5.9797,
      "step": 116000
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.776655443322111e-05,
      "loss": 6.1174,
      "step": 118000
    },
    {
      "epoch": 0.81,
      "learning_rate": 8.754208754208755e-05,
      "loss": 6.2284,
      "step": 120000
    },
    {
      "epoch": 0.82,
      "learning_rate": 8.731762065095399e-05,
      "loss": 6.1204,
      "step": 122000
    },
    {
      "epoch": 0.84,
      "learning_rate": 8.709315375982043e-05,
      "loss": 5.6589,
      "step": 124000
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.686868686868688e-05,
      "loss": 5.7295,
      "step": 126000
    },
    {
      "epoch": 0.86,
      "learning_rate": 8.664421997755332e-05,
      "loss": 5.7535,
      "step": 128000
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.641975308641975e-05,
      "loss": 5.6045,
      "step": 130000
    },
    {
      "epoch": 0.89,
      "learning_rate": 8.619528619528619e-05,
      "loss": 5.6091,
      "step": 132000
    },
    {
      "epoch": 0.91,
      "learning_rate": 8.597081930415264e-05,
      "loss": 5.5933,
      "step": 134000
    },
    {
      "epoch": 0.92,
      "learning_rate": 8.574635241301908e-05,
      "loss": 5.6344,
      "step": 136000
    },
    {
      "epoch": 0.93,
      "learning_rate": 8.552188552188552e-05,
      "loss": 5.6063,
      "step": 138000
    },
    {
      "epoch": 0.95,
      "learning_rate": 8.529741863075196e-05,
      "loss": 5.4933,
      "step": 140000
    },
    {
      "epoch": 0.96,
      "learning_rate": 8.507295173961842e-05,
      "loss": 5.5386,
      "step": 142000
    },
    {
      "epoch": 0.97,
      "learning_rate": 8.484848484848486e-05,
      "loss": 5.5085,
      "step": 144000
    },
    {
      "epoch": 0.99,
      "learning_rate": 8.46240179573513e-05,
      "loss": 5.4098,
      "step": 146000
    },
    {
      "epoch": 1.0,
      "learning_rate": 8.439955106621774e-05,
      "loss": 5.0913,
      "step": 148000
    },
    {
      "epoch": 1.01,
      "learning_rate": 8.417508417508418e-05,
      "loss": 5.6115,
      "step": 150000
    },
    {
      "epoch": 1.03,
      "learning_rate": 8.395061728395062e-05,
      "loss": 5.5311,
      "step": 152000
    },
    {
      "epoch": 1.04,
      "learning_rate": 8.372615039281706e-05,
      "loss": 5.4583,
      "step": 154000
    },
    {
      "epoch": 1.05,
      "learning_rate": 8.35016835016835e-05,
      "loss": 5.4188,
      "step": 156000
    },
    {
      "epoch": 1.07,
      "learning_rate": 8.327721661054994e-05,
      "loss": 5.3814,
      "step": 158000
    },
    {
      "epoch": 1.08,
      "learning_rate": 8.30527497194164e-05,
      "loss": 5.3292,
      "step": 160000
    },
    {
      "epoch": 1.09,
      "learning_rate": 8.282828282828283e-05,
      "loss": 5.291,
      "step": 162000
    },
    {
      "epoch": 1.11,
      "learning_rate": 8.260381593714927e-05,
      "loss": 5.2601,
      "step": 164000
    },
    {
      "epoch": 1.12,
      "learning_rate": 8.237934904601572e-05,
      "loss": 5.2253,
      "step": 166000
    },
    {
      "epoch": 1.13,
      "learning_rate": 8.215488215488217e-05,
      "loss": 5.2076,
      "step": 168000
    },
    {
      "epoch": 1.15,
      "learning_rate": 8.193041526374861e-05,
      "loss": 5.1529,
      "step": 170000
    },
    {
      "epoch": 1.16,
      "learning_rate": 8.170594837261504e-05,
      "loss": 5.1493,
      "step": 172000
    },
    {
      "epoch": 1.18,
      "learning_rate": 8.148148148148148e-05,
      "loss": 5.1449,
      "step": 174000
    },
    {
      "epoch": 1.19,
      "learning_rate": 8.125701459034793e-05,
      "loss": 5.0768,
      "step": 176000
    },
    {
      "epoch": 1.2,
      "learning_rate": 8.103254769921437e-05,
      "loss": 5.0558,
      "step": 178000
    },
    {
      "epoch": 1.22,
      "learning_rate": 8.080808080808081e-05,
      "loss": 5.0232,
      "step": 180000
    },
    {
      "epoch": 1.23,
      "learning_rate": 8.058361391694725e-05,
      "loss": 5.0037,
      "step": 182000
    },
    {
      "epoch": 1.24,
      "learning_rate": 8.035914702581369e-05,
      "loss": 4.9835,
      "step": 184000
    },
    {
      "epoch": 1.26,
      "learning_rate": 8.013468013468015e-05,
      "loss": 4.9832,
      "step": 186000
    },
    {
      "epoch": 1.27,
      "learning_rate": 7.991021324354659e-05,
      "loss": 4.9235,
      "step": 188000
    },
    {
      "epoch": 1.28,
      "learning_rate": 7.968574635241303e-05,
      "loss": 4.9121,
      "step": 190000
    },
    {
      "epoch": 1.3,
      "learning_rate": 7.946127946127947e-05,
      "loss": 4.8973,
      "step": 192000
    },
    {
      "epoch": 1.31,
      "learning_rate": 7.92368125701459e-05,
      "loss": 4.8992,
      "step": 194000
    },
    {
      "epoch": 1.32,
      "learning_rate": 7.901234567901235e-05,
      "loss": 4.8595,
      "step": 196000
    },
    {
      "epoch": 1.34,
      "learning_rate": 7.878787878787879e-05,
      "loss": 4.84,
      "step": 198000
    },
    {
      "epoch": 1.35,
      "learning_rate": 7.856341189674523e-05,
      "loss": 4.8424,
      "step": 200000
    },
    {
      "epoch": 1.36,
      "learning_rate": 7.833894500561168e-05,
      "loss": 4.8017,
      "step": 202000
    },
    {
      "epoch": 1.38,
      "learning_rate": 7.811447811447812e-05,
      "loss": 4.8083,
      "step": 204000
    },
    {
      "epoch": 1.39,
      "learning_rate": 7.789001122334456e-05,
      "loss": 4.7732,
      "step": 206000
    },
    {
      "epoch": 1.41,
      "learning_rate": 7.7665544332211e-05,
      "loss": 4.7528,
      "step": 208000
    },
    {
      "epoch": 1.42,
      "learning_rate": 7.744107744107744e-05,
      "loss": 4.7499,
      "step": 210000
    },
    {
      "epoch": 1.43,
      "learning_rate": 7.72166105499439e-05,
      "loss": 4.7275,
      "step": 212000
    },
    {
      "epoch": 1.45,
      "learning_rate": 7.699214365881034e-05,
      "loss": 4.7118,
      "step": 214000
    },
    {
      "epoch": 1.46,
      "learning_rate": 7.676767676767676e-05,
      "loss": 4.6839,
      "step": 216000
    },
    {
      "epoch": 1.47,
      "learning_rate": 7.65432098765432e-05,
      "loss": 4.6654,
      "step": 218000
    },
    {
      "epoch": 1.49,
      "learning_rate": 7.631874298540966e-05,
      "loss": 4.6509,
      "step": 220000
    },
    {
      "epoch": 1.5,
      "learning_rate": 7.60942760942761e-05,
      "loss": 4.6436,
      "step": 222000
    },
    {
      "epoch": 1.51,
      "learning_rate": 7.586980920314254e-05,
      "loss": 4.6321,
      "step": 224000
    },
    {
      "epoch": 1.53,
      "learning_rate": 7.564534231200898e-05,
      "loss": 4.6257,
      "step": 226000
    },
    {
      "epoch": 1.54,
      "learning_rate": 7.542087542087543e-05,
      "loss": 4.6259,
      "step": 228000
    },
    {
      "epoch": 1.55,
      "learning_rate": 7.519640852974187e-05,
      "loss": 4.5876,
      "step": 230000
    },
    {
      "epoch": 1.57,
      "learning_rate": 7.497194163860831e-05,
      "loss": 4.5918,
      "step": 232000
    },
    {
      "epoch": 1.58,
      "learning_rate": 7.474747474747475e-05,
      "loss": 4.7349,
      "step": 234000
    },
    {
      "epoch": 1.59,
      "learning_rate": 7.452300785634119e-05,
      "loss": 4.7906,
      "step": 236000
    },
    {
      "epoch": 1.61,
      "learning_rate": 7.429854096520763e-05,
      "loss": 4.5217,
      "step": 238000
    },
    {
      "epoch": 1.62,
      "learning_rate": 7.407407407407407e-05,
      "loss": 4.2357,
      "step": 240000
    },
    {
      "epoch": 1.63,
      "learning_rate": 7.384960718294051e-05,
      "loss": 4.1207,
      "step": 242000
    },
    {
      "epoch": 1.65,
      "learning_rate": 7.362514029180695e-05,
      "loss": 4.1877,
      "step": 244000
    },
    {
      "epoch": 1.66,
      "learning_rate": 7.340067340067341e-05,
      "loss": 4.4264,
      "step": 246000
    },
    {
      "epoch": 1.68,
      "learning_rate": 7.317620650953985e-05,
      "loss": 4.3763,
      "step": 248000
    },
    {
      "epoch": 1.69,
      "learning_rate": 7.295173961840629e-05,
      "loss": 4.3764,
      "step": 250000
    },
    {
      "epoch": 1.7,
      "learning_rate": 7.272727272727273e-05,
      "loss": 4.6481,
      "step": 252000
    },
    {
      "epoch": 1.72,
      "learning_rate": 7.250280583613918e-05,
      "loss": 4.6296,
      "step": 254000
    },
    {
      "epoch": 1.73,
      "learning_rate": 7.227833894500562e-05,
      "loss": 4.6029,
      "step": 256000
    },
    {
      "epoch": 1.74,
      "learning_rate": 7.205387205387206e-05,
      "loss": 4.5606,
      "step": 258000
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.182940516273849e-05,
      "loss": 4.6015,
      "step": 260000
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.160493827160494e-05,
      "loss": 4.6519,
      "step": 262000
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.138047138047138e-05,
      "loss": 4.7609,
      "step": 264000
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.115600448933782e-05,
      "loss": 4.9615,
      "step": 266000
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.093153759820426e-05,
      "loss": 5.1392,
      "step": 268000
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.07070707070707e-05,
      "loss": 5.0403,
      "step": 270000
    },
    {
      "epoch": 1.84,
      "learning_rate": 7.048260381593716e-05,
      "loss": 4.3987,
      "step": 272000
    },
    {
      "epoch": 1.85,
      "learning_rate": 7.02581369248036e-05,
      "loss": 4.6271,
      "step": 274000
    },
    {
      "epoch": 1.86,
      "learning_rate": 7.003367003367004e-05,
      "loss": 4.6795,
      "step": 276000
    },
    {
      "epoch": 1.88,
      "learning_rate": 6.980920314253648e-05,
      "loss": 4.5634,
      "step": 278000
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.958473625140293e-05,
      "loss": 4.5767,
      "step": 280000
    },
    {
      "epoch": 1.9,
      "learning_rate": 6.936026936026936e-05,
      "loss": 4.5943,
      "step": 282000
    },
    {
      "epoch": 1.92,
      "learning_rate": 6.91358024691358e-05,
      "loss": 4.6575,
      "step": 284000
    },
    {
      "epoch": 1.93,
      "learning_rate": 6.891133557800224e-05,
      "loss": 4.6512,
      "step": 286000
    },
    {
      "epoch": 1.95,
      "learning_rate": 6.86868686868687e-05,
      "loss": 4.5433,
      "step": 288000
    },
    {
      "epoch": 1.96,
      "learning_rate": 6.846240179573513e-05,
      "loss": 4.6108,
      "step": 290000
    },
    {
      "epoch": 1.97,
      "learning_rate": 6.823793490460157e-05,
      "loss": 4.6175,
      "step": 292000
    },
    {
      "epoch": 1.99,
      "learning_rate": 6.801346801346801e-05,
      "loss": 4.5076,
      "step": 294000
    },
    {
      "epoch": 2.0,
      "learning_rate": 6.778900112233445e-05,
      "loss": 4.131,
      "step": 296000
    },
    {
      "epoch": 2.01,
      "learning_rate": 6.756453423120091e-05,
      "loss": 4.4705,
      "step": 298000
    },
    {
      "epoch": 2.03,
      "learning_rate": 6.734006734006735e-05,
      "loss": 4.4578,
      "step": 300000
    },
    {
      "epoch": 2.04,
      "learning_rate": 6.711560044893379e-05,
      "loss": 4.4142,
      "step": 302000
    },
    {
      "epoch": 2.05,
      "learning_rate": 6.689113355780023e-05,
      "loss": 4.3968,
      "step": 304000
    },
    {
      "epoch": 2.07,
      "learning_rate": 6.666666666666667e-05,
      "loss": 4.3893,
      "step": 306000
    },
    {
      "epoch": 2.08,
      "learning_rate": 6.644219977553311e-05,
      "loss": 4.3628,
      "step": 308000
    },
    {
      "epoch": 2.09,
      "learning_rate": 6.621773288439955e-05,
      "loss": 4.3452,
      "step": 310000
    },
    {
      "epoch": 2.11,
      "learning_rate": 6.599326599326599e-05,
      "loss": 4.3391,
      "step": 312000
    },
    {
      "epoch": 2.12,
      "learning_rate": 6.576879910213244e-05,
      "loss": 4.3232,
      "step": 314000
    },
    {
      "epoch": 2.13,
      "learning_rate": 6.554433221099888e-05,
      "loss": 4.3286,
      "step": 316000
    },
    {
      "epoch": 2.15,
      "learning_rate": 6.531986531986533e-05,
      "loss": 4.2963,
      "step": 318000
    },
    {
      "epoch": 2.16,
      "learning_rate": 6.509539842873177e-05,
      "loss": 4.3098,
      "step": 320000
    },
    {
      "epoch": 2.18,
      "learning_rate": 6.48709315375982e-05,
      "loss": 4.3239,
      "step": 322000
    },
    {
      "epoch": 2.19,
      "learning_rate": 6.464646464646466e-05,
      "loss": 4.2606,
      "step": 324000
    },
    {
      "epoch": 2.2,
      "learning_rate": 6.442199775533109e-05,
      "loss": 4.2617,
      "step": 326000
    },
    {
      "epoch": 2.22,
      "learning_rate": 6.419753086419753e-05,
      "loss": 4.2429,
      "step": 328000
    },
    {
      "epoch": 2.23,
      "learning_rate": 6.397306397306398e-05,
      "loss": 4.2412,
      "step": 330000
    },
    {
      "epoch": 2.24,
      "learning_rate": 6.374859708193042e-05,
      "loss": 4.2415,
      "step": 332000
    },
    {
      "epoch": 2.26,
      "learning_rate": 6.352413019079686e-05,
      "loss": 4.2517,
      "step": 334000
    },
    {
      "epoch": 2.27,
      "learning_rate": 6.32996632996633e-05,
      "loss": 4.2051,
      "step": 336000
    },
    {
      "epoch": 2.28,
      "learning_rate": 6.307519640852974e-05,
      "loss": 4.2003,
      "step": 338000
    },
    {
      "epoch": 2.3,
      "learning_rate": 6.28507295173962e-05,
      "loss": 4.2039,
      "step": 340000
    },
    {
      "epoch": 2.31,
      "learning_rate": 6.262626262626264e-05,
      "loss": 4.2197,
      "step": 342000
    },
    {
      "epoch": 2.32,
      "learning_rate": 6.240179573512908e-05,
      "loss": 4.1898,
      "step": 344000
    },
    {
      "epoch": 2.34,
      "learning_rate": 6.21773288439955e-05,
      "loss": 4.1859,
      "step": 346000
    },
    {
      "epoch": 2.35,
      "learning_rate": 6.195286195286196e-05,
      "loss": 4.2003,
      "step": 348000
    },
    {
      "epoch": 2.36,
      "learning_rate": 6.17283950617284e-05,
      "loss": 4.1667,
      "step": 350000
    },
    {
      "epoch": 2.38,
      "learning_rate": 6.150392817059484e-05,
      "loss": 4.1896,
      "step": 352000
    },
    {
      "epoch": 2.39,
      "learning_rate": 6.127946127946128e-05,
      "loss": 4.1576,
      "step": 354000
    },
    {
      "epoch": 2.4,
      "learning_rate": 6.105499438832773e-05,
      "loss": 4.1515,
      "step": 356000
    },
    {
      "epoch": 2.42,
      "learning_rate": 6.0830527497194164e-05,
      "loss": 4.1666,
      "step": 358000
    },
    {
      "epoch": 2.43,
      "learning_rate": 6.060606060606061e-05,
      "loss": 4.1459,
      "step": 360000
    },
    {
      "epoch": 2.45,
      "learning_rate": 6.038159371492705e-05,
      "loss": 4.1438,
      "step": 362000
    },
    {
      "epoch": 2.46,
      "learning_rate": 6.01571268237935e-05,
      "loss": 4.1272,
      "step": 364000
    },
    {
      "epoch": 2.47,
      "learning_rate": 5.993265993265994e-05,
      "loss": 4.1105,
      "step": 366000
    },
    {
      "epoch": 2.49,
      "learning_rate": 5.970819304152637e-05,
      "loss": 4.1111,
      "step": 368000
    },
    {
      "epoch": 2.5,
      "learning_rate": 5.948372615039281e-05,
      "loss": 4.1062,
      "step": 370000
    },
    {
      "epoch": 2.51,
      "learning_rate": 5.925925925925926e-05,
      "loss": 4.1072,
      "step": 372000
    },
    {
      "epoch": 2.53,
      "learning_rate": 5.90347923681257e-05,
      "loss": 4.1112,
      "step": 374000
    },
    {
      "epoch": 2.54,
      "learning_rate": 5.881032547699215e-05,
      "loss": 4.1251,
      "step": 376000
    },
    {
      "epoch": 2.55,
      "learning_rate": 5.858585858585859e-05,
      "loss": 4.0813,
      "step": 378000
    },
    {
      "epoch": 2.57,
      "learning_rate": 5.8361391694725034e-05,
      "loss": 4.0908,
      "step": 380000
    },
    {
      "epoch": 2.58,
      "learning_rate": 5.8136924803591475e-05,
      "loss": 4.2286,
      "step": 382000
    },
    {
      "epoch": 2.59,
      "learning_rate": 5.7912457912457915e-05,
      "loss": 4.3118,
      "step": 384000
    },
    {
      "epoch": 2.61,
      "learning_rate": 5.768799102132436e-05,
      "loss": 4.0238,
      "step": 386000
    },
    {
      "epoch": 2.62,
      "learning_rate": 5.74635241301908e-05,
      "loss": 3.7601,
      "step": 388000
    },
    {
      "epoch": 2.63,
      "learning_rate": 5.7239057239057236e-05,
      "loss": 3.6653,
      "step": 390000
    },
    {
      "epoch": 2.65,
      "learning_rate": 5.701459034792368e-05,
      "loss": 3.7115,
      "step": 392000
    },
    {
      "epoch": 2.66,
      "learning_rate": 5.679012345679012e-05,
      "loss": 4.0332,
      "step": 394000
    },
    {
      "epoch": 2.67,
      "learning_rate": 5.6565656565656563e-05,
      "loss": 3.9273,
      "step": 396000
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.634118967452301e-05,
      "loss": 3.9359,
      "step": 398000
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.611672278338945e-05,
      "loss": 4.2338,
      "step": 400000
    },
    {
      "epoch": 2.72,
      "learning_rate": 5.58922558922559e-05,
      "loss": 4.2178,
      "step": 402000
    },
    {
      "epoch": 2.73,
      "learning_rate": 5.566778900112234e-05,
      "loss": 4.2029,
      "step": 404000
    },
    {
      "epoch": 2.74,
      "learning_rate": 5.5443322109988785e-05,
      "loss": 4.1545,
      "step": 406000
    },
    {
      "epoch": 2.76,
      "learning_rate": 5.5218855218855225e-05,
      "loss": 4.2009,
      "step": 408000
    },
    {
      "epoch": 2.77,
      "learning_rate": 5.4994388327721666e-05,
      "loss": 4.2519,
      "step": 410000
    },
    {
      "epoch": 2.78,
      "learning_rate": 5.47699214365881e-05,
      "loss": 4.3561,
      "step": 412000
    },
    {
      "epoch": 2.8,
      "learning_rate": 5.4545454545454546e-05,
      "loss": 4.5544,
      "step": 414000
    },
    {
      "epoch": 2.81,
      "learning_rate": 5.4320987654320986e-05,
      "loss": 4.7515,
      "step": 416000
    },
    {
      "epoch": 2.82,
      "learning_rate": 5.4096520763187434e-05,
      "loss": 4.6801,
      "step": 418000
    },
    {
      "epoch": 2.84,
      "learning_rate": 5.3872053872053874e-05,
      "loss": 3.9784,
      "step": 420000
    },
    {
      "epoch": 2.85,
      "learning_rate": 5.3647586980920314e-05,
      "loss": 4.2571,
      "step": 422000
    },
    {
      "epoch": 2.86,
      "learning_rate": 5.342312008978676e-05,
      "loss": 4.3184,
      "step": 424000
    },
    {
      "epoch": 2.88,
      "learning_rate": 5.31986531986532e-05,
      "loss": 4.2075,
      "step": 426000
    },
    {
      "epoch": 2.89,
      "learning_rate": 5.297418630751965e-05,
      "loss": 4.2215,
      "step": 428000
    },
    {
      "epoch": 2.9,
      "learning_rate": 5.274971941638609e-05,
      "loss": 4.2475,
      "step": 430000
    },
    {
      "epoch": 2.92,
      "learning_rate": 5.2525252525252536e-05,
      "loss": 4.3064,
      "step": 432000
    },
    {
      "epoch": 2.93,
      "learning_rate": 5.230078563411896e-05,
      "loss": 4.3034,
      "step": 434000
    },
    {
      "epoch": 2.95,
      "learning_rate": 5.207631874298541e-05,
      "loss": 4.2119,
      "step": 436000
    },
    {
      "epoch": 2.96,
      "learning_rate": 5.185185185185185e-05,
      "loss": 4.2714,
      "step": 438000
    },
    {
      "epoch": 2.97,
      "learning_rate": 5.16273849607183e-05,
      "loss": 4.2878,
      "step": 440000
    },
    {
      "epoch": 2.99,
      "learning_rate": 5.140291806958474e-05,
      "loss": 4.1744,
      "step": 442000
    },
    {
      "epoch": 3.0,
      "learning_rate": 5.1178451178451184e-05,
      "loss": 3.8192,
      "step": 444000
    },
    {
      "epoch": 3.01,
      "learning_rate": 5.0953984287317624e-05,
      "loss": 4.0513,
      "step": 446000
    },
    {
      "epoch": 3.03,
      "learning_rate": 5.0729517396184065e-05,
      "loss": 4.0666,
      "step": 448000
    },
    {
      "epoch": 3.04,
      "learning_rate": 5.050505050505051e-05,
      "loss": 4.0297,
      "step": 450000
    },
    {
      "epoch": 3.05,
      "learning_rate": 5.028058361391695e-05,
      "loss": 4.019,
      "step": 452000
    },
    {
      "epoch": 3.07,
      "learning_rate": 5.00561167227834e-05,
      "loss": 4.0183,
      "step": 454000
    },
    {
      "epoch": 3.08,
      "learning_rate": 4.983164983164983e-05,
      "loss": 3.9988,
      "step": 456000
    },
    {
      "epoch": 3.09,
      "learning_rate": 4.960718294051628e-05,
      "loss": 3.9882,
      "step": 458000
    },
    {
      "epoch": 3.11,
      "learning_rate": 4.938271604938271e-05,
      "loss": 3.9879,
      "step": 460000
    },
    {
      "epoch": 3.12,
      "learning_rate": 4.915824915824916e-05,
      "loss": 3.9752,
      "step": 462000
    },
    {
      "epoch": 3.13,
      "learning_rate": 4.89337822671156e-05,
      "loss": 3.9901,
      "step": 464000
    },
    {
      "epoch": 3.15,
      "learning_rate": 4.870931537598205e-05,
      "loss": 3.9586,
      "step": 466000
    },
    {
      "epoch": 3.16,
      "learning_rate": 4.848484848484849e-05,
      "loss": 3.9711,
      "step": 468000
    },
    {
      "epoch": 3.17,
      "learning_rate": 4.8260381593714935e-05,
      "loss": 4.0005,
      "step": 470000
    },
    {
      "epoch": 3.19,
      "learning_rate": 4.803591470258137e-05,
      "loss": 3.9334,
      "step": 472000
    },
    {
      "epoch": 3.2,
      "learning_rate": 4.7811447811447815e-05,
      "loss": 3.9461,
      "step": 474000
    },
    {
      "epoch": 3.22,
      "learning_rate": 4.7586980920314256e-05,
      "loss": 3.9312,
      "step": 476000
    },
    {
      "epoch": 3.23,
      "learning_rate": 4.7362514029180696e-05,
      "loss": 3.928,
      "step": 478000
    },
    {
      "epoch": 3.24,
      "learning_rate": 4.713804713804714e-05,
      "loss": 3.9341,
      "step": 480000
    },
    {
      "epoch": 3.26,
      "learning_rate": 4.691358024691358e-05,
      "loss": 3.949,
      "step": 482000
    },
    {
      "epoch": 3.27,
      "learning_rate": 4.6689113355780024e-05,
      "loss": 3.9045,
      "step": 484000
    },
    {
      "epoch": 3.28,
      "learning_rate": 4.6464646464646464e-05,
      "loss": 3.9063,
      "step": 486000
    },
    {
      "epoch": 3.3,
      "learning_rate": 4.624017957351291e-05,
      "loss": 3.9079,
      "step": 488000
    },
    {
      "epoch": 3.31,
      "learning_rate": 4.601571268237935e-05,
      "loss": 3.9365,
      "step": 490000
    },
    {
      "epoch": 3.32,
      "learning_rate": 4.57912457912458e-05,
      "loss": 3.908,
      "step": 492000
    },
    {
      "epoch": 3.34,
      "learning_rate": 4.556677890011223e-05,
      "loss": 3.9043,
      "step": 494000
    },
    {
      "epoch": 3.35,
      "learning_rate": 4.534231200897868e-05,
      "loss": 3.9217,
      "step": 496000
    },
    {
      "epoch": 3.36,
      "learning_rate": 4.511784511784512e-05,
      "loss": 3.8926,
      "step": 498000
    },
    {
      "epoch": 3.38,
      "learning_rate": 4.4893378226711566e-05,
      "loss": 3.9196,
      "step": 500000
    },
    {
      "epoch": 3.39,
      "learning_rate": 4.4668911335578006e-05,
      "loss": 3.8883,
      "step": 502000
    },
    {
      "epoch": 3.4,
      "learning_rate": 4.4444444444444447e-05,
      "loss": 3.8888,
      "step": 504000
    },
    {
      "epoch": 3.42,
      "learning_rate": 4.421997755331089e-05,
      "loss": 3.9072,
      "step": 506000
    },
    {
      "epoch": 3.43,
      "learning_rate": 4.3995510662177334e-05,
      "loss": 3.8872,
      "step": 508000
    },
    {
      "epoch": 3.44,
      "learning_rate": 4.3771043771043774e-05,
      "loss": 3.8941,
      "step": 510000
    },
    {
      "epoch": 3.46,
      "learning_rate": 4.3546576879910214e-05,
      "loss": 3.8768,
      "step": 512000
    },
    {
      "epoch": 3.47,
      "learning_rate": 4.332210998877666e-05,
      "loss": 3.8667,
      "step": 514000
    },
    {
      "epoch": 3.49,
      "learning_rate": 4.3097643097643095e-05,
      "loss": 3.867,
      "step": 516000
    },
    {
      "epoch": 3.5,
      "learning_rate": 4.287317620650954e-05,
      "loss": 3.8665,
      "step": 518000
    },
    {
      "epoch": 3.51,
      "learning_rate": 4.264870931537598e-05,
      "loss": 3.8677,
      "step": 520000
    },
    {
      "epoch": 3.53,
      "learning_rate": 4.242424242424243e-05,
      "loss": 3.8752,
      "step": 522000
    },
    {
      "epoch": 3.54,
      "learning_rate": 4.219977553310887e-05,
      "loss": 3.888,
      "step": 524000
    },
    {
      "epoch": 3.55,
      "learning_rate": 4.197530864197531e-05,
      "loss": 3.849,
      "step": 526000
    },
    {
      "epoch": 3.57,
      "learning_rate": 4.175084175084175e-05,
      "loss": 3.8591,
      "step": 528000
    },
    {
      "epoch": 3.58,
      "learning_rate": 4.15263748597082e-05,
      "loss": 3.9935,
      "step": 530000
    },
    {
      "epoch": 3.59,
      "learning_rate": 4.130190796857464e-05,
      "loss": 4.1179,
      "step": 532000
    },
    {
      "epoch": 3.61,
      "learning_rate": 4.1077441077441085e-05,
      "loss": 3.8085,
      "step": 534000
    },
    {
      "epoch": 3.62,
      "learning_rate": 4.085297418630752e-05,
      "loss": 3.5572,
      "step": 536000
    },
    {
      "epoch": 3.63,
      "learning_rate": 4.0628507295173965e-05,
      "loss": 3.4772,
      "step": 538000
    },
    {
      "epoch": 3.65,
      "learning_rate": 4.0404040404040405e-05,
      "loss": 3.4955,
      "step": 540000
    },
    {
      "epoch": 3.66,
      "learning_rate": 4.0179573512906846e-05,
      "loss": 3.8805,
      "step": 542000
    },
    {
      "epoch": 3.67,
      "learning_rate": 3.995510662177329e-05,
      "loss": 3.7327,
      "step": 544000
    },
    {
      "epoch": 3.69,
      "learning_rate": 3.973063973063973e-05,
      "loss": 3.7417,
      "step": 546000
    },
    {
      "epoch": 3.7,
      "learning_rate": 3.950617283950617e-05,
      "loss": 4.0379,
      "step": 548000
    },
    {
      "epoch": 3.72,
      "learning_rate": 3.9281705948372613e-05,
      "loss": 4.0257,
      "step": 550000
    },
    {
      "epoch": 3.73,
      "learning_rate": 3.905723905723906e-05,
      "loss": 4.0116,
      "step": 552000
    },
    {
      "epoch": 3.74,
      "learning_rate": 3.88327721661055e-05,
      "loss": 3.9664,
      "step": 554000
    },
    {
      "epoch": 3.76,
      "learning_rate": 3.860830527497195e-05,
      "loss": 4.0149,
      "step": 556000
    },
    {
      "epoch": 3.77,
      "learning_rate": 3.838383838383838e-05,
      "loss": 4.0601,
      "step": 558000
    },
    {
      "epoch": 3.78,
      "learning_rate": 3.815937149270483e-05,
      "loss": 4.1639,
      "step": 560000
    },
    {
      "epoch": 3.8,
      "learning_rate": 3.793490460157127e-05,
      "loss": 4.3604,
      "step": 562000
    },
    {
      "epoch": 3.81,
      "learning_rate": 3.7710437710437716e-05,
      "loss": 4.5598,
      "step": 564000
    },
    {
      "epoch": 3.82,
      "learning_rate": 3.7485970819304156e-05,
      "loss": 4.5141,
      "step": 566000
    },
    {
      "epoch": 3.84,
      "learning_rate": 3.7261503928170596e-05,
      "loss": 3.8061,
      "step": 568000
    },
    {
      "epoch": 3.85,
      "learning_rate": 3.7037037037037037e-05,
      "loss": 4.0652,
      "step": 570000
    },
    {
      "epoch": 3.86,
      "learning_rate": 3.681257014590348e-05,
      "loss": 4.1378,
      "step": 572000
    },
    {
      "epoch": 3.88,
      "learning_rate": 3.6588103254769924e-05,
      "loss": 4.0305,
      "step": 574000
    },
    {
      "epoch": 3.89,
      "learning_rate": 3.6363636363636364e-05,
      "loss": 4.0567,
      "step": 576000
    },
    {
      "epoch": 3.9,
      "learning_rate": 3.613916947250281e-05,
      "loss": 4.0735,
      "step": 578000
    },
    {
      "epoch": 3.92,
      "learning_rate": 3.5914702581369245e-05,
      "loss": 4.1457,
      "step": 580000
    },
    {
      "epoch": 3.93,
      "learning_rate": 3.569023569023569e-05,
      "loss": 4.1228,
      "step": 582000
    },
    {
      "epoch": 3.94,
      "learning_rate": 3.546576879910213e-05,
      "loss": 4.061,
      "step": 584000
    },
    {
      "epoch": 3.96,
      "learning_rate": 3.524130190796858e-05,
      "loss": 4.1077,
      "step": 586000
    },
    {
      "epoch": 3.97,
      "learning_rate": 3.501683501683502e-05,
      "loss": 4.1329,
      "step": 588000
    },
    {
      "epoch": 3.99,
      "learning_rate": 3.4792368125701466e-05,
      "loss": 4.0207,
      "step": 590000
    },
    {
      "epoch": 4.0,
      "learning_rate": 3.45679012345679e-05,
      "loss": 3.6761,
      "step": 592000
    },
    {
      "epoch": 4.01,
      "learning_rate": 3.434343434343435e-05,
      "loss": 3.8304,
      "step": 594000
    },
    {
      "epoch": 4.03,
      "learning_rate": 3.411896745230079e-05,
      "loss": 3.8686,
      "step": 596000
    },
    {
      "epoch": 4.04,
      "learning_rate": 3.389450056116723e-05,
      "loss": 3.8363,
      "step": 598000
    },
    {
      "epoch": 4.05,
      "learning_rate": 3.3670033670033675e-05,
      "loss": 3.8302,
      "step": 600000
    },
    {
      "epoch": 4.07,
      "learning_rate": 3.3445566778900115e-05,
      "loss": 3.8316,
      "step": 602000
    },
    {
      "epoch": 4.08,
      "learning_rate": 3.3221099887766555e-05,
      "loss": 3.8158,
      "step": 604000
    },
    {
      "epoch": 4.09,
      "learning_rate": 3.2996632996632995e-05,
      "loss": 3.8058,
      "step": 606000
    },
    {
      "epoch": 4.11,
      "learning_rate": 3.277216610549944e-05,
      "loss": 3.8102,
      "step": 608000
    },
    {
      "epoch": 4.12,
      "learning_rate": 3.254769921436588e-05,
      "loss": 3.8001,
      "step": 610000
    },
    {
      "epoch": 4.13,
      "learning_rate": 3.232323232323233e-05,
      "loss": 3.8153,
      "step": 612000
    },
    {
      "epoch": 4.15,
      "learning_rate": 3.209876543209876e-05,
      "loss": 3.7888,
      "step": 614000
    },
    {
      "epoch": 4.16,
      "learning_rate": 3.187429854096521e-05,
      "loss": 3.8019,
      "step": 616000
    },
    {
      "epoch": 4.17,
      "learning_rate": 3.164983164983165e-05,
      "loss": 3.8379,
      "step": 618000
    },
    {
      "epoch": 4.19,
      "learning_rate": 3.14253647586981e-05,
      "loss": 3.7686,
      "step": 620000
    },
    {
      "epoch": 4.2,
      "learning_rate": 3.120089786756454e-05,
      "loss": 3.7825,
      "step": 622000
    },
    {
      "epoch": 4.22,
      "learning_rate": 3.097643097643098e-05,
      "loss": 3.7695,
      "step": 624000
    },
    {
      "epoch": 4.23,
      "learning_rate": 3.075196408529742e-05,
      "loss": 3.7713,
      "step": 626000
    },
    {
      "epoch": 4.24,
      "learning_rate": 3.0527497194163865e-05,
      "loss": 3.7789,
      "step": 628000
    },
    {
      "epoch": 4.26,
      "learning_rate": 3.0303030303030306e-05,
      "loss": 3.7927,
      "step": 630000
    },
    {
      "epoch": 4.27,
      "learning_rate": 3.007856341189675e-05,
      "loss": 3.7568,
      "step": 632000
    },
    {
      "epoch": 4.28,
      "learning_rate": 2.9854096520763186e-05,
      "loss": 3.7555,
      "step": 634000
    },
    {
      "epoch": 4.3,
      "learning_rate": 2.962962962962963e-05,
      "loss": 3.761,
      "step": 636000
    },
    {
      "epoch": 4.31,
      "learning_rate": 2.9405162738496074e-05,
      "loss": 3.7894,
      "step": 638000
    },
    {
      "epoch": 4.32,
      "learning_rate": 2.9180695847362517e-05,
      "loss": 3.7595,
      "step": 640000
    },
    {
      "epoch": 4.34,
      "learning_rate": 2.8956228956228958e-05,
      "loss": 3.763,
      "step": 642000
    },
    {
      "epoch": 4.35,
      "learning_rate": 2.87317620650954e-05,
      "loss": 3.7774,
      "step": 644000
    },
    {
      "epoch": 4.36,
      "learning_rate": 2.850729517396184e-05,
      "loss": 3.7551,
      "step": 646000
    },
    {
      "epoch": 4.38,
      "learning_rate": 2.8282828282828282e-05,
      "loss": 3.7829,
      "step": 648000
    },
    {
      "epoch": 4.39,
      "learning_rate": 2.8058361391694725e-05,
      "loss": 3.7527,
      "step": 650000
    },
    {
      "epoch": 4.4,
      "learning_rate": 2.783389450056117e-05,
      "loss": 3.753,
      "step": 652000
    },
    {
      "epoch": 4.42,
      "learning_rate": 2.7609427609427613e-05,
      "loss": 3.7746,
      "step": 654000
    },
    {
      "epoch": 4.43,
      "learning_rate": 2.738496071829405e-05,
      "loss": 3.7543,
      "step": 656000
    },
    {
      "epoch": 4.44,
      "learning_rate": 2.7160493827160493e-05,
      "loss": 3.7604,
      "step": 658000
    },
    {
      "epoch": 4.46,
      "learning_rate": 2.6936026936026937e-05,
      "loss": 3.7473,
      "step": 660000
    },
    {
      "epoch": 4.47,
      "learning_rate": 2.671156004489338e-05,
      "loss": 3.7377,
      "step": 662000
    },
    {
      "epoch": 4.49,
      "learning_rate": 2.6487093153759824e-05,
      "loss": 3.7398,
      "step": 664000
    },
    {
      "epoch": 4.5,
      "learning_rate": 2.6262626262626268e-05,
      "loss": 3.7432,
      "step": 666000
    },
    {
      "epoch": 4.51,
      "learning_rate": 2.6038159371492705e-05,
      "loss": 3.7484,
      "step": 668000
    },
    {
      "epoch": 4.53,
      "learning_rate": 2.581369248035915e-05,
      "loss": 3.7542,
      "step": 670000
    },
    {
      "epoch": 4.54,
      "learning_rate": 2.5589225589225592e-05,
      "loss": 3.7721,
      "step": 672000
    },
    {
      "epoch": 4.55,
      "learning_rate": 2.5364758698092032e-05,
      "loss": 3.7291,
      "step": 674000
    },
    {
      "epoch": 4.57,
      "learning_rate": 2.5140291806958476e-05,
      "loss": 3.7455,
      "step": 676000
    },
    {
      "epoch": 4.58,
      "learning_rate": 2.4915824915824916e-05,
      "loss": 3.8808,
      "step": 678000
    },
    {
      "epoch": 4.59,
      "learning_rate": 2.4691358024691357e-05,
      "loss": 4.0555,
      "step": 680000
    },
    {
      "epoch": 4.61,
      "learning_rate": 2.44668911335578e-05,
      "loss": 3.7132,
      "step": 682000
    },
    {
      "epoch": 4.62,
      "learning_rate": 2.4242424242424244e-05,
      "loss": 3.4602,
      "step": 684000
    },
    {
      "epoch": 4.63,
      "learning_rate": 2.4017957351290684e-05,
      "loss": 3.4057,
      "step": 686000
    },
    {
      "epoch": 4.65,
      "learning_rate": 2.3793490460157128e-05,
      "loss": 3.397,
      "step": 688000
    },
    {
      "epoch": 4.66,
      "learning_rate": 2.356902356902357e-05,
      "loss": 3.8375,
      "step": 690000
    },
    {
      "epoch": 4.67,
      "learning_rate": 2.3344556677890012e-05,
      "loss": 3.6615,
      "step": 692000
    },
    {
      "epoch": 4.69,
      "learning_rate": 2.3120089786756455e-05,
      "loss": 3.6624,
      "step": 694000
    },
    {
      "epoch": 4.7,
      "learning_rate": 2.28956228956229e-05,
      "loss": 3.9217,
      "step": 696000
    },
    {
      "epoch": 4.71,
      "learning_rate": 2.267115600448934e-05,
      "loss": 3.9308,
      "step": 698000
    },
    {
      "epoch": 4.73,
      "learning_rate": 2.2446689113355783e-05,
      "loss": 3.9061,
      "step": 700000
    },
    {
      "epoch": 4.74,
      "learning_rate": 2.2222222222222223e-05,
      "loss": 3.87,
      "step": 702000
    },
    {
      "epoch": 4.76,
      "learning_rate": 2.1997755331088667e-05,
      "loss": 3.919,
      "step": 704000
    },
    {
      "epoch": 4.77,
      "learning_rate": 2.1773288439955107e-05,
      "loss": 3.9621,
      "step": 706000
    },
    {
      "epoch": 4.78,
      "learning_rate": 2.1548821548821547e-05,
      "loss": 4.0667,
      "step": 708000
    },
    {
      "epoch": 4.8,
      "learning_rate": 2.132435465768799e-05,
      "loss": 4.2579,
      "step": 710000
    },
    {
      "epoch": 4.81,
      "learning_rate": 2.1099887766554435e-05,
      "loss": 4.4613,
      "step": 712000
    },
    {
      "epoch": 4.82,
      "learning_rate": 2.0875420875420875e-05,
      "loss": 4.4471,
      "step": 714000
    },
    {
      "epoch": 4.84,
      "learning_rate": 2.065095398428732e-05,
      "loss": 3.7612,
      "step": 716000
    },
    {
      "epoch": 4.85,
      "learning_rate": 2.042648709315376e-05,
      "loss": 3.9657,
      "step": 718000
    },
    {
      "epoch": 4.86,
      "learning_rate": 2.0202020202020203e-05,
      "loss": 4.0575,
      "step": 720000
    },
    {
      "epoch": 4.88,
      "learning_rate": 1.9977553310886646e-05,
      "loss": 3.9554,
      "step": 722000
    },
    {
      "epoch": 4.89,
      "learning_rate": 1.9753086419753087e-05,
      "loss": 3.9786,
      "step": 724000
    },
    {
      "epoch": 4.9,
      "learning_rate": 1.952861952861953e-05,
      "loss": 3.9809,
      "step": 726000
    },
    {
      "epoch": 4.92,
      "learning_rate": 1.9304152637485974e-05,
      "loss": 4.0793,
      "step": 728000
    },
    {
      "epoch": 4.93,
      "learning_rate": 1.9079685746352414e-05,
      "loss": 4.0373,
      "step": 730000
    },
    {
      "epoch": 4.94,
      "learning_rate": 1.8855218855218858e-05,
      "loss": 4.0029,
      "step": 732000
    },
    {
      "epoch": 4.96,
      "learning_rate": 1.8630751964085298e-05,
      "loss": 4.0193,
      "step": 734000
    },
    {
      "epoch": 4.97,
      "learning_rate": 1.840628507295174e-05,
      "loss": 4.0719,
      "step": 736000
    },
    {
      "epoch": 4.99,
      "learning_rate": 1.8181818181818182e-05,
      "loss": 3.9453,
      "step": 738000
    },
    {
      "epoch": 5.0,
      "learning_rate": 1.7957351290684622e-05,
      "loss": 3.6204,
      "step": 740000
    },
    {
      "epoch": 5.01,
      "learning_rate": 1.7732884399551066e-05,
      "loss": 3.6973,
      "step": 742000
    },
    {
      "epoch": 5.03,
      "learning_rate": 1.750841750841751e-05,
      "loss": 3.7598,
      "step": 744000
    },
    {
      "epoch": 5.04,
      "learning_rate": 1.728395061728395e-05,
      "loss": 3.7362,
      "step": 746000
    },
    {
      "epoch": 5.05,
      "learning_rate": 1.7059483726150394e-05,
      "loss": 3.7313,
      "step": 748000
    },
    {
      "epoch": 5.07,
      "learning_rate": 1.6835016835016837e-05,
      "loss": 3.7332,
      "step": 750000
    },
    {
      "epoch": 5.08,
      "learning_rate": 1.6610549943883278e-05,
      "loss": 3.722,
      "step": 752000
    },
    {
      "epoch": 5.09,
      "learning_rate": 1.638608305274972e-05,
      "loss": 3.713,
      "step": 754000
    },
    {
      "epoch": 5.11,
      "learning_rate": 1.6161616161616165e-05,
      "loss": 3.72,
      "step": 756000
    },
    {
      "epoch": 5.12,
      "learning_rate": 1.5937149270482605e-05,
      "loss": 3.7103,
      "step": 758000
    },
    {
      "epoch": 5.13,
      "learning_rate": 1.571268237934905e-05,
      "loss": 3.7315,
      "step": 760000
    },
    {
      "epoch": 5.15,
      "learning_rate": 1.548821548821549e-05,
      "loss": 3.7024,
      "step": 762000
    },
    {
      "epoch": 5.16,
      "learning_rate": 1.5263748597081933e-05,
      "loss": 3.7156,
      "step": 764000
    },
    {
      "epoch": 5.17,
      "learning_rate": 1.5039281705948375e-05,
      "loss": 3.7549,
      "step": 766000
    },
    {
      "epoch": 5.19,
      "learning_rate": 1.4814814814814815e-05,
      "loss": 3.6887,
      "step": 768000
    },
    {
      "epoch": 5.2,
      "learning_rate": 1.4590347923681259e-05,
      "loss": 3.7033,
      "step": 770000
    },
    {
      "epoch": 5.21,
      "learning_rate": 1.43658810325477e-05,
      "loss": 3.6917,
      "step": 772000
    },
    {
      "epoch": 5.23,
      "learning_rate": 1.4141414141414141e-05,
      "loss": 3.6936,
      "step": 774000
    },
    {
      "epoch": 5.24,
      "learning_rate": 1.3916947250280585e-05,
      "loss": 3.7019,
      "step": 776000
    },
    {
      "epoch": 5.26,
      "learning_rate": 1.3692480359147025e-05,
      "loss": 3.7172,
      "step": 778000
    },
    {
      "epoch": 5.27,
      "learning_rate": 1.3468013468013468e-05,
      "loss": 3.6812,
      "step": 780000
    },
    {
      "epoch": 5.28,
      "learning_rate": 1.3243546576879912e-05,
      "loss": 3.6841,
      "step": 782000
    },
    {
      "epoch": 5.3,
      "learning_rate": 1.3019079685746352e-05,
      "loss": 3.6837,
      "step": 784000
    },
    {
      "epoch": 5.31,
      "learning_rate": 1.2794612794612796e-05,
      "loss": 3.7225,
      "step": 786000
    },
    {
      "epoch": 5.32,
      "learning_rate": 1.2570145903479238e-05,
      "loss": 3.6897,
      "step": 788000
    },
    {
      "epoch": 5.34,
      "learning_rate": 1.2345679012345678e-05,
      "loss": 3.6941,
      "step": 790000
    },
    {
      "epoch": 5.35,
      "learning_rate": 1.2121212121212122e-05,
      "loss": 3.7125,
      "step": 792000
    },
    {
      "epoch": 5.36,
      "learning_rate": 1.1896745230078564e-05,
      "loss": 3.688,
      "step": 794000
    },
    {
      "epoch": 5.38,
      "learning_rate": 1.1672278338945006e-05,
      "loss": 3.7184,
      "step": 796000
    },
    {
      "epoch": 5.39,
      "learning_rate": 1.144781144781145e-05,
      "loss": 3.6901,
      "step": 798000
    },
    {
      "epoch": 5.4,
      "learning_rate": 1.1223344556677892e-05,
      "loss": 3.6886,
      "step": 800000
    },
    {
      "epoch": 5.42,
      "learning_rate": 1.0998877665544333e-05,
      "loss": 3.7129,
      "step": 802000
    },
    {
      "epoch": 5.43,
      "learning_rate": 1.0774410774410774e-05,
      "loss": 3.6953,
      "step": 804000
    },
    {
      "epoch": 5.44,
      "learning_rate": 1.0549943883277217e-05,
      "loss": 3.7016,
      "step": 806000
    },
    {
      "epoch": 5.46,
      "learning_rate": 1.032547699214366e-05,
      "loss": 3.69,
      "step": 808000
    },
    {
      "epoch": 5.47,
      "learning_rate": 1.0101010101010101e-05,
      "loss": 3.6834,
      "step": 810000
    },
    {
      "epoch": 5.48,
      "learning_rate": 9.876543209876543e-06,
      "loss": 3.6809,
      "step": 812000
    },
    {
      "epoch": 5.5,
      "learning_rate": 9.652076318742987e-06,
      "loss": 3.6861,
      "step": 814000
    },
    {
      "epoch": 5.51,
      "learning_rate": 9.427609427609429e-06,
      "loss": 3.6977,
      "step": 816000
    },
    {
      "epoch": 5.53,
      "learning_rate": 9.20314253647587e-06,
      "loss": 3.7002,
      "step": 818000
    },
    {
      "epoch": 5.54,
      "learning_rate": 8.978675645342311e-06,
      "loss": 3.7243,
      "step": 820000
    },
    {
      "epoch": 5.55,
      "learning_rate": 8.754208754208755e-06,
      "loss": 3.6768,
      "step": 822000
    },
    {
      "epoch": 5.57,
      "learning_rate": 8.529741863075197e-06,
      "loss": 3.6947,
      "step": 824000
    },
    {
      "epoch": 5.58,
      "learning_rate": 8.305274971941639e-06,
      "loss": 3.8538,
      "step": 826000
    },
    {
      "epoch": 5.59,
      "learning_rate": 8.080808080808082e-06,
      "loss": 4.0943,
      "step": 828000
    },
    {
      "epoch": 5.61,
      "learning_rate": 7.856341189674524e-06,
      "loss": 3.7048,
      "step": 830000
    },
    {
      "epoch": 5.62,
      "learning_rate": 7.631874298540966e-06,
      "loss": 3.4471,
      "step": 832000
    },
    {
      "epoch": 5.63,
      "learning_rate": 7.4074074074074075e-06,
      "loss": 3.4131,
      "step": 834000
    },
    {
      "epoch": 5.65,
      "learning_rate": 7.18294051627385e-06,
      "loss": 3.3965,
      "step": 836000
    },
    {
      "epoch": 5.66,
      "learning_rate": 6.958473625140292e-06,
      "loss": 3.9014,
      "step": 838000
    },
    {
      "epoch": 5.67,
      "learning_rate": 6.734006734006734e-06,
      "loss": 3.6888,
      "step": 840000
    },
    {
      "epoch": 5.69,
      "learning_rate": 6.509539842873176e-06,
      "loss": 3.6862,
      "step": 842000
    },
    {
      "epoch": 5.7,
      "learning_rate": 6.285072951739619e-06,
      "loss": 3.8646,
      "step": 844000
    },
    {
      "epoch": 5.71,
      "learning_rate": 6.060606060606061e-06,
      "loss": 3.8818,
      "step": 846000
    },
    {
      "epoch": 5.73,
      "learning_rate": 5.836139169472503e-06,
      "loss": 3.8666,
      "step": 848000
    },
    {
      "epoch": 5.74,
      "learning_rate": 5.611672278338946e-06,
      "loss": 3.8303,
      "step": 850000
    },
    {
      "epoch": 5.76,
      "learning_rate": 5.387205387205387e-06,
      "loss": 3.8764,
      "step": 852000
    },
    {
      "epoch": 5.77,
      "learning_rate": 5.16273849607183e-06,
      "loss": 3.9326,
      "step": 854000
    },
    {
      "epoch": 5.78,
      "learning_rate": 4.938271604938272e-06,
      "loss": 4.0327,
      "step": 856000
    },
    {
      "epoch": 5.8,
      "learning_rate": 4.7138047138047145e-06,
      "loss": 4.2255,
      "step": 858000
    },
    {
      "epoch": 5.81,
      "learning_rate": 4.489337822671156e-06,
      "loss": 4.4367,
      "step": 860000
    },
    {
      "epoch": 5.82,
      "learning_rate": 4.264870931537598e-06,
      "loss": 4.4619,
      "step": 862000
    },
    {
      "epoch": 5.84,
      "learning_rate": 4.040404040404041e-06,
      "loss": 3.8376,
      "step": 864000
    },
    {
      "epoch": 5.85,
      "learning_rate": 3.815937149270483e-06,
      "loss": 3.9389,
      "step": 866000
    },
    {
      "epoch": 5.86,
      "learning_rate": 3.591470258136925e-06,
      "loss": 4.051,
      "step": 868000
    },
    {
      "epoch": 5.88,
      "learning_rate": 3.367003367003367e-06,
      "loss": 3.9391,
      "step": 870000
    },
    {
      "epoch": 5.89,
      "learning_rate": 3.1425364758698095e-06,
      "loss": 3.969,
      "step": 872000
    },
    {
      "epoch": 5.9,
      "learning_rate": 2.9180695847362515e-06,
      "loss": 3.9726,
      "step": 874000
    },
    {
      "epoch": 5.92,
      "learning_rate": 2.6936026936026934e-06,
      "loss": 4.081,
      "step": 876000
    },
    {
      "epoch": 5.93,
      "learning_rate": 2.469135802469136e-06,
      "loss": 4.0299,
      "step": 878000
    },
    {
      "epoch": 5.94,
      "learning_rate": 2.244668911335578e-06,
      "loss": 4.0083,
      "step": 880000
    },
    {
      "epoch": 5.96,
      "learning_rate": 2.0202020202020206e-06,
      "loss": 4.0233,
      "step": 882000
    },
    {
      "epoch": 5.97,
      "learning_rate": 1.7957351290684626e-06,
      "loss": 4.0916,
      "step": 884000
    },
    {
      "epoch": 5.98,
      "learning_rate": 1.5712682379349048e-06,
      "loss": 3.9541,
      "step": 886000
    },
    {
      "epoch": 6.0,
      "learning_rate": 1.3468013468013467e-06,
      "loss": 3.6427,
      "step": 888000
    },
    {
      "epoch": 6.01,
      "learning_rate": 1.122334455667789e-06,
      "loss": 3.6296,
      "step": 890000
    },
    {
      "epoch": 6.03,
      "learning_rate": 8.978675645342313e-07,
      "loss": 3.7085,
      "step": 892000
    },
    {
      "epoch": 6.04,
      "learning_rate": 6.734006734006734e-07,
      "loss": 3.6918,
      "step": 894000
    },
    {
      "epoch": 6.05,
      "learning_rate": 4.4893378226711564e-07,
      "loss": 3.6907,
      "step": 896000
    },
    {
      "epoch": 6.07,
      "learning_rate": 2.2446689113355782e-07,
      "loss": 3.6985,
      "step": 898000
    },
    {
      "epoch": 6.08,
      "learning_rate": 0.0,
      "loss": 3.6945,
      "step": 900000
    }
  ],
  "max_steps": 900000,
  "num_train_epochs": 7,
  "total_flos": 1.894127697690476e+18,
  "trial_name": null,
  "trial_params": null
}