{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 287076,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 4.991291504688654e-05,
      "loss": 5.9932,
      "step": 500
    },
    {
      "epoch": 0.01,
      "learning_rate": 4.982583009377308e-05,
      "loss": 4.5752,
      "step": 1000
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.973874514065962e-05,
      "loss": 3.9864,
      "step": 1500
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.9651660187546154e-05,
      "loss": 3.7187,
      "step": 2000
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.95645752344327e-05,
      "loss": 3.5442,
      "step": 2500
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.947749028131923e-05,
      "loss": 3.378,
      "step": 3000
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.939040532820577e-05,
      "loss": 3.2869,
      "step": 3500
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.930332037509231e-05,
      "loss": 3.2232,
      "step": 4000
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.921623542197885e-05,
      "loss": 3.1339,
      "step": 4500
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.912915046886539e-05,
      "loss": 3.1246,
      "step": 5000
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.904206551575193e-05,
      "loss": 3.0439,
      "step": 5500
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.895498056263847e-05,
      "loss": 2.9875,
      "step": 6000
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.8867895609525004e-05,
      "loss": 2.9797,
      "step": 6500
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.878081065641154e-05,
      "loss": 2.9155,
      "step": 7000
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.869372570329808e-05,
      "loss": 2.9334,
      "step": 7500
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.860664075018462e-05,
      "loss": 2.8763,
      "step": 8000
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.851955579707116e-05,
      "loss": 2.8501,
      "step": 8500
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.84324708439577e-05,
      "loss": 2.825,
      "step": 9000
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.834538589084424e-05,
      "loss": 2.8267,
      "step": 9500
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.8258300937730774e-05,
      "loss": 2.7971,
      "step": 10000
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.817121598461732e-05,
      "loss": 2.7697,
      "step": 10500
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.808413103150385e-05,
      "loss": 2.7516,
      "step": 11000
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.799704607839039e-05,
      "loss": 2.721,
      "step": 11500
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.790996112527693e-05,
      "loss": 2.6979,
      "step": 12000
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.782287617216347e-05,
      "loss": 2.6872,
      "step": 12500
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.7735791219050005e-05,
      "loss": 2.6909,
      "step": 13000
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.764870626593655e-05,
      "loss": 2.664,
      "step": 13500
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.7561621312823084e-05,
      "loss": 2.6515,
      "step": 14000
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.747453635970962e-05,
      "loss": 2.655,
      "step": 14500
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.738745140659617e-05,
      "loss": 2.6327,
      "step": 15000
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.73003664534827e-05,
      "loss": 2.6356,
      "step": 15500
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.721328150036924e-05,
      "loss": 2.6308,
      "step": 16000
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.712619654725578e-05,
      "loss": 2.6302,
      "step": 16500
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.703911159414232e-05,
      "loss": 2.6208,
      "step": 17000
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.6952026641028854e-05,
      "loss": 2.6159,
      "step": 17500
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.68649416879154e-05,
      "loss": 2.5992,
      "step": 18000
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.677785673480193e-05,
      "loss": 2.5901,
      "step": 18500
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.669077178168847e-05,
      "loss": 2.5911,
      "step": 19000
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.660368682857502e-05,
      "loss": 2.5278,
      "step": 19500
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.651660187546155e-05,
      "loss": 2.5569,
      "step": 20000
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.642951692234809e-05,
      "loss": 2.5522,
      "step": 20500
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.634243196923463e-05,
      "loss": 2.5293,
      "step": 21000
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.625534701612117e-05,
      "loss": 2.5006,
      "step": 21500
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.6168262063007704e-05,
      "loss": 2.5361,
      "step": 22000
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.608117710989425e-05,
      "loss": 2.5116,
      "step": 22500
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.599409215678078e-05,
      "loss": 2.4744,
      "step": 23000
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.590700720366732e-05,
      "loss": 2.5232,
      "step": 23500
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.581992225055386e-05,
      "loss": 2.4808,
      "step": 24000
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.57328372974404e-05,
      "loss": 2.4899,
      "step": 24500
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.564575234432694e-05,
      "loss": 2.4656,
      "step": 25000
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.555866739121348e-05,
      "loss": 2.4816,
      "step": 25500
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.547158243810002e-05,
      "loss": 2.4736,
      "step": 26000
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.538449748498655e-05,
      "loss": 2.4767,
      "step": 26500
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.529741253187309e-05,
      "loss": 2.4841,
      "step": 27000
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.521032757875963e-05,
      "loss": 2.4883,
      "step": 27500
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.512324262564617e-05,
      "loss": 2.4851,
      "step": 28000
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.503615767253271e-05,
      "loss": 2.4384,
      "step": 28500
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.494907271941925e-05,
      "loss": 2.4828,
      "step": 29000
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.486198776630579e-05,
      "loss": 2.4509,
      "step": 29500
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.4774902813192324e-05,
      "loss": 2.4772,
      "step": 30000
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.468781786007887e-05,
      "loss": 2.4236,
      "step": 30500
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.46007329069654e-05,
      "loss": 2.4359,
      "step": 31000
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.451364795385194e-05,
      "loss": 2.4348,
      "step": 31500
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.442656300073848e-05,
      "loss": 2.4512,
      "step": 32000
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.433947804762502e-05,
      "loss": 2.4132,
      "step": 32500
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.425239309451156e-05,
      "loss": 2.4363,
      "step": 33000
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.41653081413981e-05,
      "loss": 2.4347,
      "step": 33500
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.407822318828464e-05,
      "loss": 2.3936,
      "step": 34000
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.399113823517117e-05,
      "loss": 2.4141,
      "step": 34500
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.390405328205772e-05,
      "loss": 2.4454,
      "step": 35000
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.381696832894425e-05,
      "loss": 2.398,
      "step": 35500
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.372988337583079e-05,
      "loss": 2.4122,
      "step": 36000
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.364279842271733e-05,
      "loss": 2.4089,
      "step": 36500
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.355571346960387e-05,
      "loss": 2.3848,
      "step": 37000
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.3468628516490404e-05,
      "loss": 2.3909,
      "step": 37500
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.338154356337695e-05,
      "loss": 2.4077,
      "step": 38000
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.329445861026349e-05,
      "loss": 2.3876,
      "step": 38500
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.320737365715002e-05,
      "loss": 2.3946,
      "step": 39000
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.312028870403656e-05,
      "loss": 2.3676,
      "step": 39500
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.30332037509231e-05,
      "loss": 2.3862,
      "step": 40000
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.294611879780964e-05,
      "loss": 2.3724,
      "step": 40500
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.285903384469618e-05,
      "loss": 2.3735,
      "step": 41000
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.277194889158272e-05,
      "loss": 2.3679,
      "step": 41500
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.268486393846925e-05,
      "loss": 2.3536,
      "step": 42000
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.259777898535579e-05,
      "loss": 2.3705,
      "step": 42500
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.251069403224234e-05,
      "loss": 2.3386,
      "step": 43000
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.242360907912887e-05,
      "loss": 2.3387,
      "step": 43500
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.233652412601541e-05,
      "loss": 2.3283,
      "step": 44000
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.224943917290195e-05,
      "loss": 2.3456,
      "step": 44500
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.216235421978849e-05,
      "loss": 2.3318,
      "step": 45000
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.2075269266675024e-05,
      "loss": 2.3478,
      "step": 45500
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.198818431356157e-05,
      "loss": 2.3435,
      "step": 46000
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.19010993604481e-05,
      "loss": 2.3509,
      "step": 46500
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.181401440733464e-05,
      "loss": 2.3666,
      "step": 47000
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.172692945422119e-05,
      "loss": 2.3244,
      "step": 47500
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.163984450110772e-05,
      "loss": 2.3507,
      "step": 48000
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.155275954799426e-05,
      "loss": 2.2845,
      "step": 48500
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.14656745948808e-05,
      "loss": 2.3202,
      "step": 49000
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.137858964176734e-05,
      "loss": 2.3355,
      "step": 49500
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.129150468865387e-05,
      "loss": 2.31,
      "step": 50000
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.120441973554042e-05,
      "loss": 2.2929,
      "step": 50500
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.111733478242695e-05,
      "loss": 2.2909,
      "step": 51000
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.103024982931349e-05,
      "loss": 2.2868,
      "step": 51500
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.094316487620004e-05,
      "loss": 2.2951,
      "step": 52000
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.085607992308657e-05,
      "loss": 2.2917,
      "step": 52500
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.076899496997311e-05,
      "loss": 2.3179,
      "step": 53000
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.068191001685965e-05,
      "loss": 2.2887,
      "step": 53500
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.059482506374619e-05,
      "loss": 2.2864,
      "step": 54000
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.050774011063272e-05,
      "loss": 2.2646,
      "step": 54500
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.042065515751927e-05,
      "loss": 2.3021,
      "step": 55000
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.03335702044058e-05,
      "loss": 2.2817,
      "step": 55500
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.024648525129234e-05,
      "loss": 2.2674,
      "step": 56000
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.015940029817888e-05,
      "loss": 2.296,
      "step": 56500
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.007231534506542e-05,
      "loss": 2.2639,
      "step": 57000
    },
    {
      "epoch": 0.6,
      "learning_rate": 3.998523039195196e-05,
      "loss": 2.2967,
      "step": 57500
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.98981454388385e-05,
      "loss": 2.245,
      "step": 58000
    },
    {
      "epoch": 0.61,
      "learning_rate": 3.981106048572504e-05,
      "loss": 2.2482,
      "step": 58500
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.972397553261157e-05,
      "loss": 2.2715,
      "step": 59000
    },
    {
      "epoch": 0.62,
      "learning_rate": 3.963689057949811e-05,
      "loss": 2.251,
      "step": 59500
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.954980562638465e-05,
      "loss": 2.2595,
      "step": 60000
    },
    {
      "epoch": 0.63,
      "learning_rate": 3.946272067327119e-05,
      "loss": 2.2864,
      "step": 60500
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.937563572015773e-05,
      "loss": 2.2546,
      "step": 61000
    },
    {
      "epoch": 0.64,
      "learning_rate": 3.928855076704427e-05,
      "loss": 2.2461,
      "step": 61500
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.920146581393081e-05,
      "loss": 2.2758,
      "step": 62000
    },
    {
      "epoch": 0.65,
      "learning_rate": 3.911438086081734e-05,
      "loss": 2.2608,
      "step": 62500
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.902729590770389e-05,
      "loss": 2.2611,
      "step": 63000
    },
    {
      "epoch": 0.66,
      "learning_rate": 3.894021095459042e-05,
      "loss": 2.252,
      "step": 63500
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.885312600147696e-05,
      "loss": 2.2721,
      "step": 64000
    },
    {
      "epoch": 0.67,
      "learning_rate": 3.87660410483635e-05,
      "loss": 2.2638,
      "step": 64500
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.867895609525004e-05,
      "loss": 2.2493,
      "step": 65000
    },
    {
      "epoch": 0.68,
      "learning_rate": 3.859187114213657e-05,
      "loss": 2.2404,
      "step": 65500
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.850478618902312e-05,
      "loss": 2.2588,
      "step": 66000
    },
    {
      "epoch": 0.69,
      "learning_rate": 3.841770123590966e-05,
      "loss": 2.2567,
      "step": 66500
    },
    {
      "epoch": 0.7,
      "learning_rate": 3.833061628279619e-05,
      "loss": 2.2333,
      "step": 67000
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.824353132968274e-05,
      "loss": 2.2266,
      "step": 67500
    },
    {
      "epoch": 0.71,
      "learning_rate": 3.815644637656927e-05,
      "loss": 2.2122,
      "step": 68000
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.806936142345581e-05,
      "loss": 2.2624,
      "step": 68500
    },
    {
      "epoch": 0.72,
      "learning_rate": 3.798227647034235e-05,
      "loss": 2.1801,
      "step": 69000
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.789519151722889e-05,
      "loss": 2.2352,
      "step": 69500
    },
    {
      "epoch": 0.73,
      "learning_rate": 3.780810656411542e-05,
      "loss": 2.2211,
      "step": 70000
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.772102161100197e-05,
      "loss": 2.2461,
      "step": 70500
    },
    {
      "epoch": 0.74,
      "learning_rate": 3.763393665788851e-05,
      "loss": 2.2082,
      "step": 71000
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.754685170477504e-05,
      "loss": 2.2128,
      "step": 71500
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.745976675166159e-05,
      "loss": 2.2142,
      "step": 72000
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.737268179854812e-05,
      "loss": 2.2365,
      "step": 72500
    },
    {
      "epoch": 0.76,
      "learning_rate": 3.728559684543466e-05,
      "loss": 2.2143,
      "step": 73000
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.71985118923212e-05,
      "loss": 2.2392,
      "step": 73500
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.711142693920774e-05,
      "loss": 2.2237,
      "step": 74000
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.702434198609427e-05,
      "loss": 2.1729,
      "step": 74500
    },
    {
      "epoch": 0.78,
      "learning_rate": 3.693725703298082e-05,
      "loss": 2.2192,
      "step": 75000
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.685017207986736e-05,
      "loss": 2.2138,
      "step": 75500
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.676308712675389e-05,
      "loss": 2.2359,
      "step": 76000
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.667600217364043e-05,
      "loss": 2.1912,
      "step": 76500
    },
    {
      "epoch": 0.8,
      "learning_rate": 3.658891722052697e-05,
      "loss": 2.2056,
      "step": 77000
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.650183226741351e-05,
      "loss": 2.1702,
      "step": 77500
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.641474731430004e-05,
      "loss": 2.1527,
      "step": 78000
    },
    {
      "epoch": 0.82,
      "learning_rate": 3.632766236118659e-05,
      "loss": 2.1872,
      "step": 78500
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.624057740807312e-05,
      "loss": 2.1901,
      "step": 79000
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.615349245495966e-05,
      "loss": 2.1773,
      "step": 79500
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.60664075018462e-05,
      "loss": 2.1708,
      "step": 80000
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.597932254873274e-05,
      "loss": 2.2007,
      "step": 80500
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.589223759561928e-05,
      "loss": 2.1826,
      "step": 81000
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.580515264250582e-05,
      "loss": 2.1822,
      "step": 81500
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.571806768939236e-05,
      "loss": 2.1832,
      "step": 82000
    },
    {
      "epoch": 0.86,
      "learning_rate": 3.563098273627889e-05,
      "loss": 2.1715,
      "step": 82500
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.554389778316544e-05,
      "loss": 2.1889,
      "step": 83000
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.545681283005197e-05,
      "loss": 2.1901,
      "step": 83500
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.536972787693851e-05,
      "loss": 2.1805,
      "step": 84000
    },
    {
      "epoch": 0.88,
      "learning_rate": 3.528264292382505e-05,
      "loss": 2.1482,
      "step": 84500
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.519555797071159e-05,
      "loss": 2.189,
      "step": 85000
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.510847301759813e-05,
      "loss": 2.1871,
      "step": 85500
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.502138806448467e-05,
      "loss": 2.1471,
      "step": 86000
    },
    {
      "epoch": 0.9,
      "learning_rate": 3.493430311137121e-05,
      "loss": 2.1751,
      "step": 86500
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.484721815825774e-05,
      "loss": 2.1785,
      "step": 87000
    },
    {
      "epoch": 0.91,
      "learning_rate": 3.476013320514429e-05,
      "loss": 2.1597,
      "step": 87500
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.467304825203082e-05,
      "loss": 2.1878,
      "step": 88000
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.458596329891736e-05,
      "loss": 2.168,
      "step": 88500
    },
    {
      "epoch": 0.93,
      "learning_rate": 3.44988783458039e-05,
      "loss": 2.1748,
      "step": 89000
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.441179339269044e-05,
      "loss": 2.2079,
      "step": 89500
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.432470843957698e-05,
      "loss": 2.1503,
      "step": 90000
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.423762348646352e-05,
      "loss": 2.1362,
      "step": 90500
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.415053853335006e-05,
      "loss": 2.1269,
      "step": 91000
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.406345358023659e-05,
      "loss": 2.1609,
      "step": 91500
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.397636862712313e-05,
      "loss": 2.1505,
      "step": 92000
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.388928367400967e-05,
      "loss": 2.1131,
      "step": 92500
    },
    {
      "epoch": 0.97,
      "learning_rate": 3.380219872089621e-05,
      "loss": 2.1455,
      "step": 93000
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.371511376778275e-05,
      "loss": 2.1455,
      "step": 93500
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.362802881466929e-05,
      "loss": 2.1504,
      "step": 94000
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.354094386155583e-05,
      "loss": 2.1506,
      "step": 94500
    },
    {
      "epoch": 0.99,
      "learning_rate": 3.345385890844236e-05,
      "loss": 2.1657,
      "step": 95000
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.336677395532891e-05,
      "loss": 2.1465,
      "step": 95500
    },
    {
      "epoch": 1.0,
      "learning_rate": 3.327968900221544e-05,
      "loss": 2.1235,
      "step": 96000
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.319260404910198e-05,
      "loss": 2.101,
      "step": 96500
    },
    {
      "epoch": 1.01,
      "learning_rate": 3.310551909598852e-05,
      "loss": 2.1058,
      "step": 97000
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.301843414287506e-05,
      "loss": 2.0882,
      "step": 97500
    },
    {
      "epoch": 1.02,
      "learning_rate": 3.293134918976159e-05,
      "loss": 2.0591,
      "step": 98000
    },
    {
      "epoch": 1.03,
      "learning_rate": 3.284426423664814e-05,
      "loss": 2.123,
      "step": 98500
    },
    {
      "epoch": 1.03,
      "learning_rate": 3.275717928353468e-05,
      "loss": 2.07,
      "step": 99000
    },
    {
      "epoch": 1.04,
      "learning_rate": 3.267009433042121e-05,
      "loss": 2.1077,
      "step": 99500
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.258300937730776e-05,
      "loss": 2.0761,
      "step": 100000
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.249592442419429e-05,
      "loss": 2.0634,
      "step": 100500
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.240883947108083e-05,
      "loss": 2.1005,
      "step": 101000
    },
    {
      "epoch": 1.06,
      "learning_rate": 3.232175451796737e-05,
      "loss": 2.108,
      "step": 101500
    },
    {
      "epoch": 1.07,
      "learning_rate": 3.223466956485391e-05,
      "loss": 2.0803,
      "step": 102000
    },
    {
      "epoch": 1.07,
      "learning_rate": 3.214758461174044e-05,
      "loss": 2.1421,
      "step": 102500
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.206049965862699e-05,
      "loss": 2.1066,
      "step": 103000
    },
    {
      "epoch": 1.08,
      "learning_rate": 3.197341470551352e-05,
      "loss": 2.0903,
      "step": 103500
    },
    {
      "epoch": 1.09,
      "learning_rate": 3.188632975240006e-05,
      "loss": 2.0872,
      "step": 104000
    },
    {
      "epoch": 1.09,
      "learning_rate": 3.179924479928661e-05,
      "loss": 2.1146,
      "step": 104500
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.171215984617314e-05,
      "loss": 2.1223,
      "step": 105000
    },
    {
      "epoch": 1.1,
      "learning_rate": 3.162507489305968e-05,
      "loss": 2.1103,
      "step": 105500
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.153798993994622e-05,
      "loss": 2.0907,
      "step": 106000
    },
    {
      "epoch": 1.11,
      "learning_rate": 3.145090498683276e-05,
      "loss": 2.0805,
      "step": 106500
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.136382003371929e-05,
      "loss": 2.0863,
      "step": 107000
    },
    {
      "epoch": 1.12,
      "learning_rate": 3.127673508060584e-05,
      "loss": 2.0719,
      "step": 107500
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.118965012749237e-05,
      "loss": 2.0808,
      "step": 108000
    },
    {
      "epoch": 1.13,
      "learning_rate": 3.110256517437891e-05,
      "loss": 2.0602,
      "step": 108500
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.101548022126545e-05,
      "loss": 2.0837,
      "step": 109000
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.092839526815199e-05,
      "loss": 2.0494,
      "step": 109500
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.084131031503853e-05,
      "loss": 2.0456,
      "step": 110000
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.075422536192507e-05,
      "loss": 2.0486,
      "step": 110500
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.066714040881161e-05,
      "loss": 2.0681,
      "step": 111000
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.058005545569814e-05,
      "loss": 2.071,
      "step": 111500
    },
    {
      "epoch": 1.17,
      "learning_rate": 3.0492970502584684e-05,
      "loss": 2.0743,
      "step": 112000
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.040588554947122e-05,
      "loss": 2.0461,
      "step": 112500
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.031880059635776e-05,
      "loss": 2.0534,
      "step": 113000
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.0231715643244303e-05,
      "loss": 2.0593,
      "step": 113500
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.014463069013084e-05,
      "loss": 2.0678,
      "step": 114000
    },
    {
      "epoch": 1.2,
      "learning_rate": 3.005754573701738e-05,
      "loss": 2.0111,
      "step": 114500
    },
    {
      "epoch": 1.2,
      "learning_rate": 2.9970460783903915e-05,
      "loss": 2.0568,
      "step": 115000
    },
    {
      "epoch": 1.21,
      "learning_rate": 2.9883375830790454e-05,
      "loss": 2.0486,
      "step": 115500
    },
    {
      "epoch": 1.21,
      "learning_rate": 2.979629087767699e-05,
      "loss": 2.07,
      "step": 116000
    },
    {
      "epoch": 1.22,
      "learning_rate": 2.9709205924563534e-05,
      "loss": 2.0438,
      "step": 116500
    },
    {
      "epoch": 1.22,
      "learning_rate": 2.9622120971450066e-05,
      "loss": 2.063,
      "step": 117000
    },
    {
      "epoch": 1.23,
      "learning_rate": 2.953503601833661e-05,
      "loss": 2.0837,
      "step": 117500
    },
    {
      "epoch": 1.23,
      "learning_rate": 2.944795106522315e-05,
      "loss": 2.0608,
      "step": 118000
    },
    {
      "epoch": 1.24,
      "learning_rate": 2.9360866112109685e-05,
      "loss": 2.0656,
      "step": 118500
    },
    {
      "epoch": 1.24,
      "learning_rate": 2.9273781158996228e-05,
      "loss": 2.0592,
      "step": 119000
    },
    {
      "epoch": 1.25,
      "learning_rate": 2.9186696205882764e-05,
      "loss": 2.0735,
      "step": 119500
    },
    {
      "epoch": 1.25,
      "learning_rate": 2.9099611252769304e-05,
      "loss": 2.0314,
      "step": 120000
    },
    {
      "epoch": 1.26,
      "learning_rate": 2.901252629965584e-05,
      "loss": 2.0442,
      "step": 120500
    },
    {
      "epoch": 1.26,
      "learning_rate": 2.892544134654238e-05,
      "loss": 2.0758,
      "step": 121000
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.8838356393428916e-05,
      "loss": 2.0281,
      "step": 121500
    },
    {
      "epoch": 1.27,
      "learning_rate": 2.875127144031546e-05,
      "loss": 2.0676,
      "step": 122000
    },
    {
      "epoch": 1.28,
      "learning_rate": 2.8664186487202e-05,
      "loss": 2.0562,
      "step": 122500
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.8577101534088535e-05,
      "loss": 2.0334,
      "step": 123000
    },
    {
      "epoch": 1.29,
      "learning_rate": 2.8490016580975078e-05,
      "loss": 2.04,
      "step": 123500
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.840293162786161e-05,
      "loss": 2.0521,
      "step": 124000
    },
    {
      "epoch": 1.3,
      "learning_rate": 2.8315846674748153e-05,
      "loss": 2.0218,
      "step": 124500
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.822876172163469e-05,
      "loss": 2.0267,
      "step": 125000
    },
    {
      "epoch": 1.31,
      "learning_rate": 2.814167676852123e-05,
      "loss": 2.0198,
      "step": 125500
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.8054591815407765e-05,
      "loss": 2.0346,
      "step": 126000
    },
    {
      "epoch": 1.32,
      "learning_rate": 2.796750686229431e-05,
      "loss": 2.0275,
      "step": 126500
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.788042190918084e-05,
      "loss": 2.0263,
      "step": 127000
    },
    {
      "epoch": 1.33,
      "learning_rate": 2.7793336956067384e-05,
      "loss": 2.0065,
      "step": 127500
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.7706252002953924e-05,
      "loss": 2.0181,
      "step": 128000
    },
    {
      "epoch": 1.34,
      "learning_rate": 2.761916704984046e-05,
      "loss": 2.0094,
      "step": 128500
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.7532082096727003e-05,
      "loss": 2.0347,
      "step": 129000
    },
    {
      "epoch": 1.35,
      "learning_rate": 2.744499714361354e-05,
      "loss": 2.0015,
      "step": 129500
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.735791219050008e-05,
      "loss": 1.9911,
      "step": 130000
    },
    {
      "epoch": 1.36,
      "learning_rate": 2.7270827237386615e-05,
      "loss": 2.0332,
      "step": 130500
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.7183742284273155e-05,
      "loss": 2.0351,
      "step": 131000
    },
    {
      "epoch": 1.37,
      "learning_rate": 2.709665733115969e-05,
      "loss": 2.0185,
      "step": 131500
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.7009572378046234e-05,
      "loss": 2.0057,
      "step": 132000
    },
    {
      "epoch": 1.38,
      "learning_rate": 2.6922487424932773e-05,
      "loss": 2.0095,
      "step": 132500
    },
    {
      "epoch": 1.39,
      "learning_rate": 2.683540247181931e-05,
      "loss": 2.0363,
      "step": 133000
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.6748317518705852e-05,
      "loss": 2.0017,
      "step": 133500
    },
    {
      "epoch": 1.4,
      "learning_rate": 2.6661232565592385e-05,
      "loss": 2.0176,
      "step": 134000
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.6574147612478928e-05,
      "loss": 1.9988,
      "step": 134500
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.6487062659365464e-05,
      "loss": 2.0006,
      "step": 135000
    },
    {
      "epoch": 1.42,
      "learning_rate": 2.6399977706252004e-05,
      "loss": 1.9966,
      "step": 135500
    },
    {
      "epoch": 1.42,
      "learning_rate": 2.631289275313854e-05,
      "loss": 1.9746,
      "step": 136000
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.6225807800025083e-05,
      "loss": 2.0186,
      "step": 136500
    },
    {
      "epoch": 1.43,
      "learning_rate": 2.6138722846911623e-05,
      "loss": 1.9999,
      "step": 137000
    },
    {
      "epoch": 1.44,
      "learning_rate": 2.605163789379816e-05,
      "loss": 2.0056,
      "step": 137500
    },
    {
      "epoch": 1.44,
      "learning_rate": 2.59645529406847e-05,
      "loss": 1.9837,
      "step": 138000
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.5877467987571235e-05,
      "loss": 1.9683,
      "step": 138500
    },
    {
      "epoch": 1.45,
      "learning_rate": 2.5790383034457778e-05,
      "loss": 2.0063,
      "step": 139000
    },
    {
      "epoch": 1.46,
      "learning_rate": 2.5703298081344314e-05,
      "loss": 1.9752,
      "step": 139500
    },
    {
      "epoch": 1.46,
      "learning_rate": 2.5616213128230854e-05,
      "loss": 1.977,
      "step": 140000
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.552912817511739e-05,
      "loss": 1.9953,
      "step": 140500
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.544204322200393e-05,
      "loss": 1.9789,
      "step": 141000
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.5354958268890472e-05,
      "loss": 2.0108,
      "step": 141500
    },
    {
      "epoch": 1.48,
      "learning_rate": 2.526787331577701e-05,
      "loss": 1.9867,
      "step": 142000
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.5180788362663548e-05,
      "loss": 1.993,
      "step": 142500
    },
    {
      "epoch": 1.49,
      "learning_rate": 2.5093703409550084e-05,
      "loss": 1.9721,
      "step": 143000
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.5006618456436627e-05,
      "loss": 1.9846,
      "step": 143500
    },
    {
      "epoch": 1.5,
      "learning_rate": 2.4919533503323164e-05,
      "loss": 1.9779,
      "step": 144000
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.4832448550209703e-05,
      "loss": 1.9963,
      "step": 144500
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.4745363597096243e-05,
      "loss": 1.9521,
      "step": 145000
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.465827864398278e-05,
      "loss": 1.9644,
      "step": 145500
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.457119369086932e-05,
      "loss": 1.97,
      "step": 146000
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.4484108737755858e-05,
      "loss": 1.9816,
      "step": 146500
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.4397023784642394e-05,
      "loss": 1.9872,
      "step": 147000
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.4309938831528934e-05,
      "loss": 1.9639,
      "step": 147500
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.4222853878415473e-05,
      "loss": 1.9679,
      "step": 148000
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.4135768925302013e-05,
      "loss": 1.966,
      "step": 148500
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.4048683972188553e-05,
      "loss": 1.9513,
      "step": 149000
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.396159901907509e-05,
      "loss": 1.9739,
      "step": 149500
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.387451406596163e-05,
      "loss": 1.9763,
      "step": 150000
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.3787429112848168e-05,
      "loss": 1.943,
      "step": 150500
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.3700344159734704e-05,
      "loss": 1.9663,
      "step": 151000
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.3613259206621244e-05,
      "loss": 1.9904,
      "step": 151500
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.3526174253507783e-05,
      "loss": 1.9495,
      "step": 152000
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.343908930039432e-05,
      "loss": 1.9218,
      "step": 152500
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.335200434728086e-05,
      "loss": 1.9479,
      "step": 153000
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.32649193941674e-05,
      "loss": 1.9625,
      "step": 153500
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.317783444105394e-05,
      "loss": 1.9426,
      "step": 154000
    },
    {
      "epoch": 1.61,
      "learning_rate": 2.3090749487940478e-05,
      "loss": 1.94,
      "step": 154500
    },
    {
      "epoch": 1.62,
      "learning_rate": 2.3003664534827014e-05,
      "loss": 1.9457,
      "step": 155000
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.2916579581713554e-05,
      "loss": 1.9589,
      "step": 155500
    },
    {
      "epoch": 1.63,
      "learning_rate": 2.2829494628600093e-05,
      "loss": 1.9571,
      "step": 156000
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.274240967548663e-05,
      "loss": 1.9398,
      "step": 156500
    },
    {
      "epoch": 1.64,
      "learning_rate": 2.265532472237317e-05,
      "loss": 1.9584,
      "step": 157000
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.256823976925971e-05,
      "loss": 1.9351,
      "step": 157500
    },
    {
      "epoch": 1.65,
      "learning_rate": 2.2481154816146248e-05,
      "loss": 1.9612,
      "step": 158000
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.2394069863032788e-05,
      "loss": 1.9398,
      "step": 158500
    },
    {
      "epoch": 1.66,
      "learning_rate": 2.2306984909919327e-05,
      "loss": 1.9425,
      "step": 159000
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.2219899956805864e-05,
      "loss": 1.9639,
      "step": 159500
    },
    {
      "epoch": 1.67,
      "learning_rate": 2.2132815003692403e-05,
      "loss": 1.9281,
      "step": 160000
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.2045730050578943e-05,
      "loss": 1.9447,
      "step": 160500
    },
    {
      "epoch": 1.68,
      "learning_rate": 2.195864509746548e-05,
      "loss": 1.9255,
      "step": 161000
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.187156014435202e-05,
      "loss": 1.9457,
      "step": 161500
    },
    {
      "epoch": 1.69,
      "learning_rate": 2.1784475191238558e-05,
      "loss": 1.945,
      "step": 162000
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.1697390238125098e-05,
      "loss": 1.9559,
      "step": 162500
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.1610305285011637e-05,
      "loss": 1.9303,
      "step": 163000
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.1523220331898174e-05,
      "loss": 1.9174,
      "step": 163500
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.1436135378784713e-05,
      "loss": 1.9374,
      "step": 164000
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.1349050425671253e-05,
      "loss": 1.941,
      "step": 164500
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.126196547255779e-05,
      "loss": 1.9235,
      "step": 165000
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.117488051944433e-05,
      "loss": 1.9325,
      "step": 165500
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.1087795566330868e-05,
      "loss": 1.9497,
      "step": 166000
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.1000710613217404e-05,
      "loss": 1.9277,
      "step": 166500
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.0913625660103944e-05,
      "loss": 1.9326,
      "step": 167000
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.0826540706990487e-05,
      "loss": 1.9026,
      "step": 167500
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.0739455753877023e-05,
      "loss": 1.9175,
      "step": 168000
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.0652370800763563e-05,
      "loss": 1.9002,
      "step": 168500
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.0565285847650102e-05,
      "loss": 1.9534,
      "step": 169000
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.047820089453664e-05,
      "loss": 1.9303,
      "step": 169500
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.0391115941423178e-05,
      "loss": 1.9324,
      "step": 170000
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.0304030988309718e-05,
      "loss": 1.9322,
      "step": 170500
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.0216946035196254e-05,
      "loss": 1.9185,
      "step": 171000
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.0129861082082793e-05,
      "loss": 1.9078,
      "step": 171500
    },
    {
      "epoch": 1.8,
      "learning_rate": 2.0042776128969333e-05,
      "loss": 1.8907,
      "step": 172000
    },
    {
      "epoch": 1.8,
      "learning_rate": 1.9955691175855873e-05,
      "loss": 1.8814,
      "step": 172500
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.9868606222742412e-05,
      "loss": 1.9388,
      "step": 173000
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.978152126962895e-05,
      "loss": 1.898,
      "step": 173500
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.9694436316515488e-05,
      "loss": 1.9195,
      "step": 174000
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.9607351363402028e-05,
      "loss": 1.911,
      "step": 174500
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.9520266410288564e-05,
      "loss": 1.8908,
      "step": 175000
    },
    {
      "epoch": 1.83,
      "learning_rate": 1.9433181457175103e-05,
      "loss": 1.8934,
      "step": 175500
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.9346096504061643e-05,
      "loss": 1.8823,
      "step": 176000
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.925901155094818e-05,
      "loss": 1.8952,
      "step": 176500
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.9171926597834722e-05,
      "loss": 1.917,
      "step": 177000
    },
    {
      "epoch": 1.85,
      "learning_rate": 1.9084841644721262e-05,
      "loss": 1.9191,
      "step": 177500
    },
    {
      "epoch": 1.86,
      "learning_rate": 1.8997756691607798e-05,
      "loss": 1.8784,
      "step": 178000
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.8910671738494338e-05,
      "loss": 1.8598,
      "step": 178500
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.8823586785380877e-05,
      "loss": 1.9008,
      "step": 179000
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.8736501832267413e-05,
      "loss": 1.9135,
      "step": 179500
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.8649416879153953e-05,
      "loss": 1.8975,
      "step": 180000
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.8562331926040493e-05,
      "loss": 1.8977,
      "step": 180500
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.847524697292703e-05,
      "loss": 1.9034,
      "step": 181000
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.8388162019813572e-05,
      "loss": 1.8742,
      "step": 181500
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.8301077066700108e-05,
      "loss": 1.9012,
      "step": 182000
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.8213992113586647e-05,
      "loss": 1.8909,
      "step": 182500
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.8126907160473187e-05,
      "loss": 1.8979,
      "step": 183000
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.8039822207359723e-05,
      "loss": 1.8946,
      "step": 183500
    },
    {
      "epoch": 1.92,
      "learning_rate": 1.7952737254246263e-05,
      "loss": 1.9101,
      "step": 184000
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.7865652301132802e-05,
      "loss": 1.8848,
      "step": 184500
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.777856734801934e-05,
      "loss": 1.8875,
      "step": 185000
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.7691482394905878e-05,
      "loss": 1.8915,
      "step": 185500
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.760439744179242e-05,
      "loss": 1.9019,
      "step": 186000
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.7517312488678957e-05,
      "loss": 1.8698,
      "step": 186500
    },
    {
      "epoch": 1.95,
      "learning_rate": 1.7430227535565497e-05,
      "loss": 1.887,
      "step": 187000
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.7343142582452037e-05,
      "loss": 1.8691,
      "step": 187500
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.7256057629338573e-05,
      "loss": 1.8669,
      "step": 188000
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.7168972676225112e-05,
      "loss": 1.9039,
      "step": 188500
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.7081887723111652e-05,
      "loss": 1.893,
      "step": 189000
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.6994802769998188e-05,
      "loss": 1.8628,
      "step": 189500
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.6907717816884728e-05,
      "loss": 1.8672,
      "step": 190000
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.6820632863771267e-05,
      "loss": 1.8714,
      "step": 190500
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.6733547910657807e-05,
      "loss": 1.8499,
      "step": 191000
    },
    {
      "epoch": 2.0,
      "learning_rate": 1.6646462957544347e-05,
      "loss": 1.8405,
      "step": 191500
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.6559378004430883e-05,
      "loss": 1.8392,
      "step": 192000
    },
    {
      "epoch": 2.01,
      "learning_rate": 1.6472293051317422e-05,
      "loss": 1.8528,
      "step": 192500
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.6385208098203962e-05,
      "loss": 1.8385,
      "step": 193000
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.6298123145090498e-05,
      "loss": 1.8326,
      "step": 193500
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.6211038191977038e-05,
      "loss": 1.8424,
      "step": 194000
    },
    {
      "epoch": 2.03,
      "learning_rate": 1.6123953238863577e-05,
      "loss": 1.8147,
      "step": 194500
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.6036868285750114e-05,
      "loss": 1.8671,
      "step": 195000
    },
    {
      "epoch": 2.04,
      "learning_rate": 1.5949783332636656e-05,
      "loss": 1.823,
      "step": 195500
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.5862698379523193e-05,
      "loss": 1.8271,
      "step": 196000
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.5775613426409732e-05,
      "loss": 1.8385,
      "step": 196500
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.5688528473296272e-05,
      "loss": 1.8401,
      "step": 197000
    },
    {
      "epoch": 2.06,
      "learning_rate": 1.5601443520182808e-05,
      "loss": 1.8166,
      "step": 197500
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.5514358567069348e-05,
      "loss": 1.8358,
      "step": 198000
    },
    {
      "epoch": 2.07,
      "learning_rate": 1.5427273613955887e-05,
      "loss": 1.8174,
      "step": 198500
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.5340188660842423e-05,
      "loss": 1.8661,
      "step": 199000
    },
    {
      "epoch": 2.08,
      "learning_rate": 1.5253103707728963e-05,
      "loss": 1.8418,
      "step": 199500
    },
    {
      "epoch": 2.09,
      "learning_rate": 1.5166018754615503e-05,
      "loss": 1.8458,
      "step": 200000
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.5078933801502044e-05,
      "loss": 1.8425,
      "step": 200500
    },
    {
      "epoch": 2.1,
      "learning_rate": 1.4991848848388582e-05,
      "loss": 1.8042,
      "step": 201000
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.490476389527512e-05,
      "loss": 1.811,
      "step": 201500
    },
    {
      "epoch": 2.11,
      "learning_rate": 1.481767894216166e-05,
      "loss": 1.827,
      "step": 202000
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.4730593989048197e-05,
      "loss": 1.8273,
      "step": 202500
    },
    {
      "epoch": 2.12,
      "learning_rate": 1.4643509035934735e-05,
      "loss": 1.8124,
      "step": 203000
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.4556424082821275e-05,
      "loss": 1.8035,
      "step": 203500
    },
    {
      "epoch": 2.13,
      "learning_rate": 1.4469339129707813e-05,
      "loss": 1.8408,
      "step": 204000
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.438225417659435e-05,
      "loss": 1.7842,
      "step": 204500
    },
    {
      "epoch": 2.14,
      "learning_rate": 1.4295169223480892e-05,
      "loss": 1.8058,
      "step": 205000
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.4208084270367431e-05,
      "loss": 1.8098,
      "step": 205500
    },
    {
      "epoch": 2.15,
      "learning_rate": 1.412099931725397e-05,
      "loss": 1.8061,
      "step": 206000
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.4033914364140507e-05,
      "loss": 1.8077,
      "step": 206500
    },
    {
      "epoch": 2.16,
      "learning_rate": 1.3946829411027047e-05,
      "loss": 1.8174,
      "step": 207000
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.3859744457913585e-05,
      "loss": 1.7933,
      "step": 207500
    },
    {
      "epoch": 2.17,
      "learning_rate": 1.3772659504800122e-05,
      "loss": 1.7864,
      "step": 208000
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.3685574551686662e-05,
      "loss": 1.8072,
      "step": 208500
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.35984895985732e-05,
      "loss": 1.8226,
      "step": 209000
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.3511404645459738e-05,
      "loss": 1.8134,
      "step": 209500
    },
    {
      "epoch": 2.19,
      "learning_rate": 1.342431969234628e-05,
      "loss": 1.8193,
      "step": 210000
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.3337234739232817e-05,
      "loss": 1.8274,
      "step": 210500
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.3250149786119357e-05,
      "loss": 1.8083,
      "step": 211000
    },
    {
      "epoch": 2.21,
      "learning_rate": 1.3163064833005895e-05,
      "loss": 1.8299,
      "step": 211500
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.3075979879892432e-05,
      "loss": 1.7945,
      "step": 212000
    },
    {
      "epoch": 2.22,
      "learning_rate": 1.2988894926778972e-05,
      "loss": 1.8197,
      "step": 212500
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.290180997366551e-05,
      "loss": 1.801,
      "step": 213000
    },
    {
      "epoch": 2.23,
      "learning_rate": 1.2814725020552048e-05,
      "loss": 1.7869,
      "step": 213500
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.2727640067438587e-05,
      "loss": 1.787,
      "step": 214000
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.2640555114325129e-05,
      "loss": 1.795,
      "step": 214500
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.2553470161211667e-05,
      "loss": 1.8039,
      "step": 215000
    },
    {
      "epoch": 2.25,
      "learning_rate": 1.2466385208098204e-05,
      "loss": 1.7862,
      "step": 215500
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.2379300254984744e-05,
      "loss": 1.7948,
      "step": 216000
    },
    {
      "epoch": 2.26,
      "learning_rate": 1.2292215301871282e-05,
      "loss": 1.7791,
      "step": 216500
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.220513034875782e-05,
      "loss": 1.7978,
      "step": 217000
    },
    {
      "epoch": 2.27,
      "learning_rate": 1.211804539564436e-05,
      "loss": 1.8121,
      "step": 217500
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.2030960442530899e-05,
      "loss": 1.8106,
      "step": 218000
    },
    {
      "epoch": 2.28,
      "learning_rate": 1.1943875489417437e-05,
      "loss": 1.812,
      "step": 218500
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.1856790536303977e-05,
      "loss": 1.7548,
      "step": 219000
    },
    {
      "epoch": 2.29,
      "learning_rate": 1.1769705583190514e-05,
      "loss": 1.7806,
      "step": 219500
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.1682620630077052e-05,
      "loss": 1.7942,
      "step": 220000
    },
    {
      "epoch": 2.3,
      "learning_rate": 1.1595535676963592e-05,
      "loss": 1.8271,
      "step": 220500
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.1508450723850131e-05,
      "loss": 1.765,
      "step": 221000
    },
    {
      "epoch": 2.31,
      "learning_rate": 1.142136577073667e-05,
      "loss": 1.8056,
      "step": 221500
    },
    {
      "epoch": 2.32,
      "learning_rate": 1.1334280817623207e-05,
      "loss": 1.7853,
      "step": 222000
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.1247195864509749e-05,
      "loss": 1.7754,
      "step": 222500
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.1160110911396286e-05,
      "loss": 1.7931,
      "step": 223000
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.1073025958282824e-05,
      "loss": 1.7868,
      "step": 223500
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.0985941005169364e-05,
      "loss": 1.7744,
      "step": 224000
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.0898856052055902e-05,
      "loss": 1.7919,
      "step": 224500
    },
    {
      "epoch": 2.35,
      "learning_rate": 1.0811771098942441e-05,
      "loss": 1.7822,
      "step": 225000
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.072468614582898e-05,
      "loss": 1.7738,
      "step": 225500
    },
    {
      "epoch": 2.36,
      "learning_rate": 1.0637601192715519e-05,
      "loss": 1.779,
      "step": 226000
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0550516239602057e-05,
      "loss": 1.7856,
      "step": 226500
    },
    {
      "epoch": 2.37,
      "learning_rate": 1.0463431286488595e-05,
      "loss": 1.8012,
      "step": 227000
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0376346333375134e-05,
      "loss": 1.7557,
      "step": 227500
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.0289261380261674e-05,
      "loss": 1.7674,
      "step": 228000
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.0202176427148212e-05,
      "loss": 1.7739,
      "step": 228500
    },
    {
      "epoch": 2.39,
      "learning_rate": 1.011509147403475e-05,
      "loss": 1.7393,
      "step": 229000
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.002800652092129e-05,
      "loss": 1.7869,
      "step": 229500
    },
    {
      "epoch": 2.4,
      "learning_rate": 9.940921567807829e-06,
      "loss": 1.7764,
      "step": 230000
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.853836614694367e-06,
      "loss": 1.7298,
      "step": 230500
    },
    {
      "epoch": 2.41,
      "learning_rate": 9.766751661580906e-06,
      "loss": 1.7476,
      "step": 231000
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.679666708467444e-06,
      "loss": 1.7576,
      "step": 231500
    },
    {
      "epoch": 2.42,
      "learning_rate": 9.592581755353984e-06,
      "loss": 1.7482,
      "step": 232000
    },
    {
      "epoch": 2.43,
      "learning_rate": 9.505496802240522e-06,
      "loss": 1.7533,
      "step": 232500
    },
    {
      "epoch": 2.43,
      "learning_rate": 9.418411849127061e-06,
      "loss": 1.7642,
      "step": 233000
    },
    {
      "epoch": 2.44,
      "learning_rate": 9.3313268960136e-06,
      "loss": 1.7546,
      "step": 233500
    },
    {
      "epoch": 2.45,
      "learning_rate": 9.244241942900137e-06,
      "loss": 1.7569,
      "step": 234000
    },
    {
      "epoch": 2.45,
      "learning_rate": 9.157156989786678e-06,
      "loss": 1.7453,
      "step": 234500
    },
    {
      "epoch": 2.46,
      "learning_rate": 9.070072036673216e-06,
      "loss": 1.7449,
      "step": 235000
    },
    {
      "epoch": 2.46,
      "learning_rate": 8.982987083559754e-06,
      "loss": 1.7682,
      "step": 235500
    },
    {
      "epoch": 2.47,
      "learning_rate": 8.895902130446294e-06,
      "loss": 1.7584,
      "step": 236000
    },
    {
      "epoch": 2.47,
      "learning_rate": 8.808817177332832e-06,
      "loss": 1.7557,
      "step": 236500
    },
    {
      "epoch": 2.48,
      "learning_rate": 8.721732224219371e-06,
      "loss": 1.773,
      "step": 237000
    },
    {
      "epoch": 2.48,
      "learning_rate": 8.634647271105909e-06,
      "loss": 1.773,
      "step": 237500
    },
    {
      "epoch": 2.49,
      "learning_rate": 8.547562317992449e-06,
      "loss": 1.744,
      "step": 238000
    },
    {
      "epoch": 2.49,
      "learning_rate": 8.460477364878987e-06,
      "loss": 1.7379,
      "step": 238500
    },
    {
      "epoch": 2.5,
      "learning_rate": 8.373392411765526e-06,
      "loss": 1.7635,
      "step": 239000
    },
    {
      "epoch": 2.5,
      "learning_rate": 8.286307458652066e-06,
      "loss": 1.7585,
      "step": 239500
    },
    {
      "epoch": 2.51,
      "learning_rate": 8.199222505538604e-06,
      "loss": 1.7488,
      "step": 240000
    },
    {
      "epoch": 2.51,
      "learning_rate": 8.112137552425142e-06,
      "loss": 1.7532,
      "step": 240500
    },
    {
      "epoch": 2.52,
      "learning_rate": 8.025052599311681e-06,
      "loss": 1.758,
      "step": 241000
    },
    {
      "epoch": 2.52,
      "learning_rate": 7.93796764619822e-06,
      "loss": 1.7485,
      "step": 241500
    },
    {
      "epoch": 2.53,
      "learning_rate": 7.850882693084759e-06,
      "loss": 1.7409,
      "step": 242000
    },
    {
      "epoch": 2.53,
      "learning_rate": 7.763797739971297e-06,
      "loss": 1.7494,
      "step": 242500
    },
    {
      "epoch": 2.54,
      "learning_rate": 7.676712786857836e-06,
      "loss": 1.7264,
      "step": 243000
    },
    {
      "epoch": 2.54,
      "learning_rate": 7.589627833744374e-06,
      "loss": 1.7465,
      "step": 243500
    },
    {
      "epoch": 2.55,
      "learning_rate": 7.502542880630914e-06,
      "loss": 1.7654,
      "step": 244000
    },
    {
      "epoch": 2.56,
      "learning_rate": 7.415457927517452e-06,
      "loss": 1.7422,
      "step": 244500
    },
    {
      "epoch": 2.56,
      "learning_rate": 7.328372974403991e-06,
      "loss": 1.7418,
      "step": 245000
    },
    {
      "epoch": 2.57,
      "learning_rate": 7.241288021290529e-06,
      "loss": 1.7664,
      "step": 245500
    },
    {
      "epoch": 2.57,
      "learning_rate": 7.154203068177068e-06,
      "loss": 1.7324,
      "step": 246000
    },
    {
      "epoch": 2.58,
      "learning_rate": 7.067118115063607e-06,
      "loss": 1.7143,
      "step": 246500
    },
    {
      "epoch": 2.58,
      "learning_rate": 6.980033161950146e-06,
      "loss": 1.6917,
      "step": 247000
    },
    {
      "epoch": 2.59,
      "learning_rate": 6.892948208836685e-06,
      "loss": 1.7498,
      "step": 247500
    },
    {
      "epoch": 2.59,
      "learning_rate": 6.805863255723223e-06,
      "loss": 1.7161,
      "step": 248000
    },
    {
      "epoch": 2.6,
      "learning_rate": 6.718778302609763e-06,
      "loss": 1.7315,
      "step": 248500
    },
    {
      "epoch": 2.6,
      "learning_rate": 6.631693349496301e-06,
      "loss": 1.7591,
      "step": 249000
    },
    {
      "epoch": 2.61,
      "learning_rate": 6.54460839638284e-06,
      "loss": 1.7388,
      "step": 249500
    },
    {
      "epoch": 2.61,
      "learning_rate": 6.4575234432693785e-06,
      "loss": 1.7236,
      "step": 250000
    },
    {
      "epoch": 2.62,
      "learning_rate": 6.370438490155916e-06,
      "loss": 1.7396,
      "step": 250500
    },
    {
      "epoch": 2.62,
      "learning_rate": 6.283353537042457e-06,
      "loss": 1.7287,
      "step": 251000
    },
    {
      "epoch": 2.63,
      "learning_rate": 6.196268583928995e-06,
      "loss": 1.7231,
      "step": 251500
    },
    {
      "epoch": 2.63,
      "learning_rate": 6.1091836308155335e-06,
      "loss": 1.723,
      "step": 252000
    },
    {
      "epoch": 2.64,
      "learning_rate": 6.022098677702072e-06,
      "loss": 1.7419,
      "step": 252500
    },
    {
      "epoch": 2.64,
      "learning_rate": 5.935013724588611e-06,
      "loss": 1.7434,
      "step": 253000
    },
    {
      "epoch": 2.65,
      "learning_rate": 5.84792877147515e-06,
      "loss": 1.7221,
      "step": 253500
    },
    {
      "epoch": 2.65,
      "learning_rate": 5.7608438183616885e-06,
      "loss": 1.7131,
      "step": 254000
    },
    {
      "epoch": 2.66,
      "learning_rate": 5.673758865248227e-06,
      "loss": 1.7402,
      "step": 254500
    },
    {
      "epoch": 2.66,
      "learning_rate": 5.586673912134766e-06,
      "loss": 1.7157,
      "step": 255000
    },
    {
      "epoch": 2.67,
      "learning_rate": 5.499588959021305e-06,
      "loss": 1.7488,
      "step": 255500
    },
    {
      "epoch": 2.68,
      "learning_rate": 5.4125040059078434e-06,
      "loss": 1.7135,
      "step": 256000
    },
    {
      "epoch": 2.68,
      "learning_rate": 5.325419052794382e-06,
      "loss": 1.7125,
      "step": 256500
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.238334099680921e-06,
      "loss": 1.713,
      "step": 257000
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.15124914656746e-06,
      "loss": 1.7231,
      "step": 257500
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.064164193453998e-06,
      "loss": 1.724,
      "step": 258000
    },
    {
      "epoch": 2.7,
      "learning_rate": 4.977079240340537e-06,
      "loss": 1.7196,
      "step": 258500
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.889994287227076e-06,
      "loss": 1.7282,
      "step": 259000
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.802909334113615e-06,
      "loss": 1.7245,
      "step": 259500
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.715824381000153e-06,
      "loss": 1.7129,
      "step": 260000
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.628739427886692e-06,
      "loss": 1.7114,
      "step": 260500
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.541654474773231e-06,
      "loss": 1.72,
      "step": 261000
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.45456952165977e-06,
      "loss": 1.7321,
      "step": 261500
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.367484568546309e-06,
      "loss": 1.7439,
      "step": 262000
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.280399615432847e-06,
      "loss": 1.6934,
      "step": 262500
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.193314662319386e-06,
      "loss": 1.7236,
      "step": 263000
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.1062297092059246e-06,
      "loss": 1.7333,
      "step": 263500
    },
    {
      "epoch": 2.76,
      "learning_rate": 4.019144756092463e-06,
      "loss": 1.7389,
      "step": 264000
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.932059802979003e-06,
      "loss": 1.6828,
      "step": 264500
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.844974849865541e-06,
      "loss": 1.7122,
      "step": 265000
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.75788989675208e-06,
      "loss": 1.7356,
      "step": 265500
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.6708049436386187e-06,
      "loss": 1.7251,
      "step": 266000
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.583719990525157e-06,
      "loss": 1.7069,
      "step": 266500
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.496635037411696e-06,
      "loss": 1.6999,
      "step": 267000
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.4095500842982345e-06,
      "loss": 1.6997,
      "step": 267500
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.3224651311847737e-06,
      "loss": 1.6997,
      "step": 268000
    },
    {
      "epoch": 2.81,
      "learning_rate": 3.2353801780713124e-06,
      "loss": 1.7099,
      "step": 268500
    },
    {
      "epoch": 2.81,
      "learning_rate": 3.1482952249578516e-06,
      "loss": 1.7147,
      "step": 269000
    },
    {
      "epoch": 2.82,
      "learning_rate": 3.06121027184439e-06,
      "loss": 1.7139,
      "step": 269500
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.9741253187309286e-06,
      "loss": 1.7218,
      "step": 270000
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.8870403656174674e-06,
      "loss": 1.7158,
      "step": 270500
    },
    {
      "epoch": 2.83,
      "learning_rate": 2.7999554125040057e-06,
      "loss": 1.6819,
      "step": 271000
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.712870459390545e-06,
      "loss": 1.7058,
      "step": 271500
    },
    {
      "epoch": 2.84,
      "learning_rate": 2.6257855062770836e-06,
      "loss": 1.708,
      "step": 272000
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.5387005531636223e-06,
      "loss": 1.7167,
      "step": 272500
    },
    {
      "epoch": 2.85,
      "learning_rate": 2.451615600050161e-06,
      "loss": 1.7075,
      "step": 273000
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.3645306469367e-06,
      "loss": 1.7125,
      "step": 273500
    },
    {
      "epoch": 2.86,
      "learning_rate": 2.2774456938232386e-06,
      "loss": 1.7116,
      "step": 274000
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.1903607407097773e-06,
      "loss": 1.6871,
      "step": 274500
    },
    {
      "epoch": 2.87,
      "learning_rate": 2.103275787596316e-06,
      "loss": 1.6659,
      "step": 275000
    },
    {
      "epoch": 2.88,
      "learning_rate": 2.016190834482855e-06,
      "loss": 1.6994,
      "step": 275500
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.9291058813693935e-06,
      "loss": 1.6782,
      "step": 276000
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.8420209282559325e-06,
      "loss": 1.6992,
      "step": 276500
    },
    {
      "epoch": 2.89,
      "learning_rate": 1.7549359751424712e-06,
      "loss": 1.6839,
      "step": 277000
    },
    {
      "epoch": 2.9,
      "learning_rate": 1.6678510220290096e-06,
      "loss": 1.7042,
      "step": 277500
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.5807660689155485e-06,
      "loss": 1.6993,
      "step": 278000
    },
    {
      "epoch": 2.91,
      "learning_rate": 1.4936811158020872e-06,
      "loss": 1.7004,
      "step": 278500
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.406596162688626e-06,
      "loss": 1.6947,
      "step": 279000
    },
    {
      "epoch": 2.92,
      "learning_rate": 1.319511209575165e-06,
      "loss": 1.7011,
      "step": 279500
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.2324262564617035e-06,
      "loss": 1.6973,
      "step": 280000
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.1453413033482422e-06,
      "loss": 1.6965,
      "step": 280500
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.0582563502347812e-06,
      "loss": 1.687,
      "step": 281000
    },
    {
      "epoch": 2.94,
      "learning_rate": 9.7117139712132e-07,
      "loss": 1.6907,
      "step": 281500
    },
    {
      "epoch": 2.95,
      "learning_rate": 8.840864440078585e-07,
      "loss": 1.6931,
      "step": 282000
    },
    {
      "epoch": 2.95,
      "learning_rate": 7.970014908943974e-07,
      "loss": 1.7063,
      "step": 282500
    },
    {
      "epoch": 2.96,
      "learning_rate": 7.09916537780936e-07,
      "loss": 1.7135,
      "step": 283000
    },
    {
      "epoch": 2.96,
      "learning_rate": 6.228315846674749e-07,
      "loss": 1.689,
      "step": 283500
    },
    {
      "epoch": 2.97,
      "learning_rate": 5.357466315540136e-07,
      "loss": 1.6676,
      "step": 284000
    },
    {
      "epoch": 2.97,
      "learning_rate": 4.486616784405523e-07,
      "loss": 1.7108,
      "step": 284500
    },
    {
      "epoch": 2.98,
      "learning_rate": 3.615767253270911e-07,
      "loss": 1.6767,
      "step": 285000
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.7449177221362985e-07,
      "loss": 1.6986,
      "step": 285500
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.8740681910016862e-07,
      "loss": 1.6615,
      "step": 286000
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.0032186598670735e-07,
      "loss": 1.6876,
      "step": 286500
    },
    {
      "epoch": 3.0,
      "learning_rate": 1.323691287324611e-08,
      "loss": 1.7034,
      "step": 287000
    },
    {
      "epoch": 3.0,
      "step": 287076,
      "total_flos": 4.9083067456945856e+17,
      "train_loss": 2.060935147813305,
      "train_runtime": 51665.6711,
      "train_samples_per_second": 44.451,
      "train_steps_per_second": 5.556
    }
  ],
  "logging_steps": 500,
  "max_steps": 287076,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 4.9083067456945856e+17,
  "trial_name": null,
  "trial_params": null
}