{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.944640753828033,
  "global_step": 160000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 4.995398998822144e-05,
      "loss": 0.2082,
      "step": 500
    },
    {
      "epoch": 0.02,
      "learning_rate": 4.990797997644288e-05,
      "loss": 0.1548,
      "step": 1000
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.9861969964664316e-05,
      "loss": 0.1389,
      "step": 1500
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.9815959952885747e-05,
      "loss": 0.1344,
      "step": 2000
    },
    {
      "epoch": 0.05,
      "learning_rate": 4.9769949941107184e-05,
      "loss": 0.1306,
      "step": 2500
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.972393992932862e-05,
      "loss": 0.1204,
      "step": 3000
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.967792991755006e-05,
      "loss": 0.1265,
      "step": 3500
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.96319199057715e-05,
      "loss": 0.1253,
      "step": 4000
    },
    {
      "epoch": 0.08,
      "learning_rate": 4.9585909893992935e-05,
      "loss": 0.1151,
      "step": 4500
    },
    {
      "epoch": 0.09,
      "learning_rate": 4.953989988221437e-05,
      "loss": 0.1133,
      "step": 5000
    },
    {
      "epoch": 0.1,
      "learning_rate": 4.949388987043581e-05,
      "loss": 0.1193,
      "step": 5500
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.944787985865725e-05,
      "loss": 0.1181,
      "step": 6000
    },
    {
      "epoch": 0.12,
      "learning_rate": 4.9401869846878686e-05,
      "loss": 0.1124,
      "step": 6500
    },
    {
      "epoch": 0.13,
      "learning_rate": 4.9355859835100124e-05,
      "loss": 0.1117,
      "step": 7000
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.9309849823321555e-05,
      "loss": 0.1117,
      "step": 7500
    },
    {
      "epoch": 0.15,
      "learning_rate": 4.926383981154299e-05,
      "loss": 0.1064,
      "step": 8000
    },
    {
      "epoch": 0.16,
      "learning_rate": 4.921782979976443e-05,
      "loss": 0.1073,
      "step": 8500
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.917181978798587e-05,
      "loss": 0.1064,
      "step": 9000
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.9125809776207306e-05,
      "loss": 0.103,
      "step": 9500
    },
    {
      "epoch": 0.18,
      "learning_rate": 4.9079799764428744e-05,
      "loss": 0.1039,
      "step": 10000
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.9033789752650175e-05,
      "loss": 0.0993,
      "step": 10500
    },
    {
      "epoch": 0.2,
      "learning_rate": 4.898777974087161e-05,
      "loss": 0.0978,
      "step": 11000
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.894176972909305e-05,
      "loss": 0.1037,
      "step": 11500
    },
    {
      "epoch": 0.22,
      "learning_rate": 4.889575971731449e-05,
      "loss": 0.1021,
      "step": 12000
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.8849749705535926e-05,
      "loss": 0.104,
      "step": 12500
    },
    {
      "epoch": 0.24,
      "learning_rate": 4.880373969375736e-05,
      "loss": 0.1,
      "step": 13000
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.87577296819788e-05,
      "loss": 0.1069,
      "step": 13500
    },
    {
      "epoch": 0.26,
      "learning_rate": 4.871171967020024e-05,
      "loss": 0.1063,
      "step": 14000
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.8665709658421677e-05,
      "loss": 0.1061,
      "step": 14500
    },
    {
      "epoch": 0.28,
      "learning_rate": 4.8619699646643114e-05,
      "loss": 0.1037,
      "step": 15000
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.857368963486455e-05,
      "loss": 0.0995,
      "step": 15500
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.852767962308599e-05,
      "loss": 0.0999,
      "step": 16000
    },
    {
      "epoch": 0.3,
      "learning_rate": 4.848166961130742e-05,
      "loss": 0.0985,
      "step": 16500
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.843565959952886e-05,
      "loss": 0.0925,
      "step": 17000
    },
    {
      "epoch": 0.32,
      "learning_rate": 4.8389649587750296e-05,
      "loss": 0.0996,
      "step": 17500
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.8343639575971734e-05,
      "loss": 0.1022,
      "step": 18000
    },
    {
      "epoch": 0.34,
      "learning_rate": 4.829762956419317e-05,
      "loss": 0.0979,
      "step": 18500
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.825161955241461e-05,
      "loss": 0.0969,
      "step": 19000
    },
    {
      "epoch": 0.36,
      "learning_rate": 4.820560954063604e-05,
      "loss": 0.109,
      "step": 19500
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.815959952885748e-05,
      "loss": 0.1,
      "step": 20000
    },
    {
      "epoch": 0.38,
      "learning_rate": 4.8113589517078916e-05,
      "loss": 0.0998,
      "step": 20500
    },
    {
      "epoch": 0.39,
      "learning_rate": 4.8067579505300354e-05,
      "loss": 0.0954,
      "step": 21000
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.80215694935218e-05,
      "loss": 0.1004,
      "step": 21500
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.797555948174323e-05,
      "loss": 0.097,
      "step": 22000
    },
    {
      "epoch": 0.41,
      "learning_rate": 4.792954946996467e-05,
      "loss": 0.0972,
      "step": 22500
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.7883539458186105e-05,
      "loss": 0.0958,
      "step": 23000
    },
    {
      "epoch": 0.43,
      "learning_rate": 4.783752944640754e-05,
      "loss": 0.0981,
      "step": 23500
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.779151943462898e-05,
      "loss": 0.0962,
      "step": 24000
    },
    {
      "epoch": 0.45,
      "learning_rate": 4.774550942285042e-05,
      "loss": 0.0994,
      "step": 24500
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.769949941107185e-05,
      "loss": 0.0958,
      "step": 25000
    },
    {
      "epoch": 0.47,
      "learning_rate": 4.7653489399293287e-05,
      "loss": 0.0921,
      "step": 25500
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.7607479387514724e-05,
      "loss": 0.0985,
      "step": 26000
    },
    {
      "epoch": 0.49,
      "learning_rate": 4.756146937573616e-05,
      "loss": 0.0957,
      "step": 26500
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.75154593639576e-05,
      "loss": 0.0962,
      "step": 27000
    },
    {
      "epoch": 0.51,
      "learning_rate": 4.746944935217904e-05,
      "loss": 0.1003,
      "step": 27500
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.742343934040047e-05,
      "loss": 0.1016,
      "step": 28000
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.7377429328621906e-05,
      "loss": 0.0943,
      "step": 28500
    },
    {
      "epoch": 0.53,
      "learning_rate": 4.7331419316843344e-05,
      "loss": 0.0997,
      "step": 29000
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.728540930506478e-05,
      "loss": 0.0924,
      "step": 29500
    },
    {
      "epoch": 0.55,
      "learning_rate": 4.7239399293286226e-05,
      "loss": 0.0959,
      "step": 30000
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.719338928150766e-05,
      "loss": 0.0915,
      "step": 30500
    },
    {
      "epoch": 0.57,
      "learning_rate": 4.7147379269729095e-05,
      "loss": 0.0932,
      "step": 31000
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.710136925795053e-05,
      "loss": 0.0964,
      "step": 31500
    },
    {
      "epoch": 0.59,
      "learning_rate": 4.705535924617197e-05,
      "loss": 0.1011,
      "step": 32000
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.700934923439341e-05,
      "loss": 0.091,
      "step": 32500
    },
    {
      "epoch": 0.61,
      "learning_rate": 4.6963339222614846e-05,
      "loss": 0.0986,
      "step": 33000
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.6917329210836284e-05,
      "loss": 0.0911,
      "step": 33500
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.6871319199057715e-05,
      "loss": 0.0857,
      "step": 34000
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.682530918727915e-05,
      "loss": 0.0983,
      "step": 34500
    },
    {
      "epoch": 0.64,
      "learning_rate": 4.677929917550059e-05,
      "loss": 0.0974,
      "step": 35000
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.673328916372203e-05,
      "loss": 0.0911,
      "step": 35500
    },
    {
      "epoch": 0.66,
      "learning_rate": 4.6687279151943466e-05,
      "loss": 0.0945,
      "step": 36000
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.66412691401649e-05,
      "loss": 0.0932,
      "step": 36500
    },
    {
      "epoch": 0.68,
      "learning_rate": 4.6595259128386334e-05,
      "loss": 0.0998,
      "step": 37000
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.654924911660777e-05,
      "loss": 0.0896,
      "step": 37500
    },
    {
      "epoch": 0.7,
      "learning_rate": 4.650323910482921e-05,
      "loss": 0.0893,
      "step": 38000
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.6457229093050654e-05,
      "loss": 0.0984,
      "step": 38500
    },
    {
      "epoch": 0.72,
      "learning_rate": 4.641121908127209e-05,
      "loss": 0.0867,
      "step": 39000
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.636520906949352e-05,
      "loss": 0.0944,
      "step": 39500
    },
    {
      "epoch": 0.74,
      "learning_rate": 4.631919905771496e-05,
      "loss": 0.0918,
      "step": 40000
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.62731890459364e-05,
      "loss": 0.0884,
      "step": 40500
    },
    {
      "epoch": 0.75,
      "learning_rate": 4.6227179034157836e-05,
      "loss": 0.0927,
      "step": 41000
    },
    {
      "epoch": 0.76,
      "learning_rate": 4.6181169022379274e-05,
      "loss": 0.0927,
      "step": 41500
    },
    {
      "epoch": 0.77,
      "learning_rate": 4.613515901060071e-05,
      "loss": 0.0978,
      "step": 42000
    },
    {
      "epoch": 0.78,
      "learning_rate": 4.608914899882214e-05,
      "loss": 0.0944,
      "step": 42500
    },
    {
      "epoch": 0.79,
      "learning_rate": 4.604313898704358e-05,
      "loss": 0.0918,
      "step": 43000
    },
    {
      "epoch": 0.8,
      "learning_rate": 4.599712897526502e-05,
      "loss": 0.0886,
      "step": 43500
    },
    {
      "epoch": 0.81,
      "learning_rate": 4.5951118963486456e-05,
      "loss": 0.0897,
      "step": 44000
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.5905108951707894e-05,
      "loss": 0.0894,
      "step": 44500
    },
    {
      "epoch": 0.83,
      "learning_rate": 4.585909893992933e-05,
      "loss": 0.09,
      "step": 45000
    },
    {
      "epoch": 0.84,
      "learning_rate": 4.581308892815077e-05,
      "loss": 0.0934,
      "step": 45500
    },
    {
      "epoch": 0.85,
      "learning_rate": 4.57670789163722e-05,
      "loss": 0.0874,
      "step": 46000
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.572106890459364e-05,
      "loss": 0.0929,
      "step": 46500
    },
    {
      "epoch": 0.86,
      "learning_rate": 4.567505889281508e-05,
      "loss": 0.0883,
      "step": 47000
    },
    {
      "epoch": 0.87,
      "learning_rate": 4.562904888103652e-05,
      "loss": 0.0929,
      "step": 47500
    },
    {
      "epoch": 0.88,
      "learning_rate": 4.558303886925796e-05,
      "loss": 0.0835,
      "step": 48000
    },
    {
      "epoch": 0.89,
      "learning_rate": 4.553702885747939e-05,
      "loss": 0.0925,
      "step": 48500
    },
    {
      "epoch": 0.9,
      "learning_rate": 4.5491018845700827e-05,
      "loss": 0.0934,
      "step": 49000
    },
    {
      "epoch": 0.91,
      "learning_rate": 4.5445008833922264e-05,
      "loss": 0.095,
      "step": 49500
    },
    {
      "epoch": 0.92,
      "learning_rate": 4.53989988221437e-05,
      "loss": 0.0861,
      "step": 50000
    },
    {
      "epoch": 0.93,
      "learning_rate": 4.535298881036514e-05,
      "loss": 0.089,
      "step": 50500
    },
    {
      "epoch": 0.94,
      "learning_rate": 4.530697879858658e-05,
      "loss": 0.0878,
      "step": 51000
    },
    {
      "epoch": 0.95,
      "learning_rate": 4.526096878680801e-05,
      "loss": 0.094,
      "step": 51500
    },
    {
      "epoch": 0.96,
      "learning_rate": 4.5214958775029446e-05,
      "loss": 0.0888,
      "step": 52000
    },
    {
      "epoch": 0.97,
      "learning_rate": 4.5168948763250884e-05,
      "loss": 0.0802,
      "step": 52500
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.512293875147232e-05,
      "loss": 0.0945,
      "step": 53000
    },
    {
      "epoch": 0.98,
      "learning_rate": 4.507692873969376e-05,
      "loss": 0.0909,
      "step": 53500
    },
    {
      "epoch": 0.99,
      "learning_rate": 4.50309187279152e-05,
      "loss": 0.0863,
      "step": 54000
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9518667922735654,
      "eval_loss": 0.08465953916311264,
      "eval_runtime": 917.0055,
      "eval_samples_per_second": 270.875,
      "eval_steps_per_second": 16.93,
      "step": 54336
    },
    {
      "epoch": 1.0,
      "learning_rate": 4.498490871613663e-05,
      "loss": 0.0784,
      "step": 54500
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.4938898704358066e-05,
      "loss": 0.0789,
      "step": 55000
    },
    {
      "epoch": 1.02,
      "learning_rate": 4.489288869257951e-05,
      "loss": 0.0799,
      "step": 55500
    },
    {
      "epoch": 1.03,
      "learning_rate": 4.484687868080095e-05,
      "loss": 0.0813,
      "step": 56000
    },
    {
      "epoch": 1.04,
      "learning_rate": 4.4800868669022386e-05,
      "loss": 0.0784,
      "step": 56500
    },
    {
      "epoch": 1.05,
      "learning_rate": 4.475485865724382e-05,
      "loss": 0.081,
      "step": 57000
    },
    {
      "epoch": 1.06,
      "learning_rate": 4.4708848645465255e-05,
      "loss": 0.0781,
      "step": 57500
    },
    {
      "epoch": 1.07,
      "learning_rate": 4.466283863368669e-05,
      "loss": 0.0764,
      "step": 58000
    },
    {
      "epoch": 1.08,
      "learning_rate": 4.461682862190813e-05,
      "loss": 0.0765,
      "step": 58500
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.457081861012957e-05,
      "loss": 0.0798,
      "step": 59000
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.4524808598351006e-05,
      "loss": 0.0805,
      "step": 59500
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.4478798586572437e-05,
      "loss": 0.0795,
      "step": 60000
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.4432788574793874e-05,
      "loss": 0.0802,
      "step": 60500
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.438677856301531e-05,
      "loss": 0.0761,
      "step": 61000
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.434076855123675e-05,
      "loss": 0.0847,
      "step": 61500
    },
    {
      "epoch": 1.14,
      "learning_rate": 4.429475853945819e-05,
      "loss": 0.0816,
      "step": 62000
    },
    {
      "epoch": 1.15,
      "learning_rate": 4.4248748527679625e-05,
      "loss": 0.0886,
      "step": 62500
    },
    {
      "epoch": 1.16,
      "learning_rate": 4.420273851590106e-05,
      "loss": 0.0796,
      "step": 63000
    },
    {
      "epoch": 1.17,
      "learning_rate": 4.4156728504122494e-05,
      "loss": 0.0809,
      "step": 63500
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.411071849234394e-05,
      "loss": 0.0852,
      "step": 64000
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.4064708480565376e-05,
      "loss": 0.0798,
      "step": 64500
    },
    {
      "epoch": 1.2,
      "learning_rate": 4.4018698468786814e-05,
      "loss": 0.0792,
      "step": 65000
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.397268845700825e-05,
      "loss": 0.0814,
      "step": 65500
    },
    {
      "epoch": 1.21,
      "learning_rate": 4.392667844522968e-05,
      "loss": 0.0865,
      "step": 66000
    },
    {
      "epoch": 1.22,
      "learning_rate": 4.388066843345112e-05,
      "loss": 0.0796,
      "step": 66500
    },
    {
      "epoch": 1.23,
      "learning_rate": 4.383465842167256e-05,
      "loss": 0.0813,
      "step": 67000
    },
    {
      "epoch": 1.24,
      "learning_rate": 4.3788648409893996e-05,
      "loss": 0.0778,
      "step": 67500
    },
    {
      "epoch": 1.25,
      "learning_rate": 4.3742638398115434e-05,
      "loss": 0.0754,
      "step": 68000
    },
    {
      "epoch": 1.26,
      "learning_rate": 4.369662838633687e-05,
      "loss": 0.0824,
      "step": 68500
    },
    {
      "epoch": 1.27,
      "learning_rate": 4.36506183745583e-05,
      "loss": 0.0804,
      "step": 69000
    },
    {
      "epoch": 1.28,
      "learning_rate": 4.360460836277974e-05,
      "loss": 0.082,
      "step": 69500
    },
    {
      "epoch": 1.29,
      "learning_rate": 4.355859835100118e-05,
      "loss": 0.0823,
      "step": 70000
    },
    {
      "epoch": 1.3,
      "learning_rate": 4.3512588339222616e-05,
      "loss": 0.0795,
      "step": 70500
    },
    {
      "epoch": 1.31,
      "learning_rate": 4.346657832744405e-05,
      "loss": 0.0811,
      "step": 71000
    },
    {
      "epoch": 1.32,
      "learning_rate": 4.342056831566549e-05,
      "loss": 0.0838,
      "step": 71500
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.337455830388692e-05,
      "loss": 0.0827,
      "step": 72000
    },
    {
      "epoch": 1.33,
      "learning_rate": 4.3328548292108367e-05,
      "loss": 0.0824,
      "step": 72500
    },
    {
      "epoch": 1.34,
      "learning_rate": 4.3282538280329804e-05,
      "loss": 0.0816,
      "step": 73000
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.323652826855124e-05,
      "loss": 0.0796,
      "step": 73500
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.319051825677268e-05,
      "loss": 0.0784,
      "step": 74000
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.314450824499411e-05,
      "loss": 0.076,
      "step": 74500
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.309849823321555e-05,
      "loss": 0.079,
      "step": 75000
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.3052488221436986e-05,
      "loss": 0.0787,
      "step": 75500
    },
    {
      "epoch": 1.4,
      "learning_rate": 4.3006478209658424e-05,
      "loss": 0.0759,
      "step": 76000
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.296046819787986e-05,
      "loss": 0.0797,
      "step": 76500
    },
    {
      "epoch": 1.42,
      "learning_rate": 4.29144581861013e-05,
      "loss": 0.0802,
      "step": 77000
    },
    {
      "epoch": 1.43,
      "learning_rate": 4.286844817432274e-05,
      "loss": 0.0839,
      "step": 77500
    },
    {
      "epoch": 1.44,
      "learning_rate": 4.282243816254417e-05,
      "loss": 0.082,
      "step": 78000
    },
    {
      "epoch": 1.44,
      "learning_rate": 4.2776428150765606e-05,
      "loss": 0.0759,
      "step": 78500
    },
    {
      "epoch": 1.45,
      "learning_rate": 4.2730418138987044e-05,
      "loss": 0.0814,
      "step": 79000
    },
    {
      "epoch": 1.46,
      "learning_rate": 4.268440812720848e-05,
      "loss": 0.0832,
      "step": 79500
    },
    {
      "epoch": 1.47,
      "learning_rate": 4.263839811542992e-05,
      "loss": 0.0779,
      "step": 80000
    },
    {
      "epoch": 1.48,
      "learning_rate": 4.259238810365136e-05,
      "loss": 0.0796,
      "step": 80500
    },
    {
      "epoch": 1.49,
      "learning_rate": 4.2546378091872795e-05,
      "loss": 0.084,
      "step": 81000
    },
    {
      "epoch": 1.5,
      "learning_rate": 4.250036808009423e-05,
      "loss": 0.08,
      "step": 81500
    },
    {
      "epoch": 1.51,
      "learning_rate": 4.245435806831567e-05,
      "loss": 0.0796,
      "step": 82000
    },
    {
      "epoch": 1.52,
      "learning_rate": 4.240834805653711e-05,
      "loss": 0.0784,
      "step": 82500
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.2362338044758546e-05,
      "loss": 0.0783,
      "step": 83000
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.2316328032979977e-05,
      "loss": 0.0859,
      "step": 83500
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.2270318021201414e-05,
      "loss": 0.0802,
      "step": 84000
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.222430800942285e-05,
      "loss": 0.0773,
      "step": 84500
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.217829799764429e-05,
      "loss": 0.0831,
      "step": 85000
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.213228798586573e-05,
      "loss": 0.0803,
      "step": 85500
    },
    {
      "epoch": 1.58,
      "learning_rate": 4.2086277974087165e-05,
      "loss": 0.0826,
      "step": 86000
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.2040267962308596e-05,
      "loss": 0.0773,
      "step": 86500
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.1994257950530034e-05,
      "loss": 0.0765,
      "step": 87000
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.194824793875147e-05,
      "loss": 0.0801,
      "step": 87500
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.190223792697291e-05,
      "loss": 0.0806,
      "step": 88000
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.185622791519435e-05,
      "loss": 0.0814,
      "step": 88500
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.1810217903415785e-05,
      "loss": 0.0734,
      "step": 89000
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.176420789163722e-05,
      "loss": 0.0784,
      "step": 89500
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.171819787985866e-05,
      "loss": 0.0776,
      "step": 90000
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.16721878680801e-05,
      "loss": 0.0713,
      "step": 90500
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.1626177856301536e-05,
      "loss": 0.0854,
      "step": 91000
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.1580167844522974e-05,
      "loss": 0.0827,
      "step": 91500
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.153415783274441e-05,
      "loss": 0.0778,
      "step": 92000
    },
    {
      "epoch": 1.7,
      "learning_rate": 4.148814782096584e-05,
      "loss": 0.0778,
      "step": 92500
    },
    {
      "epoch": 1.71,
      "learning_rate": 4.144213780918728e-05,
      "loss": 0.082,
      "step": 93000
    },
    {
      "epoch": 1.72,
      "learning_rate": 4.139612779740872e-05,
      "loss": 0.0789,
      "step": 93500
    },
    {
      "epoch": 1.73,
      "learning_rate": 4.1350117785630156e-05,
      "loss": 0.0823,
      "step": 94000
    },
    {
      "epoch": 1.74,
      "learning_rate": 4.130410777385159e-05,
      "loss": 0.0812,
      "step": 94500
    },
    {
      "epoch": 1.75,
      "learning_rate": 4.125809776207303e-05,
      "loss": 0.0852,
      "step": 95000
    },
    {
      "epoch": 1.76,
      "learning_rate": 4.121208775029446e-05,
      "loss": 0.0767,
      "step": 95500
    },
    {
      "epoch": 1.77,
      "learning_rate": 4.11660777385159e-05,
      "loss": 0.0775,
      "step": 96000
    },
    {
      "epoch": 1.78,
      "learning_rate": 4.112006772673734e-05,
      "loss": 0.0823,
      "step": 96500
    },
    {
      "epoch": 1.79,
      "learning_rate": 4.1074057714958775e-05,
      "loss": 0.0761,
      "step": 97000
    },
    {
      "epoch": 1.79,
      "learning_rate": 4.102804770318021e-05,
      "loss": 0.0715,
      "step": 97500
    },
    {
      "epoch": 1.8,
      "learning_rate": 4.098203769140165e-05,
      "loss": 0.0725,
      "step": 98000
    },
    {
      "epoch": 1.81,
      "learning_rate": 4.093602767962309e-05,
      "loss": 0.0818,
      "step": 98500
    },
    {
      "epoch": 1.82,
      "learning_rate": 4.0890017667844526e-05,
      "loss": 0.0827,
      "step": 99000
    },
    {
      "epoch": 1.83,
      "learning_rate": 4.0844007656065964e-05,
      "loss": 0.0795,
      "step": 99500
    },
    {
      "epoch": 1.84,
      "learning_rate": 4.07979976442874e-05,
      "loss": 0.0823,
      "step": 100000
    },
    {
      "epoch": 1.85,
      "learning_rate": 4.075198763250884e-05,
      "loss": 0.0778,
      "step": 100500
    },
    {
      "epoch": 1.86,
      "learning_rate": 4.070597762073027e-05,
      "loss": 0.0806,
      "step": 101000
    },
    {
      "epoch": 1.87,
      "learning_rate": 4.065996760895171e-05,
      "loss": 0.0857,
      "step": 101500
    },
    {
      "epoch": 1.88,
      "learning_rate": 4.0613957597173146e-05,
      "loss": 0.082,
      "step": 102000
    },
    {
      "epoch": 1.89,
      "learning_rate": 4.0567947585394584e-05,
      "loss": 0.0777,
      "step": 102500
    },
    {
      "epoch": 1.9,
      "learning_rate": 4.052193757361602e-05,
      "loss": 0.0705,
      "step": 103000
    },
    {
      "epoch": 1.9,
      "learning_rate": 4.047592756183746e-05,
      "loss": 0.0795,
      "step": 103500
    },
    {
      "epoch": 1.91,
      "learning_rate": 4.042991755005889e-05,
      "loss": 0.0761,
      "step": 104000
    },
    {
      "epoch": 1.92,
      "learning_rate": 4.038390753828033e-05,
      "loss": 0.0784,
      "step": 104500
    },
    {
      "epoch": 1.93,
      "learning_rate": 4.0337897526501766e-05,
      "loss": 0.0804,
      "step": 105000
    },
    {
      "epoch": 1.94,
      "learning_rate": 4.02918875147232e-05,
      "loss": 0.0789,
      "step": 105500
    },
    {
      "epoch": 1.95,
      "learning_rate": 4.024587750294464e-05,
      "loss": 0.0699,
      "step": 106000
    },
    {
      "epoch": 1.96,
      "learning_rate": 4.0199867491166086e-05,
      "loss": 0.0748,
      "step": 106500
    },
    {
      "epoch": 1.97,
      "learning_rate": 4.0153857479387517e-05,
      "loss": 0.0801,
      "step": 107000
    },
    {
      "epoch": 1.98,
      "learning_rate": 4.0107847467608954e-05,
      "loss": 0.0795,
      "step": 107500
    },
    {
      "epoch": 1.99,
      "learning_rate": 4.006183745583039e-05,
      "loss": 0.0834,
      "step": 108000
    },
    {
      "epoch": 2.0,
      "learning_rate": 4.001582744405183e-05,
      "loss": 0.0786,
      "step": 108500
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9535093440260232,
      "eval_loss": 0.08472639322280884,
      "eval_runtime": 916.4191,
      "eval_samples_per_second": 271.048,
      "eval_steps_per_second": 16.941,
      "step": 108672
    },
    {
      "epoch": 2.01,
      "learning_rate": 3.996981743227327e-05,
      "loss": 0.0723,
      "step": 109000
    },
    {
      "epoch": 2.02,
      "learning_rate": 3.9923807420494705e-05,
      "loss": 0.0672,
      "step": 109500
    },
    {
      "epoch": 2.02,
      "learning_rate": 3.9877797408716136e-05,
      "loss": 0.0694,
      "step": 110000
    },
    {
      "epoch": 2.03,
      "learning_rate": 3.9831787396937574e-05,
      "loss": 0.0671,
      "step": 110500
    },
    {
      "epoch": 2.04,
      "learning_rate": 3.978577738515901e-05,
      "loss": 0.0714,
      "step": 111000
    },
    {
      "epoch": 2.05,
      "learning_rate": 3.973976737338045e-05,
      "loss": 0.0668,
      "step": 111500
    },
    {
      "epoch": 2.06,
      "learning_rate": 3.969375736160189e-05,
      "loss": 0.0641,
      "step": 112000
    },
    {
      "epoch": 2.07,
      "learning_rate": 3.9647747349823325e-05,
      "loss": 0.0654,
      "step": 112500
    },
    {
      "epoch": 2.08,
      "learning_rate": 3.9601737338044756e-05,
      "loss": 0.0737,
      "step": 113000
    },
    {
      "epoch": 2.09,
      "learning_rate": 3.9555727326266194e-05,
      "loss": 0.0629,
      "step": 113500
    },
    {
      "epoch": 2.1,
      "learning_rate": 3.950971731448763e-05,
      "loss": 0.0648,
      "step": 114000
    },
    {
      "epoch": 2.11,
      "learning_rate": 3.946370730270907e-05,
      "loss": 0.0658,
      "step": 114500
    },
    {
      "epoch": 2.12,
      "learning_rate": 3.9417697290930514e-05,
      "loss": 0.0691,
      "step": 115000
    },
    {
      "epoch": 2.13,
      "learning_rate": 3.9371687279151945e-05,
      "loss": 0.0681,
      "step": 115500
    },
    {
      "epoch": 2.13,
      "learning_rate": 3.932567726737338e-05,
      "loss": 0.0689,
      "step": 116000
    },
    {
      "epoch": 2.14,
      "learning_rate": 3.927966725559482e-05,
      "loss": 0.0683,
      "step": 116500
    },
    {
      "epoch": 2.15,
      "learning_rate": 3.923365724381626e-05,
      "loss": 0.0701,
      "step": 117000
    },
    {
      "epoch": 2.16,
      "learning_rate": 3.9187647232037696e-05,
      "loss": 0.0669,
      "step": 117500
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.914163722025913e-05,
      "loss": 0.0716,
      "step": 118000
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.9095627208480564e-05,
      "loss": 0.0685,
      "step": 118500
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.9049617196702e-05,
      "loss": 0.0669,
      "step": 119000
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.900360718492344e-05,
      "loss": 0.0786,
      "step": 119500
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.895759717314488e-05,
      "loss": 0.0675,
      "step": 120000
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.8911587161366315e-05,
      "loss": 0.0692,
      "step": 120500
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.886557714958775e-05,
      "loss": 0.0668,
      "step": 121000
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.881956713780919e-05,
      "loss": 0.0715,
      "step": 121500
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.877355712603062e-05,
      "loss": 0.0696,
      "step": 122000
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.872754711425206e-05,
      "loss": 0.0727,
      "step": 122500
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.86815371024735e-05,
      "loss": 0.0739,
      "step": 123000
    },
    {
      "epoch": 2.27,
      "learning_rate": 3.863552709069494e-05,
      "loss": 0.0696,
      "step": 123500
    },
    {
      "epoch": 2.28,
      "learning_rate": 3.858951707891638e-05,
      "loss": 0.0735,
      "step": 124000
    },
    {
      "epoch": 2.29,
      "learning_rate": 3.854350706713781e-05,
      "loss": 0.0699,
      "step": 124500
    },
    {
      "epoch": 2.3,
      "learning_rate": 3.849749705535925e-05,
      "loss": 0.0666,
      "step": 125000
    },
    {
      "epoch": 2.31,
      "learning_rate": 3.8451487043580686e-05,
      "loss": 0.0752,
      "step": 125500
    },
    {
      "epoch": 2.32,
      "learning_rate": 3.8405477031802124e-05,
      "loss": 0.0676,
      "step": 126000
    },
    {
      "epoch": 2.33,
      "learning_rate": 3.835946702002356e-05,
      "loss": 0.0672,
      "step": 126500
    },
    {
      "epoch": 2.34,
      "learning_rate": 3.8313457008245e-05,
      "loss": 0.0681,
      "step": 127000
    },
    {
      "epoch": 2.35,
      "learning_rate": 3.826744699646643e-05,
      "loss": 0.0756,
      "step": 127500
    },
    {
      "epoch": 2.36,
      "learning_rate": 3.822143698468787e-05,
      "loss": 0.0782,
      "step": 128000
    },
    {
      "epoch": 2.36,
      "learning_rate": 3.8175426972909306e-05,
      "loss": 0.072,
      "step": 128500
    },
    {
      "epoch": 2.37,
      "learning_rate": 3.812941696113074e-05,
      "loss": 0.0694,
      "step": 129000
    },
    {
      "epoch": 2.38,
      "learning_rate": 3.808340694935218e-05,
      "loss": 0.0686,
      "step": 129500
    },
    {
      "epoch": 2.39,
      "learning_rate": 3.803739693757362e-05,
      "loss": 0.0631,
      "step": 130000
    },
    {
      "epoch": 2.4,
      "learning_rate": 3.799138692579505e-05,
      "loss": 0.074,
      "step": 130500
    },
    {
      "epoch": 2.41,
      "learning_rate": 3.794537691401649e-05,
      "loss": 0.073,
      "step": 131000
    },
    {
      "epoch": 2.42,
      "learning_rate": 3.7899366902237925e-05,
      "loss": 0.0688,
      "step": 131500
    },
    {
      "epoch": 2.43,
      "learning_rate": 3.785335689045937e-05,
      "loss": 0.0688,
      "step": 132000
    },
    {
      "epoch": 2.44,
      "learning_rate": 3.780734687868081e-05,
      "loss": 0.064,
      "step": 132500
    },
    {
      "epoch": 2.45,
      "learning_rate": 3.776133686690224e-05,
      "loss": 0.0658,
      "step": 133000
    },
    {
      "epoch": 2.46,
      "learning_rate": 3.7715326855123676e-05,
      "loss": 0.0726,
      "step": 133500
    },
    {
      "epoch": 2.47,
      "learning_rate": 3.7669316843345114e-05,
      "loss": 0.0678,
      "step": 134000
    },
    {
      "epoch": 2.48,
      "learning_rate": 3.762330683156655e-05,
      "loss": 0.0704,
      "step": 134500
    },
    {
      "epoch": 2.48,
      "learning_rate": 3.757729681978799e-05,
      "loss": 0.0714,
      "step": 135000
    },
    {
      "epoch": 2.49,
      "learning_rate": 3.753128680800943e-05,
      "loss": 0.0678,
      "step": 135500
    },
    {
      "epoch": 2.5,
      "learning_rate": 3.748527679623086e-05,
      "loss": 0.0709,
      "step": 136000
    },
    {
      "epoch": 2.51,
      "learning_rate": 3.7439266784452296e-05,
      "loss": 0.0709,
      "step": 136500
    },
    {
      "epoch": 2.52,
      "learning_rate": 3.7393256772673734e-05,
      "loss": 0.0671,
      "step": 137000
    },
    {
      "epoch": 2.53,
      "learning_rate": 3.734724676089517e-05,
      "loss": 0.0724,
      "step": 137500
    },
    {
      "epoch": 2.54,
      "learning_rate": 3.730123674911661e-05,
      "loss": 0.068,
      "step": 138000
    },
    {
      "epoch": 2.55,
      "learning_rate": 3.725522673733805e-05,
      "loss": 0.0727,
      "step": 138500
    },
    {
      "epoch": 2.56,
      "learning_rate": 3.7209216725559485e-05,
      "loss": 0.0704,
      "step": 139000
    },
    {
      "epoch": 2.57,
      "learning_rate": 3.7163206713780916e-05,
      "loss": 0.0714,
      "step": 139500
    },
    {
      "epoch": 2.58,
      "learning_rate": 3.711719670200235e-05,
      "loss": 0.0634,
      "step": 140000
    },
    {
      "epoch": 2.59,
      "learning_rate": 3.70711866902238e-05,
      "loss": 0.0703,
      "step": 140500
    },
    {
      "epoch": 2.59,
      "learning_rate": 3.7025176678445236e-05,
      "loss": 0.069,
      "step": 141000
    },
    {
      "epoch": 2.6,
      "learning_rate": 3.697916666666667e-05,
      "loss": 0.0731,
      "step": 141500
    },
    {
      "epoch": 2.61,
      "learning_rate": 3.6933156654888104e-05,
      "loss": 0.0644,
      "step": 142000
    },
    {
      "epoch": 2.62,
      "learning_rate": 3.688714664310954e-05,
      "loss": 0.0685,
      "step": 142500
    },
    {
      "epoch": 2.63,
      "learning_rate": 3.684113663133098e-05,
      "loss": 0.072,
      "step": 143000
    },
    {
      "epoch": 2.64,
      "learning_rate": 3.679512661955242e-05,
      "loss": 0.075,
      "step": 143500
    },
    {
      "epoch": 2.65,
      "learning_rate": 3.6749116607773855e-05,
      "loss": 0.0674,
      "step": 144000
    },
    {
      "epoch": 2.66,
      "learning_rate": 3.670310659599529e-05,
      "loss": 0.0703,
      "step": 144500
    },
    {
      "epoch": 2.67,
      "learning_rate": 3.6657096584216724e-05,
      "loss": 0.0705,
      "step": 145000
    },
    {
      "epoch": 2.68,
      "learning_rate": 3.661108657243816e-05,
      "loss": 0.0729,
      "step": 145500
    },
    {
      "epoch": 2.69,
      "learning_rate": 3.65650765606596e-05,
      "loss": 0.067,
      "step": 146000
    },
    {
      "epoch": 2.7,
      "learning_rate": 3.651906654888104e-05,
      "loss": 0.0694,
      "step": 146500
    },
    {
      "epoch": 2.71,
      "learning_rate": 3.6473056537102475e-05,
      "loss": 0.0692,
      "step": 147000
    },
    {
      "epoch": 2.71,
      "learning_rate": 3.642704652532391e-05,
      "loss": 0.0709,
      "step": 147500
    },
    {
      "epoch": 2.72,
      "learning_rate": 3.6381036513545344e-05,
      "loss": 0.0681,
      "step": 148000
    },
    {
      "epoch": 2.73,
      "learning_rate": 3.633502650176678e-05,
      "loss": 0.0703,
      "step": 148500
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.6289016489988226e-05,
      "loss": 0.0633,
      "step": 149000
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.6243006478209664e-05,
      "loss": 0.0664,
      "step": 149500
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.61969964664311e-05,
      "loss": 0.0694,
      "step": 150000
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.615098645465253e-05,
      "loss": 0.0716,
      "step": 150500
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.610497644287397e-05,
      "loss": 0.0702,
      "step": 151000
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.605896643109541e-05,
      "loss": 0.0743,
      "step": 151500
    },
    {
      "epoch": 2.8,
      "learning_rate": 3.6012956419316846e-05,
      "loss": 0.0705,
      "step": 152000
    },
    {
      "epoch": 2.81,
      "learning_rate": 3.596694640753828e-05,
      "loss": 0.0667,
      "step": 152500
    },
    {
      "epoch": 2.82,
      "learning_rate": 3.592093639575972e-05,
      "loss": 0.0662,
      "step": 153000
    },
    {
      "epoch": 2.83,
      "learning_rate": 3.587492638398116e-05,
      "loss": 0.0725,
      "step": 153500
    },
    {
      "epoch": 2.83,
      "learning_rate": 3.582891637220259e-05,
      "loss": 0.0665,
      "step": 154000
    },
    {
      "epoch": 2.84,
      "learning_rate": 3.578290636042403e-05,
      "loss": 0.0683,
      "step": 154500
    },
    {
      "epoch": 2.85,
      "learning_rate": 3.5736896348645465e-05,
      "loss": 0.0712,
      "step": 155000
    },
    {
      "epoch": 2.86,
      "learning_rate": 3.56908863368669e-05,
      "loss": 0.0703,
      "step": 155500
    },
    {
      "epoch": 2.87,
      "learning_rate": 3.564487632508834e-05,
      "loss": 0.0695,
      "step": 156000
    },
    {
      "epoch": 2.88,
      "learning_rate": 3.559886631330978e-05,
      "loss": 0.0672,
      "step": 156500
    },
    {
      "epoch": 2.89,
      "learning_rate": 3.555285630153121e-05,
      "loss": 0.0673,
      "step": 157000
    },
    {
      "epoch": 2.9,
      "learning_rate": 3.5506846289752654e-05,
      "loss": 0.0661,
      "step": 157500
    },
    {
      "epoch": 2.91,
      "learning_rate": 3.546083627797409e-05,
      "loss": 0.0694,
      "step": 158000
    },
    {
      "epoch": 2.92,
      "learning_rate": 3.541482626619553e-05,
      "loss": 0.0668,
      "step": 158500
    },
    {
      "epoch": 2.93,
      "learning_rate": 3.536881625441697e-05,
      "loss": 0.0725,
      "step": 159000
    },
    {
      "epoch": 2.94,
      "learning_rate": 3.53228062426384e-05,
      "loss": 0.0661,
      "step": 159500
    },
    {
      "epoch": 2.94,
      "learning_rate": 3.5276796230859836e-05,
      "loss": 0.073,
      "step": 160000
    }
  ],
  "max_steps": 543360,
  "num_train_epochs": 10,
  "total_flos": 6.6892572131328e+17,
  "trial_name": null,
  "trial_params": null
}