{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 20.0,
  "eval_steps": 2.0,
  "global_step": 2680,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.19,
      "learning_rate": 1.9813432835820897e-05,
      "loss": 0.6233,
      "step": 25
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.9626865671641793e-05,
      "loss": 0.6221,
      "step": 50
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.9440298507462686e-05,
      "loss": 0.6235,
      "step": 75
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.9253731343283585e-05,
      "loss": 0.5969,
      "step": 100
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.9067164179104477e-05,
      "loss": 0.6,
      "step": 125
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.8880597014925376e-05,
      "loss": 0.615,
      "step": 150
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.869402985074627e-05,
      "loss": 0.5995,
      "step": 175
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.8507462686567165e-05,
      "loss": 0.5833,
      "step": 200
    },
    {
      "epoch": 1.68,
      "learning_rate": 1.832089552238806e-05,
      "loss": 0.6055,
      "step": 225
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.8134328358208956e-05,
      "loss": 0.5892,
      "step": 250
    },
    {
      "epoch": 2.05,
      "learning_rate": 1.7947761194029852e-05,
      "loss": 0.5942,
      "step": 275
    },
    {
      "epoch": 2.24,
      "learning_rate": 1.7761194029850748e-05,
      "loss": 0.5543,
      "step": 300
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.7574626865671644e-05,
      "loss": 0.5465,
      "step": 325
    },
    {
      "epoch": 2.61,
      "learning_rate": 1.738805970149254e-05,
      "loss": 0.5416,
      "step": 350
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.7201492537313432e-05,
      "loss": 0.5653,
      "step": 375
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.701492537313433e-05,
      "loss": 0.5215,
      "step": 400
    },
    {
      "epoch": 3.17,
      "learning_rate": 1.6828358208955223e-05,
      "loss": 0.4565,
      "step": 425
    },
    {
      "epoch": 3.36,
      "learning_rate": 1.6641791044776122e-05,
      "loss": 0.4517,
      "step": 450
    },
    {
      "epoch": 3.54,
      "learning_rate": 1.6455223880597015e-05,
      "loss": 0.4791,
      "step": 475
    },
    {
      "epoch": 3.73,
      "learning_rate": 1.626865671641791e-05,
      "loss": 0.4519,
      "step": 500
    },
    {
      "epoch": 3.92,
      "learning_rate": 1.6082089552238806e-05,
      "loss": 0.4671,
      "step": 525
    },
    {
      "epoch": 4.1,
      "learning_rate": 1.5895522388059702e-05,
      "loss": 0.438,
      "step": 550
    },
    {
      "epoch": 4.29,
      "learning_rate": 1.5708955223880598e-05,
      "loss": 0.3501,
      "step": 575
    },
    {
      "epoch": 4.48,
      "learning_rate": 1.5522388059701494e-05,
      "loss": 0.3944,
      "step": 600
    },
    {
      "epoch": 4.66,
      "learning_rate": 1.533582089552239e-05,
      "loss": 0.3881,
      "step": 625
    },
    {
      "epoch": 4.85,
      "learning_rate": 1.5149253731343285e-05,
      "loss": 0.3794,
      "step": 650
    },
    {
      "epoch": 5.04,
      "learning_rate": 1.496268656716418e-05,
      "loss": 0.3824,
      "step": 675
    },
    {
      "epoch": 5.22,
      "learning_rate": 1.4776119402985077e-05,
      "loss": 0.3098,
      "step": 700
    },
    {
      "epoch": 5.41,
      "learning_rate": 1.4589552238805971e-05,
      "loss": 0.2898,
      "step": 725
    },
    {
      "epoch": 5.6,
      "learning_rate": 1.4402985074626867e-05,
      "loss": 0.3009,
      "step": 750
    },
    {
      "epoch": 5.78,
      "learning_rate": 1.4216417910447761e-05,
      "loss": 0.3145,
      "step": 775
    },
    {
      "epoch": 5.97,
      "learning_rate": 1.4029850746268658e-05,
      "loss": 0.323,
      "step": 800
    },
    {
      "epoch": 6.16,
      "learning_rate": 1.3843283582089553e-05,
      "loss": 0.2488,
      "step": 825
    },
    {
      "epoch": 6.34,
      "learning_rate": 1.365671641791045e-05,
      "loss": 0.2145,
      "step": 850
    },
    {
      "epoch": 6.53,
      "learning_rate": 1.3470149253731344e-05,
      "loss": 0.2635,
      "step": 875
    },
    {
      "epoch": 6.72,
      "learning_rate": 1.328358208955224e-05,
      "loss": 0.2422,
      "step": 900
    },
    {
      "epoch": 6.9,
      "learning_rate": 1.3097014925373134e-05,
      "loss": 0.2691,
      "step": 925
    },
    {
      "epoch": 7.09,
      "learning_rate": 1.2910447761194032e-05,
      "loss": 0.2335,
      "step": 950
    },
    {
      "epoch": 7.28,
      "learning_rate": 1.2723880597014926e-05,
      "loss": 0.2011,
      "step": 975
    },
    {
      "epoch": 7.46,
      "learning_rate": 1.2537313432835823e-05,
      "loss": 0.1889,
      "step": 1000
    },
    {
      "epoch": 7.65,
      "learning_rate": 1.2350746268656717e-05,
      "loss": 0.2176,
      "step": 1025
    },
    {
      "epoch": 7.84,
      "learning_rate": 1.2164179104477613e-05,
      "loss": 0.2217,
      "step": 1050
    },
    {
      "epoch": 8.02,
      "learning_rate": 1.1977611940298509e-05,
      "loss": 0.2057,
      "step": 1075
    },
    {
      "epoch": 8.21,
      "learning_rate": 1.1791044776119405e-05,
      "loss": 0.1614,
      "step": 1100
    },
    {
      "epoch": 8.4,
      "learning_rate": 1.1604477611940299e-05,
      "loss": 0.1548,
      "step": 1125
    },
    {
      "epoch": 8.58,
      "learning_rate": 1.1417910447761196e-05,
      "loss": 0.1576,
      "step": 1150
    },
    {
      "epoch": 8.77,
      "learning_rate": 1.123134328358209e-05,
      "loss": 0.1666,
      "step": 1175
    },
    {
      "epoch": 8.96,
      "learning_rate": 1.1044776119402986e-05,
      "loss": 0.1633,
      "step": 1200
    },
    {
      "epoch": 9.14,
      "learning_rate": 1.0858208955223882e-05,
      "loss": 0.122,
      "step": 1225
    },
    {
      "epoch": 9.33,
      "learning_rate": 1.0671641791044778e-05,
      "loss": 0.1254,
      "step": 1250
    },
    {
      "epoch": 9.51,
      "learning_rate": 1.0485074626865672e-05,
      "loss": 0.1193,
      "step": 1275
    },
    {
      "epoch": 9.7,
      "learning_rate": 1.029850746268657e-05,
      "loss": 0.1602,
      "step": 1300
    },
    {
      "epoch": 9.89,
      "learning_rate": 1.0111940298507463e-05,
      "loss": 0.161,
      "step": 1325
    },
    {
      "epoch": 10.07,
      "learning_rate": 9.925373134328359e-06,
      "loss": 0.1128,
      "step": 1350
    },
    {
      "epoch": 10.26,
      "learning_rate": 9.738805970149255e-06,
      "loss": 0.0877,
      "step": 1375
    },
    {
      "epoch": 10.45,
      "learning_rate": 9.552238805970149e-06,
      "loss": 0.1165,
      "step": 1400
    },
    {
      "epoch": 10.63,
      "learning_rate": 9.365671641791045e-06,
      "loss": 0.1232,
      "step": 1425
    },
    {
      "epoch": 10.82,
      "learning_rate": 9.17910447761194e-06,
      "loss": 0.1285,
      "step": 1450
    },
    {
      "epoch": 11.01,
      "learning_rate": 8.992537313432836e-06,
      "loss": 0.1214,
      "step": 1475
    },
    {
      "epoch": 11.19,
      "learning_rate": 8.805970149253732e-06,
      "loss": 0.0804,
      "step": 1500
    },
    {
      "epoch": 11.38,
      "learning_rate": 8.619402985074628e-06,
      "loss": 0.105,
      "step": 1525
    },
    {
      "epoch": 11.57,
      "learning_rate": 8.432835820895524e-06,
      "loss": 0.0885,
      "step": 1550
    },
    {
      "epoch": 11.75,
      "learning_rate": 8.246268656716418e-06,
      "loss": 0.1015,
      "step": 1575
    },
    {
      "epoch": 11.94,
      "learning_rate": 8.059701492537314e-06,
      "loss": 0.101,
      "step": 1600
    },
    {
      "epoch": 12.13,
      "learning_rate": 7.87313432835821e-06,
      "loss": 0.1017,
      "step": 1625
    },
    {
      "epoch": 12.31,
      "learning_rate": 7.686567164179105e-06,
      "loss": 0.08,
      "step": 1650
    },
    {
      "epoch": 12.5,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.1013,
      "step": 1675
    },
    {
      "epoch": 12.69,
      "learning_rate": 7.313432835820896e-06,
      "loss": 0.088,
      "step": 1700
    },
    {
      "epoch": 12.87,
      "learning_rate": 7.126865671641792e-06,
      "loss": 0.0816,
      "step": 1725
    },
    {
      "epoch": 13.06,
      "learning_rate": 6.9402985074626876e-06,
      "loss": 0.0657,
      "step": 1750
    },
    {
      "epoch": 13.25,
      "learning_rate": 6.7537313432835825e-06,
      "loss": 0.0765,
      "step": 1775
    },
    {
      "epoch": 13.43,
      "learning_rate": 6.567164179104478e-06,
      "loss": 0.0655,
      "step": 1800
    },
    {
      "epoch": 13.62,
      "learning_rate": 6.380597014925374e-06,
      "loss": 0.0854,
      "step": 1825
    },
    {
      "epoch": 13.81,
      "learning_rate": 6.194029850746269e-06,
      "loss": 0.0792,
      "step": 1850
    },
    {
      "epoch": 13.99,
      "learning_rate": 6.007462686567165e-06,
      "loss": 0.0778,
      "step": 1875
    },
    {
      "epoch": 14.18,
      "learning_rate": 5.820895522388061e-06,
      "loss": 0.049,
      "step": 1900
    },
    {
      "epoch": 14.37,
      "learning_rate": 5.6343283582089556e-06,
      "loss": 0.0625,
      "step": 1925
    },
    {
      "epoch": 14.55,
      "learning_rate": 5.447761194029851e-06,
      "loss": 0.0655,
      "step": 1950
    },
    {
      "epoch": 14.74,
      "learning_rate": 5.261194029850747e-06,
      "loss": 0.0636,
      "step": 1975
    },
    {
      "epoch": 14.93,
      "learning_rate": 5.074626865671642e-06,
      "loss": 0.0548,
      "step": 2000
    },
    {
      "epoch": 15.11,
      "learning_rate": 4.888059701492538e-06,
      "loss": 0.0509,
      "step": 2025
    },
    {
      "epoch": 15.3,
      "learning_rate": 4.701492537313434e-06,
      "loss": 0.0424,
      "step": 2050
    },
    {
      "epoch": 15.49,
      "learning_rate": 4.514925373134329e-06,
      "loss": 0.0566,
      "step": 2075
    },
    {
      "epoch": 15.67,
      "learning_rate": 4.3283582089552236e-06,
      "loss": 0.0547,
      "step": 2100
    },
    {
      "epoch": 15.86,
      "learning_rate": 4.141791044776119e-06,
      "loss": 0.0612,
      "step": 2125
    },
    {
      "epoch": 16.04,
      "learning_rate": 3.955223880597015e-06,
      "loss": 0.0502,
      "step": 2150
    },
    {
      "epoch": 16.23,
      "learning_rate": 3.7686567164179105e-06,
      "loss": 0.0392,
      "step": 2175
    },
    {
      "epoch": 16.42,
      "learning_rate": 3.582089552238806e-06,
      "loss": 0.0346,
      "step": 2200
    },
    {
      "epoch": 16.6,
      "learning_rate": 3.3955223880597017e-06,
      "loss": 0.043,
      "step": 2225
    },
    {
      "epoch": 16.79,
      "learning_rate": 3.208955223880597e-06,
      "loss": 0.0498,
      "step": 2250
    },
    {
      "epoch": 16.98,
      "learning_rate": 3.022388059701493e-06,
      "loss": 0.0511,
      "step": 2275
    },
    {
      "epoch": 17.16,
      "learning_rate": 2.835820895522388e-06,
      "loss": 0.0358,
      "step": 2300
    },
    {
      "epoch": 17.35,
      "learning_rate": 2.6492537313432836e-06,
      "loss": 0.0345,
      "step": 2325
    },
    {
      "epoch": 17.54,
      "learning_rate": 2.4626865671641794e-06,
      "loss": 0.0267,
      "step": 2350
    },
    {
      "epoch": 17.72,
      "learning_rate": 2.2761194029850747e-06,
      "loss": 0.0237,
      "step": 2375
    },
    {
      "epoch": 17.91,
      "learning_rate": 2.08955223880597e-06,
      "loss": 0.0383,
      "step": 2400
    },
    {
      "epoch": 18.1,
      "learning_rate": 1.9029850746268657e-06,
      "loss": 0.0396,
      "step": 2425
    },
    {
      "epoch": 18.28,
      "learning_rate": 1.7164179104477613e-06,
      "loss": 0.0361,
      "step": 2450
    },
    {
      "epoch": 18.47,
      "learning_rate": 1.5298507462686568e-06,
      "loss": 0.0279,
      "step": 2475
    },
    {
      "epoch": 18.66,
      "learning_rate": 1.3432835820895524e-06,
      "loss": 0.0385,
      "step": 2500
    },
    {
      "epoch": 18.84,
      "learning_rate": 1.1567164179104478e-06,
      "loss": 0.0313,
      "step": 2525
    },
    {
      "epoch": 19.03,
      "learning_rate": 9.701492537313434e-07,
      "loss": 0.0357,
      "step": 2550
    },
    {
      "epoch": 19.22,
      "learning_rate": 7.835820895522389e-07,
      "loss": 0.0366,
      "step": 2575
    },
    {
      "epoch": 19.4,
      "learning_rate": 5.970149253731343e-07,
      "loss": 0.0236,
      "step": 2600
    },
    {
      "epoch": 19.59,
      "learning_rate": 4.104477611940299e-07,
      "loss": 0.0205,
      "step": 2625
    },
    {
      "epoch": 19.78,
      "learning_rate": 2.2388059701492537e-07,
      "loss": 0.035,
      "step": 2650
    },
    {
      "epoch": 19.96,
      "learning_rate": 3.7313432835820895e-08,
      "loss": 0.0303,
      "step": 2675
    },
    {
      "epoch": 20.0,
      "step": 2680,
      "total_flos": 1.12493131719168e+16,
      "train_loss": 0.2142893238688138,
      "train_runtime": 3029.2991,
      "train_samples_per_second": 56.455,
      "train_steps_per_second": 0.885
    }
  ],
  "logging_steps": 25,
  "max_steps": 2680,
  "num_train_epochs": 20,
  "save_steps": 500,
  "total_flos": 1.12493131719168e+16,
  "trial_name": null,
  "trial_params": null
}