{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 7.99979512395001,
  "global_step": 19520,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.2,
      "learning_rate": 0.0004871926229508197,
      "loss": 0.4245,
      "step": 500
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00047438524590163935,
      "loss": 0.3278,
      "step": 1000
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.000461577868852459,
      "loss": 0.2985,
      "step": 1500
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00044877049180327874,
      "loss": 0.2814,
      "step": 2000
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0004359631147540984,
      "loss": 0.2548,
      "step": 2500
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0004231557377049181,
      "loss": 0.2388,
      "step": 3000
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.0004103483606557377,
      "loss": 0.2264,
      "step": 3500
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.00039754098360655736,
      "loss": 0.2204,
      "step": 4000
    },
    {
      "epoch": 1.84,
      "learning_rate": 0.00038473360655737703,
      "loss": 0.2129,
      "step": 4500
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.0003719262295081967,
      "loss": 0.2088,
      "step": 5000
    },
    {
      "epoch": 2.25,
      "learning_rate": 0.00035911885245901637,
      "loss": 0.1916,
      "step": 5500
    },
    {
      "epoch": 2.46,
      "learning_rate": 0.0003463114754098361,
      "loss": 0.182,
      "step": 6000
    },
    {
      "epoch": 2.66,
      "learning_rate": 0.00033350409836065576,
      "loss": 0.1856,
      "step": 6500
    },
    {
      "epoch": 2.87,
      "learning_rate": 0.00032069672131147543,
      "loss": 0.1852,
      "step": 7000
    },
    {
      "epoch": 3.07,
      "learning_rate": 0.0003078893442622951,
      "loss": 0.1718,
      "step": 7500
    },
    {
      "epoch": 3.28,
      "learning_rate": 0.00029508196721311476,
      "loss": 0.1642,
      "step": 8000
    },
    {
      "epoch": 3.48,
      "learning_rate": 0.00028227459016393443,
      "loss": 0.1619,
      "step": 8500
    },
    {
      "epoch": 3.69,
      "learning_rate": 0.0002694672131147541,
      "loss": 0.1571,
      "step": 9000
    },
    {
      "epoch": 3.89,
      "learning_rate": 0.00025665983606557377,
      "loss": 0.161,
      "step": 9500
    },
    {
      "epoch": 4.1,
      "learning_rate": 0.00024385245901639344,
      "loss": 0.1409,
      "step": 10000
    },
    {
      "epoch": 4.3,
      "learning_rate": 0.00023104508196721314,
      "loss": 0.1418,
      "step": 10500
    },
    {
      "epoch": 4.51,
      "learning_rate": 0.0002182377049180328,
      "loss": 0.1456,
      "step": 11000
    },
    {
      "epoch": 4.71,
      "learning_rate": 0.00020543032786885247,
      "loss": 0.1405,
      "step": 11500
    },
    {
      "epoch": 4.92,
      "learning_rate": 0.00019262295081967211,
      "loss": 0.1415,
      "step": 12000
    },
    {
      "epoch": 5.12,
      "learning_rate": 0.0001798155737704918,
      "loss": 0.1316,
      "step": 12500
    },
    {
      "epoch": 5.33,
      "learning_rate": 0.00016700819672131148,
      "loss": 0.1265,
      "step": 13000
    },
    {
      "epoch": 5.53,
      "learning_rate": 0.00015420081967213115,
      "loss": 0.126,
      "step": 13500
    },
    {
      "epoch": 5.74,
      "learning_rate": 0.00014139344262295082,
      "loss": 0.125,
      "step": 14000
    },
    {
      "epoch": 5.94,
      "learning_rate": 0.0001285860655737705,
      "loss": 0.1277,
      "step": 14500
    },
    {
      "epoch": 6.15,
      "learning_rate": 0.00011577868852459017,
      "loss": 0.1184,
      "step": 15000
    },
    {
      "epoch": 6.35,
      "learning_rate": 0.00010297131147540984,
      "loss": 0.1152,
      "step": 15500
    },
    {
      "epoch": 6.56,
      "learning_rate": 9.016393442622952e-05,
      "loss": 0.1173,
      "step": 16000
    },
    {
      "epoch": 6.76,
      "learning_rate": 7.735655737704917e-05,
      "loss": 0.1131,
      "step": 16500
    },
    {
      "epoch": 6.97,
      "learning_rate": 6.454918032786886e-05,
      "loss": 0.114,
      "step": 17000
    },
    {
      "epoch": 7.17,
      "learning_rate": 5.1741803278688525e-05,
      "loss": 0.1112,
      "step": 17500
    },
    {
      "epoch": 7.38,
      "learning_rate": 3.89344262295082e-05,
      "loss": 0.1097,
      "step": 18000
    },
    {
      "epoch": 7.58,
      "learning_rate": 2.612704918032787e-05,
      "loss": 0.106,
      "step": 18500
    },
    {
      "epoch": 7.79,
      "learning_rate": 1.331967213114754e-05,
      "loss": 0.108,
      "step": 19000
    },
    {
      "epoch": 7.99,
      "learning_rate": 5.122950819672132e-07,
      "loss": 0.1062,
      "step": 19500
    },
    {
      "epoch": 8.0,
      "step": 19520,
      "total_flos": 1.2620192221249536e+17,
      "train_loss": 0.1722577346397228,
      "train_runtime": 32063.6353,
      "train_samples_per_second": 19.484,
      "train_steps_per_second": 0.609
    }
  ],
  "max_steps": 19520,
  "num_train_epochs": 8,
  "total_flos": 1.2620192221249536e+17,
  "trial_name": null,
  "trial_params": null
}