{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.945945945945946,
  "global_step": 36,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.05,
      "learning_rate": 1e-05,
      "loss": 0.5752,
      "step": 1
    },
    {
      "epoch": 0.11,
      "learning_rate": 2e-05,
      "loss": 0.5373,
      "step": 2
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9957341762950346e-05,
      "loss": 0.4761,
      "step": 3
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.982973099683902e-05,
      "loss": 0.4532,
      "step": 4
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.961825643172819e-05,
      "loss": 0.433,
      "step": 5
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.932472229404356e-05,
      "loss": 0.3961,
      "step": 6
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.8951632913550625e-05,
      "loss": 0.3748,
      "step": 7
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.8502171357296144e-05,
      "loss": 0.3499,
      "step": 8
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.7980172272802398e-05,
      "loss": 0.3753,
      "step": 9
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.7390089172206594e-05,
      "loss": 0.3512,
      "step": 10
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.6736956436465573e-05,
      "loss": 0.3141,
      "step": 11
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.6026346363792565e-05,
      "loss": 0.3097,
      "step": 12
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.526432162877356e-05,
      "loss": 0.3212,
      "step": 13
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.4457383557765385e-05,
      "loss": 0.319,
      "step": 14
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.3612416661871532e-05,
      "loss": 0.2819,
      "step": 15
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.2736629900720832e-05,
      "loss": 0.2806,
      "step": 16
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.1837495178165706e-05,
      "loss": 0.3209,
      "step": 17
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.092268359463302e-05,
      "loss": 0.2699,
      "step": 18
    },
    {
      "epoch": 1.03,
      "learning_rate": 1e-05,
      "loss": 0.2524,
      "step": 19
    },
    {
      "epoch": 1.08,
      "learning_rate": 9.07731640536698e-06,
      "loss": 0.2383,
      "step": 20
    },
    {
      "epoch": 1.14,
      "learning_rate": 8.162504821834296e-06,
      "loss": 0.2116,
      "step": 21
    },
    {
      "epoch": 1.19,
      "learning_rate": 7.263370099279173e-06,
      "loss": 0.2579,
      "step": 22
    },
    {
      "epoch": 1.24,
      "learning_rate": 6.387583338128471e-06,
      "loss": 0.2247,
      "step": 23
    },
    {
      "epoch": 1.3,
      "learning_rate": 5.542616442234618e-06,
      "loss": 0.2532,
      "step": 24
    },
    {
      "epoch": 1.35,
      "learning_rate": 4.7356783712264405e-06,
      "loss": 0.2392,
      "step": 25
    },
    {
      "epoch": 1.41,
      "learning_rate": 3.973653636207437e-06,
      "loss": 0.2334,
      "step": 26
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.2630435635344283e-06,
      "loss": 0.2479,
      "step": 27
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.6099108277934105e-06,
      "loss": 0.2025,
      "step": 28
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.019827727197605e-06,
      "loss": 0.2124,
      "step": 29
    },
    {
      "epoch": 1.62,
      "learning_rate": 1.4978286427038602e-06,
      "loss": 0.1978,
      "step": 30
    },
    {
      "epoch": 1.68,
      "learning_rate": 1.0483670864493777e-06,
      "loss": 0.2144,
      "step": 31
    },
    {
      "epoch": 1.73,
      "learning_rate": 6.752777059564431e-07,
      "loss": 0.2246,
      "step": 32
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.817435682718096e-07,
      "loss": 0.1932,
      "step": 33
    },
    {
      "epoch": 1.84,
      "learning_rate": 1.7026900316098217e-07,
      "loss": 0.1926,
      "step": 34
    },
    {
      "epoch": 1.89,
      "learning_rate": 4.2658237049655325e-08,
      "loss": 0.2296,
      "step": 35
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.0,
      "loss": 0.1756,
      "step": 36
    },
    {
      "epoch": 1.95,
      "step": 36,
      "total_flos": 9.480789044704051e+16,
      "train_loss": 0.29834916690985364,
      "train_runtime": 1836.8,
      "train_samples_per_second": 2.546,
      "train_steps_per_second": 0.02
    }
  ],
  "max_steps": 36,
  "num_train_epochs": 2,
  "total_flos": 9.480789044704051e+16,
  "trial_name": null,
  "trial_params": null
}