{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 9.990557129367328,
  "eval_steps": 500,
  "global_step": 5290,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.94,
      "grad_norm": 10.44521713256836,
      "learning_rate": 4.527410207939509e-05,
      "loss": 0.5661,
      "step": 500
    },
    {
      "epoch": 1.89,
      "grad_norm": 8.467999458312988,
      "learning_rate": 4.054820415879017e-05,
      "loss": 0.2756,
      "step": 1000
    },
    {
      "epoch": 2.83,
      "grad_norm": 10.823616981506348,
      "learning_rate": 3.582230623818525e-05,
      "loss": 0.1948,
      "step": 1500
    },
    {
      "epoch": 3.78,
      "grad_norm": 5.485235214233398,
      "learning_rate": 3.1096408317580344e-05,
      "loss": 0.1482,
      "step": 2000
    },
    {
      "epoch": 4.72,
      "grad_norm": 0.24583563208580017,
      "learning_rate": 2.637051039697543e-05,
      "loss": 0.108,
      "step": 2500
    },
    {
      "epoch": 5.67,
      "grad_norm": 1.4922066926956177,
      "learning_rate": 2.1644612476370512e-05,
      "loss": 0.086,
      "step": 3000
    },
    {
      "epoch": 6.61,
      "grad_norm": 1.4137184619903564,
      "learning_rate": 1.6918714555765597e-05,
      "loss": 0.0682,
      "step": 3500
    },
    {
      "epoch": 7.55,
      "grad_norm": 1.483734369277954,
      "learning_rate": 1.2192816635160681e-05,
      "loss": 0.0465,
      "step": 4000
    },
    {
      "epoch": 8.5,
      "grad_norm": 0.06825185567140579,
      "learning_rate": 7.466918714555767e-06,
      "loss": 0.0366,
      "step": 4500
    },
    {
      "epoch": 9.44,
      "grad_norm": 2.011765956878662,
      "learning_rate": 2.741020793950851e-06,
      "loss": 0.0297,
      "step": 5000
    },
    {
      "epoch": 9.99,
      "step": 5290,
      "total_flos": 3.044185228774305e+16,
      "train_loss": 0.148528479718531,
      "train_runtime": 26099.405,
      "train_samples_per_second": 6.49,
      "train_steps_per_second": 0.203
    }
  ],
  "logging_steps": 500,
  "max_steps": 5290,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 500,
  "total_flos": 3.044185228774305e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}