trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.992,
  "eval_steps": 62,
  "global_step": 62,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.24,
      "grad_norm": 2.192871649096338,
      "learning_rate": 3.6290322580645165e-07,
      "loss": -0.0471,
      "step": 15
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.8012133986422623,
      "learning_rate": 7.258064516129033e-07,
      "loss": -0.0621,
      "step": 30
    },
    {
      "epoch": 0.72,
      "grad_norm": 2.0031979158169264,
      "learning_rate": 1.088709677419355e-06,
      "loss": -0.1129,
      "step": 45
    },
    {
      "epoch": 0.96,
      "grad_norm": 13.949909495504963,
      "learning_rate": 1.4516129032258066e-06,
      "loss": -0.2486,
      "step": 60
    }
  ],
  "logging_steps": 15,
  "max_steps": 312,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 6,
  "save_steps": 62,
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}