checkpoint-40/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 5.0,
  "eval_steps": 500,
  "global_step": 40,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.38,
      "grad_norm": 1.9970229864120483,
      "learning_rate": 0.0001611111111111111,
      "loss": 2.0118,
      "step": 11
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.440118432044983,
      "learning_rate": 0.0001,
      "loss": 1.2712,
      "step": 22
    },
    {
      "epoch": 4.12,
      "grad_norm": 1.1562683582305908,
      "learning_rate": 3.888888888888889e-05,
      "loss": 0.963,
      "step": 33
    }
  ],
  "logging_steps": 11,
  "max_steps": 40,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 500,
  "total_flos": 6778114591948800.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}