trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.0008022398536714506,
  "eval_steps": 100,
  "global_step": 25,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 7.037811279296875,
      "learning_rate": 8e-05,
      "loss": 2.3007,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.989239692687988,
      "learning_rate": 6e-05,
      "loss": 1.0676,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.793990612030029,
      "learning_rate": 4e-05,
      "loss": 0.5856,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.514509439468384,
      "learning_rate": 2e-05,
      "loss": 0.5141,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.3384485244750977,
      "learning_rate": 0.0,
      "loss": 0.5086,
      "step": 25
    },
    {
      "epoch": 0.0,
      "step": 25,
      "total_flos": 479173072435200.0,
      "train_loss": 0.9953355026245118,
      "train_runtime": 114.2038,
      "train_samples_per_second": 1.751,
      "train_steps_per_second": 0.219
    }
  ],
  "logging_steps": 5,
  "max_steps": 25,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "total_flos": 479173072435200.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}