trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "global_step": 4,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.25,
      "learning_rate": 0.0,
      "loss": 1.7578,
      "step": 1
    },
    {
      "epoch": 0.5,
      "learning_rate": 2e-05,
      "loss": 1.5381,
      "step": 2
    },
    {
      "epoch": 0.75,
      "learning_rate": 2e-05,
      "loss": 1.5361,
      "step": 3
    },
    {
      "epoch": 1.0,
      "learning_rate": 2e-05,
      "loss": 1.3584,
      "step": 4
    },
    {
      "epoch": 1.0,
      "step": 4,
      "total_flos": 5397617180672.0,
      "train_loss": 1.547607421875,
      "train_runtime": 211.0832,
      "train_samples_per_second": 4.737,
      "train_steps_per_second": 0.019
    }
  ],
  "max_steps": 4,
  "num_train_epochs": 1,
  "total_flos": 5397617180672.0,
  "trial_name": null,
  "trial_params": null
}