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": 2e-05,
      "loss": 1.25,
      "step": 1
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 1.2812,
      "step": 2
    },
    {
      "epoch": 0.75,
      "learning_rate": 5.000000000000003e-06,
      "loss": 1.2812,
      "step": 3
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.0,
      "loss": 1.1953,
      "step": 4
    },
    {
      "epoch": 1.0,
      "step": 4,
      "total_flos": 402050395406336.0,
      "train_loss": 1.251953125,
      "train_runtime": 23.5154,
      "train_samples_per_second": 2.339,
      "train_steps_per_second": 0.17
    }
  ],
  "max_steps": 4,
  "num_train_epochs": 1,
  "total_flos": 402050395406336.0,
  "trial_name": null,
  "trial_params": null
}