trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.0,
  "global_step": 52,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.77,
      "learning_rate": 0.00016153846153846155,
      "loss": 0.3633,
      "step": 10
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0001230769230769231,
      "loss": 0.2712,
      "step": 20
    },
    {
      "epoch": 2.31,
      "learning_rate": 8.461538461538461e-05,
      "loss": 0.1212,
      "step": 30
    },
    {
      "epoch": 3.08,
      "learning_rate": 4.615384615384616e-05,
      "loss": 0.0929,
      "step": 40
    },
    {
      "epoch": 3.85,
      "learning_rate": 7.692307692307694e-06,
      "loss": 0.0569,
      "step": 50
    },
    {
      "epoch": 4.0,
      "step": 52,
      "total_flos": 6.3190524837888e+16,
      "train_loss": 0.17656130974109358,
      "train_runtime": 18.7317,
      "train_samples_per_second": 42.708,
      "train_steps_per_second": 2.776
    }
  ],
  "max_steps": 52,
  "num_train_epochs": 4,
  "total_flos": 6.3190524837888e+16,
  "trial_name": null,
  "trial_params": null
}