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": 0.00016923076923076923,
      "loss": 0.6187,
      "step": 10
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.00013076923076923077,
      "loss": 0.6278,
      "step": 20
    },
    {
      "epoch": 2.31,
      "learning_rate": 9.230769230769232e-05,
      "loss": 0.5182,
      "step": 30
    },
    {
      "epoch": 3.08,
      "learning_rate": 5.384615384615385e-05,
      "loss": 0.5655,
      "step": 40
    },
    {
      "epoch": 3.85,
      "learning_rate": 1.5384615384615387e-05,
      "loss": 0.3918,
      "step": 50
    },
    {
      "epoch": 4.0,
      "step": 52,
      "total_flos": 2.191473438916608e+17,
      "train_loss": 0.5333058467278113,
      "train_runtime": 22.4478,
      "train_samples_per_second": 35.638,
      "train_steps_per_second": 2.316
    }
  ],
  "max_steps": 52,
  "num_train_epochs": 4,
  "total_flos": 2.191473438916608e+17,
  "trial_name": null,
  "trial_params": null
}