trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.0,
  "global_step": 32,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.25,
      "learning_rate": 0.0001375,
      "loss": 0.737,
      "step": 10
    },
    {
      "epoch": 2.5,
      "learning_rate": 7.500000000000001e-05,
      "loss": 0.2549,
      "step": 20
    },
    {
      "epoch": 3.75,
      "learning_rate": 1.25e-05,
      "loss": 0.1051,
      "step": 30
    },
    {
      "epoch": 4.0,
      "step": 32,
      "total_flos": 3.905631281199514e+16,
      "train_loss": 0.34820668725296855,
      "train_runtime": 967.8691,
      "train_samples_per_second": 0.521,
      "train_steps_per_second": 0.033
    }
  ],
  "max_steps": 32,
  "num_train_epochs": 4,
  "total_flos": 3.905631281199514e+16,
  "trial_name": null,
  "trial_params": null
}