trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "global_step": 1887,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.79,
      "learning_rate": 4.957769990318725e-05,
      "loss": 1.8409,
      "step": 500
    },
    {
      "epoch": 1.59,
      "learning_rate": 3.193439032909926e-05,
      "loss": 1.1028,
      "step": 1000
    },
    {
      "epoch": 2.38,
      "learning_rate": 1.429108075501127e-05,
      "loss": 0.8693,
      "step": 1500
    },
    {
      "epoch": 3.0,
      "step": 1887,
      "total_flos": 78409675833344.0,
      "train_loss": 1.1591679785829028,
      "train_runtime": 329.0719,
      "train_samples_per_second": 801.709,
      "train_steps_per_second": 5.734
    }
  ],
  "max_steps": 1887,
  "num_train_epochs": 3,
  "total_flos": 78409675833344.0,
  "trial_name": null,
  "trial_params": null
}