trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "global_step": 1952,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.51,
      "learning_rate": 3.719262295081967e-05,
      "loss": 1.5121,
      "step": 500
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.4385245901639343e-05,
      "loss": 1.2595,
      "step": 1000
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.1577868852459017e-05,
      "loss": 1.1804,
      "step": 1500
    },
    {
      "epoch": 2.0,
      "step": 1952,
      "total_flos": 252549760286720.0,
      "train_loss": 1.2798178625888512,
      "train_runtime": 94771.4204,
      "train_samples_per_second": 31.636,
      "train_steps_per_second": 0.021
    }
  ],
  "max_steps": 1952,
  "num_train_epochs": 2,
  "total_flos": 252549760286720.0,
  "trial_name": null,
  "trial_params": null
}