trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.650887573964497,
  "eval_steps": 500,
  "global_step": 3,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.5,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 2.0382,
      "step": 1
    },
    {
      "epoch": 0.5,
      "eval_loss": 1.9825799465179443,
      "eval_runtime": 660.1781,
      "eval_samples_per_second": 1.645,
      "eval_steps_per_second": 0.412,
      "step": 1
    },
    {
      "epoch": 1.65,
      "eval_loss": 1.9785882234573364,
      "eval_runtime": 659.635,
      "eval_samples_per_second": 1.646,
      "eval_steps_per_second": 0.412,
      "step": 3
    },
    {
      "epoch": 1.65,
      "step": 3,
      "total_flos": 1143998697701376.0,
      "train_loss": 2.036311705907186,
      "train_runtime": 5409.4181,
      "train_samples_per_second": 1.124,
      "train_steps_per_second": 0.001
    }
  ],
  "logging_steps": 5,
  "max_steps": 3,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 1143998697701376.0,
  "trial_name": null,
  "trial_params": null
}