trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 8000,
  "global_step": 6048,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.5,
      "learning_rate": 0.0,
      "loss": 0.0,
      "step": 1000
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.0,
      "loss": 0.0,
      "step": 2000
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.0,
      "loss": 0.0,
      "step": 3000
    },
    {
      "epoch": 1.98,
      "learning_rate": 0.0,
      "loss": 0.0,
      "step": 4000
    },
    {
      "epoch": 2.48,
      "learning_rate": 0.0,
      "loss": 0.0,
      "step": 5000
    },
    {
      "epoch": 2.98,
      "learning_rate": 0.0,
      "loss": 0.0,
      "step": 6000
    },
    {
      "epoch": 3.0,
      "step": 6048,
      "total_flos": 2247973873385472.0,
      "train_loss": 0.0,
      "train_runtime": 538.1748,
      "train_samples_per_second": 22.47,
      "train_steps_per_second": 11.238
    }
  ],
  "logging_steps": 1000,
  "max_steps": 6048,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 2247973873385472.0,
  "trial_name": null,
  "trial_params": null
}