last-checkpoint/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.2962962962962963,
  "eval_steps": 100,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.15,
      "learning_rate": 0.0,
      "loss": 8.5238,
      "step": 100
    },
    {
      "epoch": 0.15,
      "eval_runtime": 788.2037,
      "eval_samples_per_second": 1.903,
      "eval_steps_per_second": 0.239,
      "step": 100
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0,
      "loss": 8.4542,
      "step": 200
    },
    {
      "epoch": 0.3,
      "eval_runtime": 788.181,
      "eval_samples_per_second": 1.903,
      "eval_steps_per_second": 0.239,
      "step": 200
    }
  ],
  "logging_steps": 100,
  "max_steps": 2025,
  "num_train_epochs": 3,
  "save_steps": 100,
  "total_flos": 4.1648079765504e+16,
  "trial_name": null,
  "trial_params": null
}