trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.888888888888889,
  "eval_steps": 27,
  "global_step": 39,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.9629629629629629,
      "grad_norm": 6.899937152862549,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.7641,
      "step": 13
    },
    {
      "epoch": 1.925925925925926,
      "grad_norm": 4.543986797332764,
      "learning_rate": 5.000000000000003e-06,
      "loss": 0.3377,
      "step": 26
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.05286616086959839,
      "eval_runtime": 1.0501,
      "eval_samples_per_second": 12.379,
      "eval_steps_per_second": 0.952,
      "step": 27
    },
    {
      "epoch": 2.888888888888889,
      "grad_norm": 3.7801897525787354,
      "learning_rate": 0.0,
      "loss": 0.1846,
      "step": 39
    },
    {
      "epoch": 2.888888888888889,
      "step": 39,
      "total_flos": 40425900605440.0,
      "train_loss": 0.4288051434052296,
      "train_runtime": 2679.57,
      "train_samples_per_second": 2.896,
      "train_steps_per_second": 0.015
    }
  ],
  "logging_steps": 13,
  "max_steps": 39,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 27,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 40425900605440.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}