trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 1000,
  "global_step": 42,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.7142857142857143,
      "grad_norm": 6.295878887176514,
      "learning_rate": 0.0001845190085543795,
      "loss": 4.9016,
      "step": 10
    },
    {
      "epoch": 1.4285714285714286,
      "grad_norm": 4.4954118728637695,
      "learning_rate": 0.00012000256937760445,
      "loss": 2.7019,
      "step": 20
    },
    {
      "epoch": 2.142857142857143,
      "grad_norm": 4.040550231933594,
      "learning_rate": 4.3193525326884435e-05,
      "loss": 2.1165,
      "step": 30
    },
    {
      "epoch": 2.857142857142857,
      "grad_norm": 3.2141640186309814,
      "learning_rate": 1.2949737362087156e-06,
      "loss": 1.5267,
      "step": 40
    },
    {
      "epoch": 3.0,
      "step": 42,
      "total_flos": 266392691539968.0,
      "train_loss": 2.760472910744803,
      "train_runtime": 117.3664,
      "train_samples_per_second": 1.431,
      "train_steps_per_second": 0.358
    }
  ],
  "logging_steps": 10,
  "max_steps": 42,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 266392691539968.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}