trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.5,
  "eval_steps": 500,
  "global_step": 2500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.5,
      "grad_norm": 1.8131240606307983,
      "learning_rate": 4.166666666666667e-05,
      "loss": 2.8997,
      "step": 500
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7417197227478027,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 2.8097,
      "step": 1000
    },
    {
      "epoch": 1.5,
      "grad_norm": 2.1385185718536377,
      "learning_rate": 2.5e-05,
      "loss": 1.9273,
      "step": 1500
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.487093448638916,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 1.9087,
      "step": 2000
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.675701856613159,
      "learning_rate": 8.333333333333334e-06,
      "loss": 1.0467,
      "step": 2500
    }
  ],
  "logging_steps": 500,
  "max_steps": 3000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 4.016131522265088e+16,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}