trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.0,
  "eval_steps": 500.0,
  "global_step": 2290,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 2.183406113537118,
      "grad_norm": 35.835044860839844,
      "learning_rate": 1.5e-05,
      "loss": 4.3883,
      "step": 500
    },
    {
      "epoch": 4.366812227074236,
      "grad_norm": 40.83657455444336,
      "learning_rate": 3e-05,
      "loss": 2.5076,
      "step": 1000
    },
    {
      "epoch": 6.550218340611353,
      "grad_norm": 33.478370666503906,
      "learning_rate": 1.8372093023255815e-05,
      "loss": 1.07,
      "step": 1500
    },
    {
      "epoch": 8.733624454148472,
      "grad_norm": 15.102038383483887,
      "learning_rate": 6.744186046511628e-06,
      "loss": 0.2133,
      "step": 2000
    },
    {
      "epoch": 10.0,
      "step": 2290,
      "total_flos": 6.819440674535874e+16,
      "train_loss": 1.7925869004695176,
      "train_runtime": 18266.5412,
      "train_samples_per_second": 64.182,
      "train_steps_per_second": 0.125
    }
  ],
  "logging_steps": 500,
  "max_steps": 2290,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "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": 6.819440674535874e+16,
  "train_batch_size": 512,
  "trial_name": null,
  "trial_params": null
}