checkpoint-46/trainer_state.json

{
  "best_metric": 1.7903296947479248,
  "best_model_checkpoint": "outputs/checkpoint-46",
  "epoch": 0.9932523616734144,
  "eval_steps": 500,
  "global_step": 46,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.4318488529014845,
      "grad_norm": 1.7280967235565186,
      "learning_rate": 4e-05,
      "loss": 2.827,
      "step": 20
    },
    {
      "epoch": 0.863697705802969,
      "grad_norm": 0.49127423763275146,
      "learning_rate": 8e-05,
      "loss": 2.1949,
      "step": 40
    },
    {
      "epoch": 0.9932523616734144,
      "eval_loss": 1.7903296947479248,
      "eval_runtime": 12.0868,
      "eval_samples_per_second": 30.777,
      "eval_steps_per_second": 3.889,
      "step": 46
    }
  ],
  "logging_steps": 20,
  "max_steps": 690,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 15,
  "save_steps": 500,
  "total_flos": 4074622607425536.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}