trainer_state.json

{
  "best_metric": 2.176905632019043,
  "best_model_checkpoint": "out/checkpoint-2",
  "epoch": 0.0013069759843162881,
  "eval_steps": 1,
  "global_step": 2,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 3.709977626800537,
      "learning_rate": 0.0002,
      "loss": 2.7778,
      "step": 1
    },
    {
      "epoch": 0.0,
      "eval_loss": 2.3843367099761963,
      "eval_runtime": 6715.6678,
      "eval_samples_per_second": 0.911,
      "eval_steps_per_second": 0.228,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.760042667388916,
      "learning_rate": 0.0002,
      "loss": 2.3824,
      "step": 2
    },
    {
      "epoch": 0.0,
      "eval_loss": 2.176905632019043,
      "eval_runtime": 6717.7383,
      "eval_samples_per_second": 0.911,
      "eval_steps_per_second": 0.228,
      "step": 2
    }
  ],
  "logging_steps": 1,
  "max_steps": 5,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1,
  "total_flos": 701067914379264.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}