trainer_state.json

{
  "best_metric": 5.335667610168457,
  "best_model_checkpoint": "results/checkpoint-3",
  "epoch": 1.5,
  "eval_steps": 3,
  "global_step": 3,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.5,
      "learning_rate": 0.00013333333333333334,
      "loss": 12.207,
      "step": 1
    },
    {
      "epoch": 1.0,
      "learning_rate": 6.666666666666667e-05,
      "loss": 9.0941,
      "step": 2
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.0,
      "loss": 6.9347,
      "step": 3
    },
    {
      "epoch": 1.5,
      "eval_loss": 5.335667610168457,
      "eval_runtime": 17.8301,
      "eval_samples_per_second": 1.122,
      "eval_steps_per_second": 0.168,
      "step": 3
    },
    {
      "epoch": 1.5,
      "step": 3,
      "total_flos": 1506581575649280.0,
      "train_loss": 9.411954879760742,
      "train_runtime": 116.9289,
      "train_samples_per_second": 0.411,
      "train_steps_per_second": 0.026
    }
  ],
  "logging_steps": 1,
  "max_steps": 3,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 3,
  "total_flos": 1506581575649280.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}