trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.6004319654427646,
  "eval_steps": 500,
  "global_step": 139,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.11,
      "grad_norm": 0.675453782081604,
      "learning_rate": 0.0002,
      "loss": 2.08,
      "step": 25
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3721127510070801,
      "learning_rate": 0.0002,
      "loss": 0.3546,
      "step": 50
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.161994069814682,
      "learning_rate": 0.0002,
      "loss": 0.2689,
      "step": 75
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.08174678683280945,
      "learning_rate": 0.0002,
      "loss": 0.1425,
      "step": 100
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1015535444021225,
      "learning_rate": 0.0002,
      "loss": 0.2211,
      "step": 125
    },
    {
      "epoch": 0.6,
      "step": 139,
      "total_flos": 9030484680671232.0,
      "train_loss": 0.566877440582934,
      "train_runtime": 2411.1952,
      "train_samples_per_second": 0.461,
      "train_steps_per_second": 0.058
    }
  ],
  "logging_steps": 25,
  "max_steps": 139,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 25,
  "total_flos": 9030484680671232.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}