checkpoint-46/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.989247311827957,
  "eval_steps": 500,
  "global_step": 46,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.21505376344086022,
      "grad_norm": 0.09814453125,
      "learning_rate": 0.0002,
      "loss": 1.8867,
      "step": 10
    },
    {
      "epoch": 0.43010752688172044,
      "grad_norm": 0.054443359375,
      "learning_rate": 0.0002,
      "loss": 1.6558,
      "step": 20
    },
    {
      "epoch": 0.6451612903225806,
      "grad_norm": 0.04833984375,
      "learning_rate": 0.0002,
      "loss": 1.6351,
      "step": 30
    },
    {
      "epoch": 0.8602150537634409,
      "grad_norm": 0.056884765625,
      "learning_rate": 0.0002,
      "loss": 1.623,
      "step": 40
    },
    {
      "epoch": 0.989247311827957,
      "eval_loss": 1.6275275945663452,
      "eval_runtime": 77.4322,
      "eval_samples_per_second": 4.068,
      "eval_steps_per_second": 0.517,
      "step": 46
    }
  ],
  "logging_steps": 10,
  "max_steps": 138,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 1.2911847482022953e+18,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}