trainer_state.json

{
  "best_metric": 14.0723,
  "best_model_checkpoint": "models/sled/output_sled_mt5_base2/checkpoint-2397",
  "epoch": 1.0,
  "global_step": 2397,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.21,
      "learning_rate": 1.3904338153503894e-05,
      "loss": 18.6448,
      "step": 500
    },
    {
      "epoch": 0.42,
      "learning_rate": 2.7808676307007788e-05,
      "loss": 3.8147,
      "step": 1000
    },
    {
      "epoch": 0.63,
      "learning_rate": 4.171301446051168e-05,
      "loss": 2.6096,
      "step": 1500
    },
    {
      "epoch": 0.83,
      "learning_rate": 5.5617352614015575e-05,
      "loss": 2.3758,
      "step": 2000
    },
    {
      "epoch": 1.0,
      "eval_mean_prediction_length_characters": 7800.0,
      "eval_mean_prediction_length_tokens": 2047.0,
      "eval_num_predicted": 1,
      "eval_rouge/geometric_mean": 14.0723,
      "eval_rouge/rouge1": 24.3546,
      "eval_rouge/rouge2": 7.1185,
      "eval_rouge/rougeL": 16.074,
      "eval_rouge/rougeLsum": 21.4321,
      "eval_runtime": 358.1416,
      "eval_samples_per_second": 0.003,
      "eval_steps_per_second": 0.003,
      "step": 2397
    }
  ],
  "max_steps": 35955,
  "num_train_epochs": 15,
  "total_flos": 8.078322612436992e+16,
  "trial_name": null,
  "trial_params": null
}