hyperparameter-search/0.0001/trainer_state.json

{
  "best_metric": 17.7258,
  "best_model_checkpoint": "/local1/hfs/gs_stuff/ft-wmt14/0-0001/checkpoint-5000",
  "epoch": 0.01774311168046785,
  "eval_steps": 5000,
  "global_step": 5000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01774311168046785,
      "grad_norm": 2.2946131229400635,
      "learning_rate": 0.0,
      "loss": 2.5485,
      "step": 5000
    },
    {
      "epoch": 0.01774311168046785,
      "eval_bleu": 17.7258,
      "eval_gen_len": 28.078,
      "eval_loss": 2.0016322135925293,
      "eval_runtime": 247.2837,
      "eval_samples_per_second": 12.132,
      "eval_steps_per_second": 1.516,
      "step": 5000
    },
    {
      "epoch": 0.01774311168046785,
      "step": 5000,
      "total_flos": 1786854363365376.0,
      "train_loss": 2.54846015625,
      "train_runtime": 933.7101,
      "train_samples_per_second": 85.68,
      "train_steps_per_second": 5.355
    }
  ],
  "logging_steps": 5000,
  "max_steps": 5000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 5000,
  "total_flos": 1786854363365376.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}