hyperparameter-search/0.0005/trainer_state.json

{
  "best_metric": 20.3555,
  "best_model_checkpoint": "/local1/hfs/gs_stuff/ft-wmt14/0-0005/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": 1.4693390130996704,
      "learning_rate": 0.0,
      "loss": 2.4399,
      "step": 5000
    },
    {
      "epoch": 0.01774311168046785,
      "eval_bleu": 20.3555,
      "eval_gen_len": 27.4887,
      "eval_loss": 1.8954609632492065,
      "eval_runtime": 241.2918,
      "eval_samples_per_second": 12.433,
      "eval_steps_per_second": 1.554,
      "step": 5000
    },
    {
      "epoch": 0.01774311168046785,
      "step": 5000,
      "total_flos": 1786854363365376.0,
      "train_loss": 2.4398966796875,
      "train_runtime": 932.582,
      "train_samples_per_second": 85.783,
      "train_steps_per_second": 5.361
    }
  ],
  "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
}