hyperparameter-search/0.00005/trainer_state.json

{
  "best_metric": 16.4909,
  "best_model_checkpoint": "/local1/hfs/gs_stuff/ft-wmt14/0-00005/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": 2.3125038146972656,
      "learning_rate": 0.0,
      "loss": 2.6169,
      "step": 5000
    },
    {
      "epoch": 0.01774311168046785,
      "eval_bleu": 16.4909,
      "eval_gen_len": 28.2457,
      "eval_loss": 2.060305118560791,
      "eval_runtime": 244.1086,
      "eval_samples_per_second": 12.29,
      "eval_steps_per_second": 1.536,
      "step": 5000
    },
    {
      "epoch": 0.01774311168046785,
      "step": 5000,
      "total_flos": 1786854363365376.0,
      "train_loss": 2.616867578125,
      "train_runtime": 931.4391,
      "train_samples_per_second": 85.889,
      "train_steps_per_second": 5.368
    }
  ],
  "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
}