last-checkpoint/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 25.641025641025642,
  "global_step": 1000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 12.82,
      "learning_rate": 1.24e-05,
      "loss": 4.0279,
      "step": 500
    },
    {
      "epoch": 12.82,
      "eval_bleu": 49.841,
      "eval_em": 0.0,
      "eval_gen_len": 51.6403,
      "eval_loss": 2.4031472206115723,
      "eval_rm": 0.0,
      "eval_runtime": 135.1177,
      "eval_samples_per_second": 3.086,
      "eval_steps_per_second": 0.392,
      "step": 500
    },
    {
      "epoch": 25.64,
      "learning_rate": 2.4900000000000002e-05,
      "loss": 1.3442,
      "step": 1000
    },
    {
      "epoch": 25.64,
      "eval_bleu": 85.0177,
      "eval_em": 0.0,
      "eval_gen_len": 57.9784,
      "eval_loss": 0.501366138458252,
      "eval_rm": 0.0,
      "eval_runtime": 188.7983,
      "eval_samples_per_second": 2.209,
      "eval_steps_per_second": 0.281,
      "step": 1000
    }
  ],
  "max_steps": 11700,
  "num_train_epochs": 300,
  "total_flos": 810312415883520.0,
  "trial_name": null,
  "trial_params": null
}