checkpoint-2000/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 7.0033,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 1.02,
      "learning_rate": 9.960000000000001e-06,
      "loss": 0.5615,
      "step": 500
    },
    {
      "epoch": 3.02,
      "learning_rate": 1e-05,
      "loss": 0.3273,
      "step": 1000
    },
    {
      "epoch": 3.02,
      "eval_loss": 0.4225325584411621,
      "eval_runtime": 1489.9228,
      "eval_samples_per_second": 3.09,
      "eval_steps_per_second": 0.193,
      "eval_wer": 20.82533081951615,
      "step": 1000
    },
    {
      "epoch": 5.01,
      "learning_rate": 1e-05,
      "loss": 0.1859,
      "step": 1500
    },
    {
      "epoch": 7.0,
      "learning_rate": 1e-05,
      "loss": 0.0923,
      "step": 2000
    },
    {
      "epoch": 7.0,
      "eval_loss": 0.46434369683265686,
      "eval_runtime": 1504.5618,
      "eval_samples_per_second": 3.06,
      "eval_steps_per_second": 0.191,
      "eval_wer": 21.22002361108664,
      "step": 2000
    }
  ],
  "max_steps": 10000,
  "num_train_epochs": 9223372036854775807,
  "total_flos": 4.14105131483136e+18,
  "trial_name": null,
  "trial_params": null
}