trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.932551319648094,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.73,
      "learning_rate": 3.7781036168132945e-05,
      "loss": 0.5315,
      "step": 500
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9261092775944261,
      "eval_loss": 0.21968263387680054,
      "eval_runtime": 262.5074,
      "eval_samples_per_second": 41.553,
      "eval_steps_per_second": 5.196,
      "step": 682
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.5562072336265887e-05,
      "loss": 0.4054,
      "step": 1000
    },
    {
      "epoch": 2.0,
      "eval_accuracy": 0.9733223322332233,
      "eval_loss": 0.09609706699848175,
      "eval_runtime": 262.661,
      "eval_samples_per_second": 41.529,
      "eval_steps_per_second": 5.193,
      "step": 1364
    },
    {
      "epoch": 2.2,
      "learning_rate": 1.3343108504398828e-05,
      "loss": 0.2765,
      "step": 1500
    },
    {
      "epoch": 2.93,
      "learning_rate": 1.1241446725317694e-06,
      "loss": 0.165,
      "step": 2000
    }
  ],
  "max_steps": 2046,
  "num_train_epochs": 3,
  "total_flos": 2.9814448911998976e+16,
  "trial_name": null,
  "trial_params": null
}