Writing logs to ./outputs/2024-04-29-22-14-05-476099/train_log.txt.
Wrote original training args to ./outputs/2024-04-29-22-14-05-476099/training_args.json.
***** Running training *****
  Num examples = 12000
  Num epochs = 3
  Num clean epochs = 3
  Instantaneous batch size per device = 8
  Total train batch size (w. parallel, distributed & accumulation) = 8
  Gradient accumulation steps = 1
  Total optimization steps = 4500
==========================================================
Epoch 1
Running clean epoch 1/3
Train accuracy: 84.42%
Eval accuracy: 91.80%
Best score found. Saved model to ./outputs/2024-04-29-22-14-05-476099/best_model/
==========================================================
Epoch 2
Running clean epoch 2/3
Train accuracy: 95.33%
Eval accuracy: 94.57%
Best score found. Saved model to ./outputs/2024-04-29-22-14-05-476099/best_model/
==========================================================
Epoch 3
Running clean epoch 3/3
Train accuracy: 98.50%
Eval accuracy: 95.43%
Best score found. Saved model to ./outputs/2024-04-29-22-14-05-476099/best_model/
Wrote README to ./outputs/2024-04-29-22-14-05-476099/README.md.