Chinese mask language generation replacement words are too slow

QData / TextAttack

TextAttack 🐙 is a Python framework for adversarial attacks, data augmentation, and model training in NLP https://textattack.readthedocs.io/en/master/

https://textattack.readthedocs.io/en/master/

MIT License

2.98k stars 397 forks source link

Chinese mask language generation replacement words are too slow #777

Closed liuyuyan2717 closed 8 months ago

liuyuyan2717 commented 8 months ago

I conducted a Chinese adversarial sample attack experiment and found that it took too long and the GPU was not fully utilized, which was due to directly calling pipline.

liuyuyan2717 commented 8 months ago

The problem has been resolved， see #778