PID Control-Based Self-Healing to Improve the Robustness of Large Language Models

Zhuotong Chen, Zihu Wang, Yifan Yang +2

Despite the effectiveness of deep neural networks in numerous natural language processing applications, recent findings have exposed the vulnerability of these language models when…