Zhongliang Guo

Backdoor Attacks for LLMs with Weak-To-Strong Knowledge Distillation