Fetching the paper…

Unlearning Backdoor Attacks for LLMs with Weak-to-Strong Knowledge Distillation · Around