Patcher: Post-Hoc Patching of Backdoored Large Language Models 文章

ArXiv CS.AI2026-06-03NEWSen作者: Anjun Gao, Yueyang Quan, Yufei Xia, Zhuqing Liu, Minghong Fang

查看原文 →

Patcher: Post-Hoc Patching of Backdoored Large Language Models · 相关技术

相关技术

gradient-based saliency scores adaptive clustering response-conditioned gradient-based saliency scores Backdoor attacks jailbreak backdoor attacks