Critical-CoT: A Robust Defense Framework against Reasoning-Level Backdoor Attacks in Large Language Models.

Vu Tuan Truong, Long Bao Le

Browse the full ACL paper archive.