Why Safeguarded Ships Run Aground? Aligned Large Language Models' Safety Mechanisms Tend to Be Anchored in The Template Region.
Chak Tou Leong, Qingyu Yin, Jian Wang, Wenjie Li
Browse the full ACL paper archive.
Chak Tou Leong, Qingyu Yin, Jian Wang, Wenjie Li
Browse the full ACL paper archive.