作者位次 #2/3ICML 2026Poster· Social Aspects / Security
Unlearning Isn’t Forgetting: Revealing Hidden Leakage in Class Unlearning Evaluations
Ali Ebrahimpour-Boroojeny · Yian Wang · Hari Sundaram
首次获取该候选人的完整档案及 StartLux 评估共 20 积分；每位用户单独解锁，之后查看免费。主动刷新 20 积分。
实时查找电子邮箱，费用不超过20积分；未找到邮箱不扣费。
Ali Ebrahimpour-Boroojeny · Yian Wang · Hari Sundaram
Yian Wang · Yuen Chen · Agam Goyal · Hari Sundaram
Large language models (LLMs) frequently generate toxic content, posing significant risks for safe deployment. Current mitigation strategies often degrade generation quality or require costly human annotation. We propose CausalDetox, a framework that identifies and intervenes on the specific atten...