作者位次 #2/4ICML 2026Poster· Social Aspects / Safety
Safe Autoregressive Image Generation with Iterative Self-Improving Codebooks
Yunqi Xue · Zhijiang Li · Phil Torr · Jindong Gu
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。
Yunqi Xue · Zhijiang Li · Phil Torr · Jindong Gu
Wenqian Yu · Shuo Chen · Zhijiang Li · Zhipeng Wang · Jindong Gu
Large language models (LLMs) face a critical alignment challenge: balancing safety with helpfulness. Excessive safety can lead to over-refusal, where models reject harmful-looking yet benign queries, severely limiting utility.Existing training-free interventions offer an efficient way to mitigate...