Safety Mirage: How Spurious Correlations Undermine VLM Safety Fine-Tuning and Can Be Mitigated by Machine Unlearning
Yiwei Chen · Yuguang Yao · Yihua Zhang · Bingquan Shen · Gaowen Liu · Sijia Liu
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。
Yiwei Chen · Yuguang Yao · Yihua Zhang · Bingquan Shen · Gaowen Liu · Sijia Liu
Xiang Gao · Yuguang Yao · Qi Zhang · Kaiwen Dong · Avinash Baidya · Ruocheng Guo · Hilaf Hasson · Kamalika Das
Large language models (LLMs) often struggle to use tools reliably in domain-specific settings, where APIs may be idiosyncratic, under-documented, or tailored to private workflows. This highlights the need for effective adaptation to task-specific tools. We propose RIMRULE, a neuro-symbolic approa...
Dawei Li · Yuguang Yao · Zhen Tan · huan liu · Ruocheng Guo
Reward-guided search methods have demonstrated strong potential in enhancing tool-using agents by effectively guiding sampling and exploration over complex action spaces. As a core design, those search methods utilize process reward models (PRMs) to provide step-level rewards, enabling more fine-...