作者位次 #9/13ACL 2026Poster· Natural Language Processing
Act-Adaptive Margin: Dynamically Calibrating Reward Models for Subjective Ambiguity
Feiteng Fang · Dingwei Chen · Xiang Huang · Ting-En Lin · Yuchuan Wu · Xiong Liu · Jing Ye · Ziqiang Liu · Haonan Zhang · Liang Zhu · Hamid Alinejad-Rokny · Min Yang · Yongbin Li
摘要
Currently, most reinforcement learning tasks focus on domains like mathematics and programming, where verification is relatively straightforward. However, in subjective tasks such as role-playing, alignment techniques struggle to make progress, primarily because subjective reward modeling using t...