@inproceedings{fang-etal-2026-act,
    title = "Act-Adaptive Margin: Dynamically Calibrating Reward Models for Subjective Ambiguity",
    author = "Fang, Feiteng  and
      Chen, Dingwei  and
      Huang, Xiang  and
      Lin, Ting-En  and
      Wu, Yuchuan  and
      Liu, Xiong  and
      Ye, Jing  and
      Liu, Ziqiang  and
      Zhang, Haonan  and
      Zhu, Liang  and
      Alinejad-Rokny, Hamid  and
      Yang, Min  and
      Li, Yongbin",
    editor = "Liakata, Maria  and
      Moreira, Viviane P.  and
      Zhang, Jiajun  and
      Jurgens, David",
    booktitle = "Proceedings of the 64th Annual Meeting of the {A}ssociation for {C}omputational {L}inguistics (Volume 1: Long Papers)",
    month = jul,
    year = "2026",
    address = "San Diego, California, United States",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/ingest-acl/2026.acl-long.710/",
    pages = "15603--15618",
    ISBN = "979-8-89176-390-6",
    abstract = "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 the Bradley-Terry model faces significant challenges when dealing with ambiguous preferences. To improve reward modeling in subjective tasks, this paper proposes AAM (Act-Adaptive Margin), which enhances reward modeling by dynamically calibrating preference margins using the model{'}s internal parameter knowledge. We design two versions of AAM that efficiently generate contextually-appropriate preference gaps without additional human annotation. This approach fundamentally improves how reward models handle subjective rewards by better integrating generative understanding with preference scoring. To validate AAM{'}s effectiveness in subjective reward modeling, we conduct evaluations on RewardBench, JudgeBench, and challenging role-playing tasks. Results show that AAM significantly improves subjective reward modeling performance, enhancing Bradley-Terry reward models by 2.95{\%} in general tasks and 4.85{\%} in subjective role-playing tasks. Furthermore, reward models trained with AAM can help downstream alignment tasks achieve better results. Our test results show that applying rewards generated by AAM-Augmented RM to preference learning techniques (e.g., GRPO) achieves state-of-the-art results on CharacterEval and Charm. The code and dataset will be released upon acceptance."
}