Author name: Haozhe Wang, Cong Wei, Weiming Ren, Jiaming Liu, Fangzhen Lin, Wenhu Chen

arXiv:2604.11626v2 Announce Type: replace
Abstract: Most reward models for visual generation reduce rich human judgments to a single unexplained score, discarding the reasoning that underlies preference. We show that teaching reward models to produce …

arXiv:2604.11626v1 Announce Type: cross
Abstract: Most reward models for visual generation reduce rich human judgments to a single unexplained score, discarding the reasoning that underlies preference. We show that teaching reward models to produce ex…