When Preference Labels Fall Short: Aligning Diffusion Models from Real Data 事件

When Preference Labels Fall Short: Aligning Diffusion Models from Real Data arXiv:2605.19839v2 Announce Type: replace Abstract: Preference alignment aims to guide generative models by learning from comparisons between preferred and non-preferred samples. In practice, most existing approaches rely on preference pairs constructed from model-generated images. Such supervision is inherently relative and can be ambiguous when both samples exhibit artifacts or limited visual quality, making it diffic