Cross-modal representational alignment between unimodal language and vision models increases with model capability when captions are clean and info...
论文概览
- arXiv ID:
2611.12869 - 发布机构 / 来源:Technical University of Munich
- 作者:Julia Herbst, Simon Roschmann, Quentin Bouniot et al.
- 社区热度:7 Upvotes
核心摘要(Abstract / AI Summary)
Cross-modal representational alignment between unimodal language and vision models increases with model capability when captions are clean and informative, but remains below same-modality alignment and can exceed explicitly aligned vision-language models.
