arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2610.18331cs.CLcs.LG
收藏

探究多语言语言模型中颜色表征的几何结构

Probing the Geometry of Color Representations in Multilingual Language Models

Rakuten Group, Inc.(乐天集团);Graduate School of Engineering, The University of Tokyo(东京大学大学院工学系研究科)

Pablo Loyola, Edison Marrese-Taylor, Andrés Hoyos-Idrobo

AI总结:

本研究探究多语言语言模型的颜色表征几何,分解为感知对齐与跨语言对称轴,发现感知基础弱、跨语言对称随深度降低,几何特性可预测任务性能,揭示仅文本LM学颜色-语言通用几何,为感知结构保留提供诊断。

摘要:

多语言语言模型是否为颜色学习到了通用的内部几何结构,该结构是基于感知的还是主要基于语言的?我们将颜色表征分解为两个几何轴:(P1)与感知颜色空间的对齐,以及(P2)跨语言对称性。通过分析不同模型和多语言数据集的分层嵌入,我们发现感知基础较弱:与CIE-LAB的对齐在中间层达到峰值,随后向输出层崩溃;跨语言对称性随深度降低,表明后续层编码语言特定结构。功能上,几何特性可预测任务性能:在P1对齐峰值的层,模型在颜色分类任务上达到96.7%的准确率,且P1对齐(相关系数r=0.70)与下游能力显著相关。我们的结果揭示了一种分离现象:仅文本的语言模型学习到通用的颜色-语言几何结构,而非通用的颜色-感知几何结构,为识别语言模型中感知结构的保留位置提供了诊断方法。

英文摘要:

Do multilingual language models learn a universal internal geometry for color, and is it perceptually grounded or primarily linguistic? We decompose color representations into two geometric axes: (P1) alignment to perceptual color space, and (P2) cross-lingual symmetry. Analyzing layerwise embeddings across models and multilingual datasets, we find that perceptual grounding is weak: alignment to CIE-LAB peaks in intermediate layers, then collapses toward output layers. Cross-lingual symmetry degrades with depth, suggesting later layers encode language-specific structure. Functionally, we show that geometric properties predict task performance: models achieve 96.7% accuracy on color classification at layers where P1 alignment peaks, with P1 alignment (r=0.70) significantly correlating with downstream capabilities. Our results reveal a dissociation: text-only LMs learn universal color-language geometry, not universal color-perception geometry, providing diagnostics for where perceptual structure is preserved in language models.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑