双曲视觉语言模型到底有多“双曲”？研究质疑曲率的实际作用

View PDF HTML (experimental)

Abstract:Hyperbolic vision-language models (VLMs) have emerged as a rapidly growing area of research, reporting strong zero-shot performance alongside an interpretable embedding space in which radius orders representations by generality. But how hyperbolic are hyperbolic VLMs in practice? We investigate this at the exponential map, the single operation through which curvature enters an otherwise Euclidean network. We show that across MERU, HyCoCLIP, ARGENT, and PHyCLIP, the exponential map is well approximated by the identity function, i.e., pairwise distances and angles in the pre-map Euclidean space barely change under projection, and removing the map at inference leaves evaluation scores unchanged. Motivated by this observation, we train Euclidean VLMs on CC3M from scratch under the same hierarchy losses. Over roughly the final 90% of the training steps, the training statistics are nearly indistinguishable from those of their hyperbolic counterparts, and the resulting models match the evaluation scores and recover the same radial distributions. Our findings establish Euclidean space as a sufficient substrate for hierarchical vision-language learning as it is currently practiced, and motivate new paradigms in which hyperbolic geometry can deliver on its theoretical promise. Code is available at this https URL.

Subjects:	Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
Cite as:	arXiv:2611.13384 [cs.CV]
	(or arXiv:2611.13384v1 [cs.CV] for this version)
	https://doi.org/10.48550/arXiv.2611.13384 arXiv-issued DOI via DataCite (pending registration)
Journal reference:	Proceedings of the NeurIPS Workshop on Symmetry and Geometry in Neural Representations (NeurReps). PMLR, 2026

Submission history

From: Partha Das [view email]
[v1] Tue, 10 Nov 2026 00:01:33 UTC (917 KB)