Let Geometry GUIDE: Layer-wise Unrolling of Geometric Priors in Multimodal LLMs

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Wang, Chongyu, Huang, Ting, Sun, Chunyu, Ning, Xinyu, Wang, Di, Tang, Hao
Format:	Preprint
Published:	2026
Subjects:	Computer Vision and Pattern Recognition
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Internet

https://arxiv.org/abs/2604.05695