Entropy-Gradient Grounding: Training-Free Evidence Retrieval in Vision-Language Models

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Gröpl, Marcel, Jung, Jaewoo, Kim, Seungryong, Pollefeys, Marc, Hong, Sunghwan
Format:	Preprint
Published:	2026
Subjects:	Computer Vision and Pattern Recognition Computation and Language
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Internet

https://arxiv.org/abs/2604.08456