TY  - RPRT
TI  - VLMs Need Words: Vision Language Models Ignore Visual Detail In Favor of Semantic Anchors
AU  - Haz Sameen Shahgir
AU  - Xiaofu Chen
AU  - Yu Fu
AU  - Erfan Shayegani
AU  - Nael Abu-Ghazaleh
AU  - Yova Kementchedjhieva
AU  - Yue Dong
PY  - 2026
UR  - https://arxiv.org/abs/2604.02486
ID  - 2604.02486
ER  -