TY  - RPRT
TI  - Blind to Position, Biased in Language: Probing Mid-Layer Representational Bias in Vision-Language Encoders for Zero-Shot Language-Grounded Spatial Understanding
AU  - Na Min An
AU  - Inha Kang
AU  - Minhyun Lee
AU  - Hyunjung Shim
PY  - 2026
UR  - https://arxiv.org/abs/2509.23098
ID  - 2509.23098
ER  -