2 papers
cs.CV2026
The Alignment Illusion in Multimodal Large Language Models
Hong-Han Wang, Yuntao Wang, Hu Ding
Layer-wise visual-text similarity in Multimodal Large Language Models (MLLMs) is widely interpreted as evidence that the language model progressively integrates visual content into…
cs.CV2026
MIRROR: Aligning Semantic Relations from Language to Image via Gromov--Wasserstein
Hong-Han Wang, Yuntao Wang, Hu Ding
Multimodal Large Language Models (MLLMs) inherit rich relational priors from their language backbones, yet often fail when asked to apply these relationships in visual contexts. We…