VLMs Need Words: Vision Language Models Ignore Visual Detail In Favor of Semantic Anchors

Add code
Apr 02, 2026

Share this with someone who'll enjoy it:

View paper onarxiv icon

Share this with someone who'll enjoy it: