Fetching the paper…

VisOnlyQA: Large Vision Language Models Still Struggle with Visual Perception of Geometric Information · Around