This repo contains data for the paper "BLINK: Multimodal Large Language Models Can See but Not Perceive"