VCR: A Task for Pixel-Level Complex Reasoning in Vision Language Models via Restoring Occluded Text

Devs

VCR: A Task for Pixel-Level Complex Reasoning in Vision Language Models via Restoring Occluded Text | Read Paper on Bytez