Fetching the paper…

VisFocus: Prompt-Guided Vision Encoders for OCR-Free Dense Document Understanding · Around