VisFocus: Prompt-Guided Vision Encoders for OCR-Free Dense Document Understanding Paper β’ 2407.12594 β’ Published Jul 17, 2024 β’ 19