A
licenseA
qualityD
maintenanceEnables vision LLMs to read PDFs by automatically detecting text corruption and switching between text extraction and image rendering modes, while preserving reading order and filtering unnecessary images to prevent token overflow.
3
MIT