OCR scanned or image-only corpora with vision-language models, in three phases.
/vlm-ocrscdenney/open-science-skills
/vlm-ocr