view article Article How we OCR'ed 30,000 papers using Codex, open OCR models and Jobs 4 days ago β’ 38