Baidu's open OCR model digitizes all 270 pages of an 1899 Merck manual on one GPU
MaziyarPanahi · x · 2026-10-29
Developer Maziyar Panahi ran Baidu's new open-source OCR model over the 1899 Merck Manual of the Materia Medica (public domain scan from the Wellcome Collection via Internet Archive) — all 270 pages in a single request, on one GPU. Weights are on Hugging Face.
The model reproduced the book's original layout. Fun detail: the cocaine entry lists morphine as an antidote. The author notes these are 1899 doses, not medical advice.
More from Fun
- Nat Lambert Mocks Teased 'Gemini 4 Carbon' as a Graspy Preemptive Move — natolambert · 2026-10-29
- All-AI SEO playbook fails: agency earns zero backlinks from AI content pipeline — lilyraynyc · 2026-10-29
- "Recursive Sundar Improvement": the Pichai crown meme circulating in AI circles — DynamicWebPaige · 2026-10-29
- Nat Lambert calls out Google: don't announce Gemini scores nobody can try — natolambert · 2026-10-29
- "@X please give us an AI slop button": SEO expert calls for slop filtering — lilyraynyc · 2026-10-29
- "Biology seems overpowered": the carbon-based-to-silicon-based meme resurfaces on launch day — ZeroStateReflex · 2026-10-29