Medical LLM fundamentals, open-sourced chapter by chapter

What it does

This repository is a living Chinese-language textbook on medical large language models. It splits the field into five parts and eighteen PDF chapters, starting with Transformer basics and pre-training, moving through clinical decision support and drug discovery, and ending with hallucinations, privacy, and medical ethics. The authors bill it as a monthly updated, “readable, rigorous, and deep” guide for anyone entering the space.

The interesting bit

The project uses GitHub less as a codebase and more as a publishing house: the README is a table of contents, issues serve as errata, and every chapter carries a curated paper list. That turns the repo into a syllabus with a built-in reading list rather than a static download.

Key highlights

• Eighteen chapters covering fundamentals, vertical training strategies, evaluation benchmarks, and clinical applications.
• Each chapter includes a curated paper list to track the latest research.
• Authors commit to monthly updates and actively solicit corrections via GitHub issues.
• Content is organized into five thematic parts, from history to future outlook on medical AGI.
• Drawn from the authors’ own research perspective at ZJU.

Caveats

• No code or notebooks: the deliverables are PDFs only.
• Two chapters in the table of contents link to the same PDF file, and several filenames contain spaces that may break raw links.
• The text is Chinese-only and explicitly framed as the authors’ own exploration, not peer-reviewed consensus.

Verdict A solid syllabus-level bookmark for Chinese-speaking developers or researchers mapping the medical LLM landscape. If you need runnable training pipelines or English-language documentation, look elsewhere.