datawhalechina/happy-llm: 'Build Large Models From Scratch' Chinese Tutorial Reaches 27,702 Stars

GitHub·low signal

happy-llm is a structured, ground-up curriculum for building LLMs from scratch in Chinese, covering agents, LLMs, and RAG via Jupyter notebooks — companion to the successful hello-agents series from the same org (28K stars, covered last week). The datawhalechina organization is China's most prominent open ML education group and this repo's rapid growth to 27K mirrors the global appetite for deep-build tutorials rather than API-wrapper courses. Tag: llm, rag, agent.