datawhalechina/happy-llm

Covers foundational NLP concepts through practical LLM implementation, with structured chapters progressing from Transformer architecture and attention mechanisms to hands-on model building using PyTorch. Includes end-to-end training workflows (pretraining, supervised fine-tuning, LoRA optimization) and applications like RAG and agent systems, with downloadable pretrained 215M parameter models and companion code implementations.

27,292 stars. Actively maintained with 1 commit in the last 30 days.