zju-llms/foundations-of-llms

This repository is a free, open-access textbook on the foundations of large language models (LLMs), the AI systems that power tools like ChatGPT and Claude. Written in Chinese by a research team, it is designed for readers who want to understand how LLMs work at a technical level, from the basics up to cutting-edge methods. The book covers six main areas: how traditional and modern language models work at their core, how the architecture of large AI models has evolved over time, how to write effective prompts (instructions given to an AI model) to get better results, how to efficiently fine-tune (customize and adapt) a pre-existing model without retraining it from scratch, how to edit or correct specific pieces of knowledge stored inside a model, and how retrieval-augmented generation works, a technique where the model pulls in relevant information from an external source before responding. Each chapter is downloadable as a PDF, and the repo also provides curated lists of academic papers linked to each chapter's topic. The team plans monthly updates to keep the content current. This material is aimed at students, researchers, and practitioners who want a rigorous yet readable guide to LLM technology. No specific prior coding skill is assumed, but some familiarity with machine learning concepts would help.