This repository hosts the official website and resources for the book "大语言模型" (Large Language Models), authored by Zhao Xin, Li Junyi, Zhou Kun, Tang Tianyi, and Wen Jirong. It serves as a comprehensive Chinese technical reference for large language models, targeting readers with a deep learning background who seek a systematic understanding of LLM technology, from foundational concepts to advanced applications.

How It Works

The book and its associated resources provide a structured overview of LLM technology, building upon the authors' extensive research and practical experience in developing large models like Wenlan and Yulan. It systematically organizes core concepts, algorithms, and models, drawing from seminal papers, code, and academic texts, aiming for clarity and accuracy in explaining the LLM technical framework and roadmap.

Quick Start & Requirements

• Resources: The repository provides links to PDF courseware (PPT slides) for instructors and teaching materials, as well as code snippets used in the book.
• Access: Courseware requires institutional email application for teaching purposes; code is available on GitHub. Teaching videos are hosted on Bilibili.
• Official Website: https://llmbook-zh.github.io/

Highlighted Details

• Comprehensive coverage of LLM fundamentals including pre-training, fine-tuning, alignment, and prompt engineering.
• Includes detailed explanations of model architectures (e.g., Transformer), training processes, and deployment strategies.
• Features advanced topics such as complex reasoning, planning, and agents.
• Offers supplementary courseware and code for educational and practical use.

Maintenance & Community

The authors are actively involved in LLM research and development, with a commitment to updating and refining the book's content. Feedback and suggestions are welcomed via GitHub Issues or direct email to the authors.

Licensing & Compatibility

The book is published by Higher Education Press. While code snippets are available on GitHub, the repository itself does not explicitly state a license for the website content or courseware. Commercial use or redistribution of courseware requires explicit permission.

Limitations & Caveats

The authors acknowledge potential limitations and omissions due to the rapid evolution of LLM technology and their own expertise. The officially published version supersedes any prior electronic drafts, with some application-focused content not included in the final print.