Description
The "AI Infra Book" (《深入理解 AI Infra:量化分析与系统设计》), authored by Li Bojie, is an open-source manuscript dedicated to the in-depth understanding of AI infrastructure. This comprehensive resource focuses on a quantitative approach to analyzing and designing systems for Large Language Model (LLM) inference and training. It meticulously derives system designs by considering fundamental hardware constraints and various model architectures.
Read More
The GitHub repository serves as the central hub for this project, offering the complete book text, downloadable PDF versions, and a suite of配套计算工具 (accompanying calculation tools) to aid in practical application and understanding. Additionally, it includes experimental resources, allowing readers to explore and validate the concepts discussed within the book. The content is structured to guide readers through complex topics, from foundational hardware principles to advanced system design considerations.
The project is actively maintained, with recent commits indicating ongoing development and refinement across various sections, including `calculations`, `case-studies`, and `experiments`. It supports multiple languages, with dedicated folders for English (`book-en`), Simplified Chinese (`book`), and Traditional Chinese (`book-zh-tw`), ensuring broad accessibility. The book also features `case-studies` that delve into real-world applications and `experiments` that provide practical insights into runtime and device recovery.
Targeted at researchers, engineers, and students, the "AI Infra Book" aims to bridge the gap between theoretical understanding and practical implementation in the field of AI infrastructure. Its emphasis on quantitative analysis provides a rigorous framework for optimizing AI systems, making it an invaluable resource for anyone looking to deepen their knowledge in this critical area.
- CategoryEducation & Research
- Pricing ModelOpen source
- DeploymentCloud
- PlatformsWeb
AI Infra Book's Core Features
Open-source book manuscript
Quantitative analysis of AI infrastructure
LLM inference system design
LLM training system design
Derivation from hardware constraints
Derivation from model architectures
Full book text available
PDF versions included
Accompanying calculation tools
Experimental resources
Case studies on AI infrastructure
Multi-language support (English, Simplified Chinese, Traditional Chinese)
Apache-2.0 License
Getting Started with AI Infra Book
Clone Repository: Use `git clone https://github.com/bojieli/ai-infra-book.git` to get a local copy of the book and resources.
Explore Book Content: Navigate to the `book`, `book-en`, or `book-zh-tw` directories for the full text in various languages.
Access PDFs: Look for PDF releases within the repository or on the associated website for offline reading.
Utilize Calculation Tools: Explore the `calculations` directory for tools that support the quantitative analysis presented in the book.
Review Case Studies: Examine the `case-studies` folder for real-world examples and applications.
Run Experiments: Investigate the `experiments` directory to understand and replicate practical system design tests.
Contribute: Refer to the `CITATION.cff` and `LICENSE` files for information on how to cite or contribute to the project.
AI Infra Book's Use Cases
- LLM System Design
- Hardware Constraint Analysis
- Quantitative Performance Evaluation
- AI Infrastructure Research
- Educational Resource
- Practical Experimentation
- Cross-Lingual Learning







