AI工程(影印版)
AI工程(影印版)
Chip Huyen
出版时间:2025年04月
页数:509
“这本书为构建生成式AI系统的关键方面提供了全面且结构清晰的指南。对于任何希望在企业中大规模推广AI的专业人士来说,这是一本必读之作。”
——Vittorio Cretella
前宝洁(P&G)与玛氏(Mars)全球首席信息官
“Chip Huyen对生成式AI有着深刻的理解。她是一位出色的教师和作家,其工作成果在帮助团队将AI推向生产环境方面发挥了重要作用。凭借她的深厚专业知识,本书为构建生产环境中的生成式AI应用提供了全面综合指南。”
——Luke Metz
ChatGPT共创人,OpenAI前研究经理

基础模型开启了众多全新的AI应用场景,降低了构建AI产品的门槛。这将AI从一门晦涩难懂的学科转变为一种强大的开发工具,即使是没有AI经验的人也能使用。
在这本通俗易懂的指南中,作者Chip Huyen探讨了AI工程的概念:利用现成的基础模型构建应用的过程。AI应用开发者将学习如何驾驭人工智能领域,包括模型、数据集、评估基准以及看似无穷无尽的应用模式。书中还介绍了一个用于开发AI应用并高效部署的实用框架。
● 理解AI工程的概念及其与传统机器学习工程的区别
● 学习AI应用的开发过程,了解每个步骤中的挑战及其解决方法
● 探索各种模型适配技术,包括提示工程、RAG、微调、智能体以及数据集工程,并理解其原理及应用场景
● 分析基础模型在延迟和成本方面的瓶颈,学习克服这些问题的方法
● 根据需求选择合适的模型、评估指标、数据、开发模式
  1. Preface
  2. 1. Introduction to Building AI Applications with Foundation Models
  3. The Rise of AI Engineering
  4. Foundation Model Use Cases
  5. Planning AI Applications
  6. The AI Engineering Stack
  7. Summary
  8. 2. Understanding Foundation Models
  9. Training Data
  10. Modeling
  11. Post-Training
  12. Sampling
  13. Summary
  14. 3. Evaluation Methodology
  15. Challenges of Evaluating Foundation Models
  16. Understanding Language Modeling Metrics
  17. Exact Evaluation
  18. AI as a Judge
  19. Ranking Models with Comparative Evaluation
  20. Summary
  21. 4. Evaluate AI Systems
  22. Evaluation Criteria
  23. Model Selection
  24. Design Your Evaluation Pipeline
  25. Summary
  26. 5. Prompt Engineering
  27. Introduction to Prompting
  28. Prompt Engineering Best Practices
  29. Defensive Prompt Engineering
  30. Summary
  31. 6. RAG and Agents
  32. RAG
  33. Agents
  34. Memory
  35. Summary
  36. 7. Finetuning
  37. Finetuning Overview
  38. When to Finetune
  39. Memory Bottlenecks
  40. Finetuning Techniques
  41. Summary
  42. 8. Dataset Engineering
  43. Data Curation
  44. Data Augmentation and Synthesis
  45. Data Processing
  46. Summary
  47. 9. Inference Optimization
  48. Understanding Inference Optimization
  49. Inference Optimization
  50. Summary
  51. 10. AI Engineering Architecture and User Feedback
  52. AI Engineering Architecture
  53. User Feedback
  54. Summary
  55. Epilogue
  56. Index
书名:AI工程(影印版)
作者:Chip Huyen
国内出版社:东南大学出版社
出版时间:2025年04月
页数:509
书号:978-7-5766-2004-7
原版书书名:AI Engineering
原版书出版商:O'Reilly Media
Chip Huyen
 
Chip Huyen是实时机器学习平台Claypot AI的联合创始人。凭借在 NVIDIA、Netflix和Snorkel Al的工作,她帮助了一些世界上最大的组织开发和部署机器学习系统。本书是Chip根据她在斯坦福大学开设的课程“机器学习系统设计”(CS329S)的讲义撰写的。

Chip Huyen致力于人工智能、数据、叙事的交叉研究。此前,她曾供职于Snorkel AI和NVIDIA,创立了一家AI基础设施初创公司(已被收购),并在斯坦福大学教授机器学习系统设计课程。她的著作Designing Machine Learning Systems(O’Reilly出版)已被翻译成10多种语言。
 
 
The animal on the cover of AI Engineering is an Omani owl (Strix butleri), a so-called “earless owl” native to Oman, Iran, and the UAE.
An owl collected in 1878 was dubbed Strix butleri after its discoverer, ornithologist Colonel Edward Arthur Butler. This bird was commonly known as Hume’s owl and it was thought to be widespread throughout the Middle East.
In 2013, a previously unknown species of owl was discovered in Oman and given the name Strix omanensis, the Omani owl. No physical specimen was collected, but the owl was described from photographs and sound recordings. Then, in 2015, an analysis of the Strix butleri holotype (the original specimen found in 1878) revealed that the owl was actually the same as Strix omanensis, and distinct from the more common owl found throughout the Middle East. Following naming conventions, the species kept the original name Strix butleri and the more common owl was given the name Strix hadorami, the desert owl.
The Omani owl has a pale and dark gray face and orange eyes. Its upperparts are a dark grayish brown and its underparts are pale gray with narrow dark streaks. It’s a medium-sized owl with a round head and no ear tufts. As a relatively new discovery, ornithologists are still researching the owl’s behavior, ecology, and distribution.
The IUCN conservation status of the Omani owl is data deficient. Many of the animals on O’Reilly covers are endangered; all of them are important to the world.
购买选项
定价:189.00元
书号:978-7-5766-2004-7
出版社:东南大学出版社