[{"_1":2,"_3":-5,"_4":-5},"loaderData",{"_5":6,"_7":8},"actionData","errors","root",{"_29":-5,"_39":-5,"_40":41,"_42":43,"_44":45,"_46":47,"_48":49,"_50":51,"_52":53,"_54":55,"_56":57,"_58":59},"routes/_store+/_readers+/email_author.$slug/route",{"_9":10,"_11":-5,"_12":-5},"book",{"_13":14,"_15":16,"_17":18,"_19":20,"_21":22,"_23":24},"currentUser","formToken","id","Z2lkOi8vbGVhbnB1Yi9TdG9yZTo6Qm9va1Byb2R1Y3QvODQ5MjU","title","无损智能","state","published","contactAuthorMethod","email","aboutText"," 一个统一的问题：智能如何用最少的比特，保住最多的能力？

大模型的讨论常被简化为\"参数堆多大\"。这本书换一条主线——压缩与信息放置：模型如何从海量文本里压出规律，如何在没见过的样本上不露馅，如何用上下文、检索、工具和记忆，把\"记不住的\"放在\"取得到的\"地方。

18 章，一条完整的设计主线

理论起点（1–4 章）：熵、最小描述长度、能力级无损、预测下一个 token——建立\"压缩是理解智能的强主线\"这一分析视角
机制核心（5–10 章）：Transformer 信息路由、规模法则、表示叠加、泛化、推理、压缩的代价
工程工具箱（11–16 章）：量化/剪枝/蒸馏、LoRA、MoE、信息放置六仓模型、RAG、智能体
前沿收口（17–18 章）：世界模型，与\"最小描述、最大能力\"的系统设计观

这本书的四个不同

190 张原创图解：章首框架图、机制示意图、反例卡片、写实实物图——每章都\"看得见\"
真实数字：正文内嵌可运行 Python；量化数字标注\"截至 2026-10 报告值\"
失败也讲：每章有压缩失败案例与公开评测，显式暴露方法边界
前沿追踪：每章有\"2024–26 后继\"小节——从 Mamba 混合架构、SAE 电路追踪、DoRA，到 MCP 与 Genie 3

适合谁读

想搞懂大模型\"为什么这样设计\"的算法工程师与研究者；需要判断\"该压哪、该放哪\"的 AI 应用架构师；希望参与前沿问题而非只背结论的自学者。有基础 Python 即可跟上，不要求推导公式。

","authors",[25],{"_13":26,"_27":28,"_29":30},"Z2lkOi8vbGVhbnB1Yi9TdG9yZTo6UHJvZHVjdEF1dGhvci85MDU4Mg","name","kejia shao","user",{"_31":32},"avatar",{"_33":34,"_35":36,"_37":38},"url","https://d39qdlcrvnra4b.cloudfront.net/avatars/3700660/medium/avatar.jpg?1775443236","styleName","medium","exists",true,"shoppingCartId","shoppingCart",{"_13":72,"_73":74,"_75":38,"_76":77,"_78":79,"_80":81,"_82":38},"locale","en","readerTheme","light","constants",{"_70":71},"publishers",{"_66":67,"_68":69},"learnerOrganizations",[],"needsLocaleCookieSync",false,"announcements",{"_60":61,"_62":-5},"sentryEnvironment","production","sentryRelease","8ea4d29fe24d3b671122eac2eb77ac48950f1a27","announcement_store",{"_63":53,"_64":65},"announcement_authors","flag","text","

Kick off your book project, get started with GhostAI, get better at marketing, or spend the day doing all three! Free live workshops on Zoom. Saturday, June 27, 2026.

\n","companies",[],"universities",[],"paypalClientId","AX53Cwz8jR6rVVzzZiZNWWXl0dGagi-GZrjG-S3Ua14E5nrKMn7DxDQ730O4R8LusLPp49ALn6aRVaEy","no-cart","count",0,"free","updatedAt","","paymentGateway","STRIPE","cartItems",[],"loading"]

Email the Author