Event ·Multi-source

开源课程 AI Engineering from Scratch:20 阶段手写算法学 AI 工程

开源项目 AI Engineering from Scratch 在 GitHub 获得 59.8K 星,共 20 个阶段、以 Python 为主,主张在用任何框架之前先用纯数学手写算法。课程采用 Motto → Problem → Concept → Build It → Use It → Ship It 六段结构,例如 Phase 10 的分词器一课从词级、字符级到 BPE 子词切分讲演化逻辑,并包含 DeepSeek-V3 架构走读、Native Sparse Attention、speculative decoding 等论文级主题。每节课产出一个可复用 artifact(prompt、skill、agent 或 MCP server),全部学完累计 523 个作品。项目还提供六卷本 EPUB/PDF 书籍、可装入 Claude Code 的 Agent 导师模式,以及覆盖 Anthropic Claude 认证等 5 条备考路径、505 道练习题,支持包括中文在内的 12 种语言。

Current read

59.8K 星的开源课程,不调 API,从线性代数手写到 DeepSeek-V3 架构和多智能体,学完还能攒下 523 个作品,特别适合想搞懂底层原理的人。

11 sources71° AI heatLast updated 2026/9/28 08:09:52

Evidence

11 sources
Related source 2lmarena.ai
View source
Related source 5Gergely Orosz
View source
Related source 6The Information
View source
Related source 7Tibor Blaho
View source
Related source 8rohanpaul_ai
View source
Related source 9arXiv: DeepSeek
View source
Related source 10kimmonismus
View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.