voidful

voidful / hung-yi-lee-skill

Public

一個 Karpathy 風格的 LLM wiki + skill，用於以李宏毅老師的教學風格回答 AI 相關問題。

19
0
100% credibility
Found Apr 09, 2026 at 19 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

A persistent markdown knowledge base that compiles Hung-Yi Lee's YouTube AI lectures, transcripts, and references into a searchable wiki for answering questions in his distinctive teaching style.

How It Works

1
🔍 Discover the collection

You stumble upon a special collection that gathers Professor Hung-Yi Lee's AI lessons from his videos into one easy spot.

2
🛠️ Prepare your notebook

You set up a simple folder on your computer to store all the lesson notes and get the basic tools ready.

3
📋 Collect video list

You pull together the full list of his AI talks so you know what's available.

4
📥 Grab the talk scripts

You download the written transcripts from his videos, capturing every word he says like having personal class notes.

5
📚 Build your wiki book

You organize all the notes into a searchable wiki that grows smarter over time.

6
🔍 Ask AI questions

You search the wiki for any AI topic and build quick summaries to understand deeply.

🎓 Learn like in class

You get answers explained just like Professor Lee teaches: starting with intuition, then the inner workings, and common pitfalls, feeling like you're in his lecture hall.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 19 to 19 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is hung-yi-lee-skill?

This Python project builds a Karpathy-style LLM wiki and skill from Hung-Yi Lee's AI lectures, distilling 478 YouTube videos, transcripts, and topics into a 916-node knowledge graph with 10 concept communities. It lets you sync channel metadata, cache subtitles, compile a markdown wiki, search transcripts, and query the graph via CLI commands like `python scripts/hungyi_kb.py graph query "attention mechanism"`. Drop the SKILL.md into your AI coding assistant to get answers in Lee's voice: intuition first, black box overview, then mechanism and pitfalls.

Why is it gaining traction?

Unlike flat RAG or persona prompts, it traces every response to real transcripts and extracted graph edges (EXTRACTED vs INFERRED), surfacing cross-topic links like those in Karpathy's github llm-council or intro to llm karpathy videos. Developers hook into interactive graph.html viz or OpenAI agent YAML for persistent, structured queries—beats generic ChatGPT cosplay with god nodes like "語言模型" connecting ML fundamentals. Niche appeal mirrors Karpathy github ai repos: runnable pedagogy over raw knowledge dumps.

Who should use this?

ML engineers debugging transformers or agents, referencing hung-yi-lee explanations during karpathy llm youtube binges. Taiwan-based devs or speech/audio specialists querying "語音模型" with graph paths. LLM builders needing a karpathy llm knowledge base template for custom wikis.

Verdict

Grab it if Lee's lectures fuel your workflow—CLI and graph exports deliver immediate value despite 19 stars and 1.0% credibility score signaling early-stage maturity. Docs shine with examples; run `compile` post-install for a solid wiki, but expect manual transcript fetches for full coverage.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.