Karpathy's 7 essential GitHub repos: a hands-on path from micrograd to nanochat and autoresearch
huangyun_122 · x · 2026-11-02
A curated walkthrough of Karpathy's 7 standout GitHub repos, forming a complete path from fundamentals to training your own LLM:
- micrograd: 100 lines of Python implementing autodiff and backprop
- nn-zero-to-hero: notebooks for his video course, from a single neuron to GPT
- minbpe: a clean BPE tokenizer showing how text becomes tokens
- nanoGPT: the minimal trainable/finetunable GPT repo
- llm.c: the same training pipeline in pure C and CUDA, no PyTorch
- nanochat: an end-to-end chat model training stack that brings GPT-2-level training down from tens of thousands of dollars to $100
- autoresearch: an agent that edits training code, runs short training runs, keeps improvements, and loops overnight to automate research
More from coding & agent
- TLA-Verus uses LLM agents to prove temporal properties of TLA+ specs — tianyin_xu · 2026-11-02
- Vercel CEO Rauch: own your intelligence, embrace open source and local models — _AustinCalvert_ · 2026-11-02
- AI-generated App Store screenshots containing 'iPhone' trigger Guideline 5.2.5 rejection — tinyfool · 2026-11-02
- The 12GB HackerNews dataset behind Karpathy's HackerTube is freely downloadable — vista8 · 2026-11-02
- ArtCraft rebuilds Adobe's Creative Cloud in pure Rust: 15 free open-source apps on GitHub — anthara_ai · 2026-11-02
- Measure launches as fully open-source Crashlytics alternative with MCP-accessible debugging agent — Arindam_1729 · 2026-11-02