Karpathy's entire AI course is free on GitHub: 7 repos, 284k stars
alex_verem · x · 2026-11-10
Andrej Karpathy's complete AI learning path is available free as 7 GitHub repos, in recommended order:
- nn-zero-to-hero: video course building neural nets from scratch
- micrograd: tiny autograd engine showing the math of learning
- makemore: character-level autoregressive language model
- minbpe: minimal Byte Pair Encoding (LLM tokenization)
- nanoGPT: train/finetune your own medium-sized GPT
- nanochat: "the best ChatGPT that $100 can buy"
- autoresearch: AI agents running experiments autonomously on a single GPU
The repos total 284,000 GitHub stars and cover everything needed to understand how ChatGPT works, entirely free and open source.
More from coding & agent
- Amazon swaps Sonnet 4.6 for a decision model in ticket labeling, cutting cost by 98% — victormustar · 2026-11-10
- 360-trial benchmark finds code knowledge graphs give coding agents no clear accuracy gain — JeremyCMorgan · 2026-11-10
- Dev claims a $25/month Claude Code Polymarket bot scanned 14,000 wallets in 4 minutes — Aiden_Tech_Ai · 2026-11-10
- Jev Engineering's dedicated micro-decision model vs Flash subagents in Antigravity — Pokeasss · 2026-11-10
- Builder tip: give your custom AI tooling a way to calibrate responses with user feedback — menhguin · 2026-11-10
- Async agents that actually work: watchlists, autonomy levels, and context — menhguin · 2026-11-10