vipin wiki

Karpathy Public Work

2 min read

Karpathy Public Work

This hub routes future agents to Karpathy’s public projects, writing, videos, gists, and curated short-form idea links.

How To Use This Hub

Current Corpus Counts

  • blog-rss-item: 10
  • curated-status-link: 70
  • github-gist: 13
  • github-repository: 63
  • personal-homepage: 1
  • youtube-video: 15

Top GitHub Repositories By Stars At Crawl

  • autoresearch - 81258 stars, Research automation / agentic science
  • nanoGPT - 58127 stars, LLM training and inference systems
  • nanochat - 53492 stars, LLM training and inference systems
  • LLM101n - 36913 stars, LLM training and inference systems
  • llm.c - 29911 stars, LLM training and inference systems
  • minGPT - 24376 stars, LLM training and inference systems
  • nn-zero-to-hero - 21842 stars, Neural network fundamentals
  • llama2.c - 19506 stars, LLM training and inference systems
  • llm-council - 18787 stars, Research automation / agentic science
  • micrograd - 15849 stars, Neural network fundamentals
  • char-rnn - 12034 stars, Neural network fundamentals
  • convnetjs - 11160 stars, Neural network fundamentals
  • minbpe - 10482 stars, LLM training and inference systems
  • arxiv-sanity-preserver - 5681 stars, Research tooling and paper workflows
  • neuraltalk2 - 5579 stars, Neural network fundamentals
  • neuraltalk - 5488 stars, Neural network fundamentals
  • build-nanogpt - 5009 stars, LLM training and inference systems
  • ng-video-lecture - 4714 stars, LLM training and inference systems
  • makemore - 3916 stars, Neural network fundamentals
  • reader3 - 3622 stars, LLM training and inference systems

Maintenance Rule For Future Agents

  • Re-crawl before making current claims about repository counts, stars, releases, or new posts.
  • Do not rely on search snippets when an official source surface exists.
  • Keep unlicensed full text/code out of public redistribution; use summaries, hashes, and links unless the source license permits more.
  • Preserve Karpathy’s reusable heuristics as durable wiki synthesis rather than leaving them in chat.

Counterpoints And Gaps

  • This corpus is broad and source-grounded, but public feeds do not guarantee a complete historical archive of every public statement.
  • Curated status links are pointers, not a full text tweet mirror.
  • Star counts, repository metadata, and public profile numbers are time-sensitive and require a fresh crawl before use in current claims.

Graph View

Backlinks