What is paper-notes?

This is a personal note-taking repository where Andrej Karpathy (a well-known AI researcher) shares rough thoughts and summaries about academic papers he's reading. It's essentially a public thinking-out-loud space rather than a polished resource. The repo contains informal jottings, the kind of notes you might scribble while reading a research paper to capture the main ideas, interesting techniques, or questions that come up. It's not a structured database or a complete learning guide, it's more like someone's notebook that happens to be on GitHub where others can peek inside. Who would find this useful? Researchers and students working in AI and machine learning might browse through to see what papers were on Karpathy's radar and how he thinks about them. It's less about learning from the notes themselves and more about getting a window into how an experienced researcher approaches new papers, what they highlight, what questions they ask, what connections they make. The informal tone and incomplete nature actually make it feel more authentic than a polished tutorial would be. The repository description itself ("Random notes? There should be a better place for this I think...") suggests the creator sees this as somewhat temporary or ad-hoc. It's the kind of project someone maintains casually rather than as a primary contribution, which fits the label in the description that says it's "likely a short-term repo." Despite that caveat, it has attracted over 700 stars, which shows there's genuine interest in seeing how other people think through research papers, even if those thoughts are rough and incomplete.