What is machine-learning-library?

This repository is a hand-selected collection of machine learning educational material converted into a consistent, searchable format. It contains 923 documents totaling around 11 million tokens: 391 research papers from arXiv, 474 lecture transcripts from courses by Stanford, MIT, Andrej Karpathy, fast.ai, and others, and 58 explainer articles from well-known ML writers. Every document is stored as a Markdown file with structured metadata at the top covering the title, source URL, authors, date, topic tags, and difficulty level. The purpose is to bring scattered material into one place in a form that is easy to search, embed into a vector database, or feed to an AI model. The curation is deliberate: rather than dumping a broad scrape of arXiv or the web, the author selected sources that are considered foundational or widely cited across the field, ranging from introductory neural network concepts through recent 2025 and 2026 papers on topics like reasoning models, sparse attention, and diffusion language models. For human readers, the repository can be opened as an Obsidian knowledge vault. A bundled Obsidian configuration is included, and a topic navigation layer called the atlas provides 17 subject hubs and curated reading paths. You can browse by topic, difficulty level, or content type without knowing which specific papers or lectures cover a given subject. For AI tools and developers, the corpus is designed to work as a retrieval source. An included Python script demonstrates a minimal semantic search setup, and the repository includes instruction files for AI agents explaining how to navigate and cite from the corpus. The consistent frontmatter structure means the entire collection can be filtered or embedded programmatically without custom parsing per source. The research paper section includes 78 papers in full text and 313 more recent ones as abstracts with links to the originals. The lecture transcripts cover courses from Stanford CS224n, CS231n, CS229, MIT 6.S191, Karpathy's YouTube channel, and others.