Search code, repositories, users, issues, pull requests...
TinyHNSW is a tiny, simple vector database. It weighs in at a measly few hundred lines of code. It's built on a straightforward (but not fast) implementation of HNSW in Python with minimal dependencies. It has an associated set of tutorials that build up to understanding how HNSW works, and how you can build your own TinyHNSW.
⚠️ The vector database part of this works, but the tutorial is not yet complete. I will update the README checklist below as chapters and code are completed.
⚠️ tinyhnsw is NOT production-quality. Compared to literally any standard implementation, it's slow and probably buggy. If you want to use an Approximate Nearest Neighbor library in a real application, consider something like FAISS.
📝 As part of understanding how HNSW works, the tutorial walks you through how skip lists work and how to implement one. However, this implementation is not particularly robust and only works with integer keys. It's there for teaching purposes, as understanding skip lists will really help understand how HNSW works.
📝 The second part of understanding how HNSW works is understanding how NSWs work. Again, we provide a teaching implementation in this repo, but it's not meant for much more than teaching.
