ML//RAG//FAISS

FAISS (Facebook AI Similarity Search) is Meta's open-source library for efficient similarity search on dense vectors. It is the reference implementation for vector search in research, and many production systems use it internally.

FAISS (Facebook AI Similarity Search) is Meta's open-source library for efficient similarity search on dense vectors. It is the reference implementation for vector search in research, and many production systems use it internally.

It supports exact (brute-force), approximate (HNSW, IVF) and GPU-accelerated index types. Product quantization (PQ) compresses vectors to reduce memory drastically, which makes it possible to search billions of vectors on one machine.