FAISS
Library for efficient similarity search over dense vectors.
Resumen
FAISS is a library of indexes and algorithms for nearest-neighbour search in dense vector sets, including the quantised indexes that make billion-scale search fit in memory. It solves the problem that exact search is linear in the number of vectors, which stops being acceptable well before a corpus stops growing. It exists because it is the reference implementation of the approximate methods, and it is usually what a vector database is built on rather than an alternative to one.
Grafo de tecnologías
Cómo se conecta FAISS con el resto del stack.
Usa y depende de
- Quantization technique Techniques FAISS uses Quantization High ✓ source
Aplicado a
- AI Search application Applications FAISS used for AI Search High ✓ source
Classification
- HNSW algorithm Algorithms HNSW part of FAISS High ✓ source
Herramientas
Software que implementa o se integra con FAISS.