faiss

Module for indexing and searching using FAISS.

This module provides classes and configurations for indexing and searching using FAISS in the Lightning IR framework.

FAISS (Facebook AI Similarity Search) is an open-source library that provides the foundational infrastructure for making dense vector retrieval models like DPR fast and scalable in production. Rather than creating the vector embeddings itself, FAISS acts as a highly optimized database engine designed exclusively for storing and searching through massive collections of dense vectors. It uses approximate nearest neighbor algorithms, such as Product Quantization and Hierarchical Navigable Small World graphs, to drastically reduce the time and memory required to find the closest matching vectors to a search query. By intelligently grouping and compressing these mathematical representations, FAISS allows search systems to bypass exhaustive comparisons and deliver near-instant results across billions of documents.

FAISS: The Faiss library

Modules

faiss_indexer	FAISS Indexer for Lightning IR Framework
faiss_searcher	FAISS-based searcher for approximate nearest neighbor retrieval in the Lightning IR framework.