How do we store your vectors?

EigenDB stores your vectors in a Hierarchical Navigable Small World (HNSW) graph to perform similarity search with the highest level of efficiency and speed.

Under the hood

Algorithms to store and query vectors in an HNSW graph is implemented in EigenDB through our open-source Go library faissgo. This library offers lightweight Go bindings for Meta’s Faiss, a popular C++ library for efficient similarity search and clustering of dense vectors released by Facebook AI Research (FAIR). This library allows EigenDB to integrate the power of Faiss inside of its Go ecosystem. In addition, the wrapper was made into an open-source library instead of including it in EigenDB, as this allows other developers to easily integrate Faiss into their Go workflows.