Vector data types and similarity search

Hydra supports vector data types and similarity search, implemented with pgvector. For usage details, please see the README for pgvector.

Enabling the extension

You must enable the extension before using it. Note that if you create additional databases (with CREATE DATABASE) you will need to enable it in each database.

CREATE EXTENSION vector;

Use with Hydra Columnar

Hydra Columnar implicitly supports vector data types and operations. However, it does not support approximate nearest neighbor search since this requires the use of the pgvector index, ivfflat. If use of ivfflat is desired, this feature can be used on heap tables.

Current version

Hydra's current version of pgvector is 0.4.4.

Last updated