public inbox for [email protected]
help / color / mirror / Atom feedFrom: pgvector via PostgreSQL Announce <[email protected]>
To: PostgreSQL Announce <[email protected]>
Subject: pgvector 0.8.0 Released!
Date: Mon, 11 Nov 2024 14:18:45 +0000
Message-ID: <[email protected]> (raw)
[pgvector](https://github.com/pgvector/pgvector/), an open-source PostgreSQL extension that provides vector similarity search capabilities, has released [v0.8.0](https://github.com/pgvector/pgvector/releases/tag/v0.8.0). This release includes features that improve query performance and usability when using filters (e.g. the `WHERE` clause), and performance improvements for searching and building [HNSW](https://github.com/pgvector/pgvector?tab=readme-ov-file#hnsw) indexes.
This latest version of pgvector has a variety of improvements for filtering. This includes an update to how PostgreSQL estimates when to scan a approximate nearest neighbor (ANN) index like HNSW and IVFFlat, which could lead PostgreSQL to select a B-tree or other index that more efficiently executes the query. If you can achieve the same query performance without using an ANN index, this is usually preferable as it lets you achieve 100% recall, or high relevancy searches.
Additionally, this pgvector release adds [iterative index scans](https://github.com/pgvector/pgvector?tab=readme-ov-file#iterative-index-scans), which is a technique to prevent "overfiltering" or not returning enough results to satisfy the conditions of a query. You can enable iterative scanning with the `hnsw.iterative_scan` and `ivfflat.iterative_scan` parameters for HNSW and IVFFlat indexes respectively, and if an initial index scan doesn't satisfy the query conditions, pgvector will continue to search the index until it hits a configurable threshold (`hnsw.max_scan_tuples` and `ivfflat.max_probes`).
For more information, please see the [CHANGELOG for 0.8.0](https://github.com/pgvector/pgvector/blob/master/CHANGELOG.md#080-2024-10-30):
[https://github.com/pgvector/pgvector/blob/master/CHANGELOG.md#080-2024-10-30](https://github.com/pgvector/pgvector/blob/master/CHANGELOG.md#080-2024-10-30)
For more information about pgvector, including how to get started, please visit the [project repository on GitHub](https://github.com/pgvector/pgvector):
[https://github.com/pgvector/pgvector](https://github.com/pgvector/pgvector)
reply
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients using the --to and --cc options:
reply via email
To: [email protected]
Cc: [email protected], [email protected]
Subject: Re: pgvector 0.8.0 Released!
In-Reply-To: <[email protected]>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by agora; see mirroring instructions
for how to clone and mirror all data and code used for this inbox