Releases: wagtail/wagtail-vector-index
Releases · wagtail/wagtail-vector-index
Wagtail Vector Index v0.10.0
What's Changed
- Add LiteLLM support by @tomusher in #64
- Refactor vector indexes in to single type by @tomusher in #65
- Docs: Update 'Indexing across models' example to include all necessary steps by @ababic in #70
- Resolved bug in index.query() that prevented content from similar documents to be included i then LLM query by @berry in #69
Upgrade Considerations
While this package is pre-1.0 we will not be following a deprecation schedule for API changes. This means there's a few things you'll need to tweak to continue using this package after upgrading:
- The
wagtail_vector_index.index
andwagtail_vector_index.backends
modules have been merged and renamed towagtail_vector_index.storage
. - 'Vector Backends' have been renamed to 'Storage Providers' to be clearer about what they do and reduce confusion with the use of the word 'backend'. The setting
WAGTAIL_VECTOR_INDEX_VECTOR_BACKENDS
has been renamed toWAGTAIL_VECTOR_INDEX_STORAGE_PROVIDERS
to match. - The
similar
method on vector indexes has been renamed tofind_similar
.
To update:
- Update any imports as necessary, e.g.
wagtail_vector_index.index.models
becomeswagtail_vector_index.storage.models
- Update your backend configurations, e.g.
wagtail_vector_index.backends.pgvector.backend.PgvectorBackend
becomeswagtail_vector_index.storage.pgvector.PgvectorStorageProvider
- Replace any uses of the
similar
method withfind_similar
- Replace your
WAGTAIL_VECTOR_INDEX_VECTOR_BACKENDS
config, e.g.:
WAGTAIL_VECTOR_INDEX_STORAGE_PROVIDERS = {
"default": {
"STORAGE_PROVIDER": "wagtail_vector_index.storage.pgvector.backend.PgvectorStorageProvider",
}
}
New Contributors
Full Changelog: v0.9.0...v0.10.0
Wagtail Vector Index v0.9.0
The initial release of Wagtail Vector Index!
We're not quite ready to commit to a final API in a 1.0 release before we've heard from others using the package in the wild. If you get a chance to give it a go, your feedback on the docs and developer APIs would be much appreciated!
A big thanks to @tm-kn, @mgax, @zerolab and @nimasmi for their contributions to this release.