feat: native vector search support - #76
Open
rebasedming wants to merge 2 commits into
Open
Conversation
rebasedming
requested review from
isaacvando,
mdashti,
philippemnoel and
stuhood
as code owners
July 25, 2026 17:25
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## paradedb-am #76 +/- ##
===============================================
+ Coverage 89.81% 90.32% +0.51%
===============================================
Files 17 18 +1
Lines 1276 1375 +99
Branches 263 275 +12
===============================================
+ Hits 1146 1242 +96
- Misses 74 75 +1
- Partials 56 58 +2
Flags with carried forward coverage won't be shown. Click here to find out more.
🚀 New features to boost your workflow:
|
| """Top-K vector search over a BM25 index. | ||
|
|
||
| Requires a ParadeDB build with vector-in-bm25 support (pg_search branch | ||
| ``mvp/vector-search``). The ORDER BY metric must match the index opclass |
| """Integration tests for vector search inside BM25 indexes. | ||
|
|
||
| Vector-in-bm25 support is not yet in a released pg_search (it lives on branch | ||
| ``mvp/vector-search``), so every test here is gated on the server actually |
Comment on lines
+99
to
+100
| - [RAG](examples/rag/rag.py) | ||
| - [Vector Search](examples/vector_search/vector_search.py) |
Member
There was a problem hiding this comment.
Suggested change
| - [RAG](examples/rag/rag.py) | |
| - [Vector Search](examples/vector_search/vector_search.py) | |
| - [Vector Search](examples/vector_search/vector_search.py) | |
| - [RAG](examples/rag/rag.py) |
|
|
||
| ## Vector Search | ||
|
|
||
| ParadeDB indexes pgvector `vector` columns directly inside BM25 indexes, so no pgvector ORM library is needed. Declare a `vector(n)` column with the built-in `Vector` type, add it to the BM25 index with `VectorField` and a distance metric, and order by the matching distance function: |
Member
There was a problem hiding this comment.
See my comment on the EF Core PR
rebasedming
force-pushed
the
vector-search
branch
from
July 27, 2026 15:36
601f297 to
f9a62c2
Compare
rebasedming
force-pushed
the
vector-search
branch
from
July 27, 2026 16:00
f9a62c2 to
d907016
Compare
rebasedming
force-pushed
the
vector-search
branch
from
July 27, 2026 18:51
d907016 to
c1fd7d4
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Implements paradedb/paradedb#5685: native vector search support with no pgvector-python dependency.
Vector(n)column type with serialization, reflection, and Alembic autogenerate round-trip supportl2_distance/cosine_distance/inner_productexpressions (<->,<=>,<#>)VectorField(col, metric=...)for listing vector columns with their opclass (vector_l2_opsdefault) in a bm25 indexexamples/vector_search; integration tests are capability-gated until a release ships vector-in-bm25Why
ParadeDB indexes pgvector columns inside the bm25 index, so users should get the full query surface from this package alone.
Tests
165 unit tests pass; integration suite passes against 0.23.0 (7 vector tests skip via capability gate) and all 7 pass against a dev pg_search 0.25.0 build.