Open tkreuder opened 4 months ago
4725fe3daf
)[!TIP] I can email you next time I complete a pull request if you set up your email here!
I found the following snippets in your repository. I will now analyze these snippets and come up with a plan.
src/vdf_io/import_vdf/qdrant_import.py
✓ https://github.com/AI-Northstar-Tech/vector-io/commit/01405c0d8695fcdffbab2a783bfa29199c2868c5 Edit
Modify src/vdf_io/import_vdf/qdrant_import.py with contents: Update the `vectors_config` dictionary in the `upsert_data` method to correctly structure the vector configuration.
Remove the nested "vector" key and specify the vector configuration directly under the "vectors" key, with the vector column name as the key and the VectorParams
object as the value. Also remove the on_disk
parameter as it is not part of the VectorParams
configuration.
I have finished reviewing the code for completeness. I did not find errors for sweep/qdrant_import_collection_not_working
.
💡 To recreate the pull request edit the issue title or description. Something wrong? Let us know.
This is an automated message generated by Sweep AI.
The issue seems to stem from the collection configuration format used during the import process. Specifically, the vectors_config
setup in the upsert_data
method of qdrant_import.py
expects a dictionary with keys corresponding to vector column names and their configurations. However, your collection's config.json indicates a mismatch in expected structure, particularly under the params
-> vectors
section. To resolve this, ensure the collection configuration passed to self.client.create_collection
within upsert_data
matches Qdrant's expected format. This involves adjusting the vectors_config
dictionary construction to align with your collection's actual vector dimension and distance metric, ensuring it accurately reflects the structure shown in your issue description.
The
"params": {
"vectors": {
"vector": {
"size": 1536,
"distance": "Cosine"
}
},
...
is matching the format for named vectors.
I want to better understand the sequence of your operations. Did you first import a vdf dataset into your qdrant instance, and then try to do a search via the REST API?
Details
the config.json of the collection looks strange to me:
Probably the reason for this issue is:
"vectors": { "vector": { ...
instead ofBranch
No response
Checklist
- [X] Modify `src/vdf_io/import_vdf/qdrant_import.py` ✓ https://github.com/AI-Northstar-Tech/vector-io/commit/01405c0d8695fcdffbab2a783bfa29199c2868c5 [Edit](https://github.com/AI-Northstar-Tech/vector-io/edit/sweep/qdrant_import_collection_not_working/src/vdf_io/import_vdf/qdrant_import.py)