pentoai / vectory Goto Github PK

Bug

It is not possible to select any metadata column as a label in the visualizations.

Description

Vectory currently has a test coverage of 43%. Let's push to stay above 60% with good tests.

Currently, some of the most important areas we are missing tests for are:

• CLI: 0% coverage
• Indices: 10% coverage
• Demo: 34% coverage
• Elasticsearch utils: 21% coverage

Wanted Solution

Add thorough tests for the specific areas mentiones above.

Alternatives

• Adding tests for other sections of the code.

Additional Context

No response.

Description

Currently, there are no PR and issue templates. These contribute to correct documentation and communication of new features or bug reports.

Wanted Solution

Add issue and PR templates.

Describe the bug
When an experiment Query is done with train_dataset=None, it returns all the existing experiments instead of the ones that don't have a train dataset.

To Reproduce
Query the ExperimentModel with train_dataset=None and see the results

from vectory.db.models import ExperimentModel, Query

experiments = Query(ExperimentModel).get(train_dataset=None)
print(experiments)

Expected behavior
It should only return the experiments that have None on the train dataset.

Description
Currently Vectory supports Python 3.7, 3.8, 3.9, 3.10, would be nice if it also supported Python 3.11.

Wanted solution
Add Python 3.11 to automated tests.