Skip to content

SDV 2.0: The download_demo function should return the right metadata #2956

Description

@npatki

Problem Description

Starting from SDV 2.0, we will have an enhanced metadata that includes the ordinal sdtype as well as range information.

The download_demo function, which returns demo datasets, should return the updated metadata.

Expected behavior

Starting from SDV 2.0, we expect the download_demo function to return the updated metadata (stored as metadata_v2.json in our bucket).

If metadata_v2.json cannot be found, then it's ok to return v1 (aka metadata.json), but provide a warning.

from sdv.datasets.demo import download_demo

data, metadata = download_demo(
    modality='multi_table',
    dataset_name='fake_hotels'
)
Warning: An updated metadata V2 is not available for this dataset so the V1 metadata was returned.
You should be able to model and sample with the V1 metadata, but please report this issue to the DataCebo.

Additional context

No backwards compatibility is needed since this is for SDV 2.0.

Merge into a feature branch for SDV 2.0, as there will be other issues that are also needed for this version.

Metadata

Metadata

Assignees

No one assigned

    Labels

    Type

    No type

    Projects

    No projects

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions