What’s New?
This release includes major improvements to the way that SDV detects and learns ID columns (like primary keys).
SDV Enterprise is now able to auto-detect Regex formats for ID columns like primary keys. This allows you to seamlessly run SDV and get realistically-formatted synthetic data. To access this feature, simply run the metadata auto-detection using your data. You should be able to see and verify the detected Regex.
SDV Enterprise now supports embedded context within your ID columns too. This is when a portion of the ID value contains a human-readable or meaningful context. For example, if every product ID begins with a 2-character country code followed by a random 5-digit identifier. SDV Enterprise auto-detects embedded context (country code) and ensures that the synthetic data contains realistic context (no invalid countries). For more information, see the ID columns docs.
Additionally, we have made a change to our constraints. If you are an SDV Enterprise customer with the CAG bundle add-on, we recommend reaching out to us if you require a constraint that is not already part of the predefined constraints. Our team will be able to support you with writing a programmable constraint quickly.
Test it out and leave us feedback in Synthetic Data Creation. Happy synthesizing!
These release notes are for SDV Enterprise. If you are an SDV Community user, please check the community release notes on Github.
Release History
0.47.0 (May 27, 2026): This is the initial release of SDV Enterprise 0.47 that contains the Regex auto-detection and embedded context.
0.47.1 (Jun 2, 2026): In this patch release, we’ve fixed a few bugs and made internal updates to constraints. We also now allow the ability for users to load and save constraints as a separate JSON file.
0.47.2 (Jun 9, 2026): In this patch release, we’ve fixed a few bugs and made internal updates to constraints.
0.47.3 (Jun 16, 2026): In this patch release, we’ve added a few utility functions for ease-of-use. In the demo module, the generic save_resource function can be used to download and save any file type. In the constraints module, the load_constraints function can be used to load constraints from a JSON file representation into objects that are ready to be used.
0.47.4 (Jun 23, 2026): In this patch release, we’ve fixed a bug with backwards compatibility and constraints. You should now be able to load in an old, already-fitted synthesizer (that had constraints) and sample synthetic data from it using this newest version.