V3.2.1 Release Notes

Owned by Dave Trier
Last updated: Nov 27, 2023
3 min read

Full List of Enhancements and Bug Fixes:

  • UI:

    • Ensemble: added support for model ensembles in the ModelOp UI

    • Generic Model: changed the logo to represent the Generic Model (i.e. a model that does not have any source code attached upon creation)

    • Job Creation Wizard: In the Job Creation wizard, added the job extension to the filename when the file extension is selected from the drop-down

    • Threshold (DMN) Editor:

      • When adding a new set of thresholds (DMN), added support to automatically generate a default name for the DMN

      • Added ability to resize the DMN editor window

    • Documentation: added support to add a document by URL (as opposed to requiring the upload of a local file)

    • Import Models: updated the error message for end users when the git import fails for an external asset

    • Import Monitors: updated the “import monitor” screen to only allow for imports from a git repository

    • Theme: addressed an issue where the UI theme colors were inadvertently being mixed in the Job details page

    • Model Repository Info: addressed an issue where, upon adding of a git repository to a model, the repository information disappeared if tabs were switched

    • Model Test Results:

      • Added ability to set fixed custom values for the primary axis of a given graph. This includes the ability to set default values for the custom axis values.

      • Addressed an issue when a item in a graph or table contains a long name (for example, if a feature name for a model is very long, the drift graphs would not render in a user-friendly view)

      • Addressed an issue with tables that were automatically sorting the columns alphabetically. The columns in the Table will follow the natural ordering in the Model Test Result json.

      • Addressed an issue where the Summary and Performance tables repeated the test result name

      • Addressed a horizontal overflow issue for tables that are very wide in a model test result

    • Login / User Information: added support for an option to display the AzureAD group name instead of group ID

    • White-labeling ModelOp Center: added support to replace the ModelOp Center logo with a customer-specific logo

  • Tests / Monitors:

    • Model Test Result Visualizations: added support for filtering a table, graph, or chart using a drop-down menu based on keys in the table, graph, or chart

    • Test Result Pass/Fail: added support in the UI for displaying the “reason” message when a test passes

  • MLC:

    • Added support to the out-of-the-box Dashboard MLC to allow for running the dashboard jobs without opening a Jira ticket, if an issue occurs

    • Added a delegate to support Automated Documentation Generation by populating a document template with a specified test result

  • Security:

    • Added support for SAML-based integration, leveraging an Oauth2-SAML approach

    • Upgraded AzureADCustomGroupsConfig and AzureADReactiveCustomGroupsServiceImpl to receive the default WebClient exposed by the ModelOp Gateway

    • Updated WebFlux WebClient configuration making it to use proxy configurations when available

    • Updated the ModelOp Runtime’s opaque token validation mechanism from GET to POST

    • Updated the user logout routine to include ID token as a query param when redirecting the user to logout

    • Addressed an issue during Login where the connection was prematurely closed before a response was provided, leading to a “Whitelabel error page”

  • AWS SageMaker:

    • Added support to un-deploy a SageMaker model directly from the ModelOp UI

    • Addressed an issue where additional imported SageMaker models were not importing the SageMaker historical transform jobs properly

  • General / Other:

    • Jira Integration: provided workaround to address customer Jira instances that contain 500+ custom fields

    • Model Source Files: added support to upload model source files directly via the UI upload and have them categorized as Model_Source files within ModelOp (as opposed to “external assets”)

    • Runtime:

      • Added ability to acquire OAuth2 tokens through a proxy, when a customer has implemented a proxy between the ModelOp runtime and ModelOp core services

      • During model deployment, added support for extended delays for downloading very large assets to the runtime. This typically occurs when a model has very large binary artifacts that need to be loaded onto a runtime.

    • ModelOp Python Packages: Addressed an issue where the moc-python-sdk was adding an additional space to scopes when requesting a token

  • ModelOp Upgrades:

    • Addressed issue for ModelOp 3.1.x upgrades where git sync was failing if a business model or monitor was onboarded in a previous version (before 3.1.X) and the model or monitor had yaml files as assets