Skip to end of metadata
Go to start of metadata

You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 4 Current »

What’s New:

ModelOp Center v2.5 contains a powerful set of new features to increase AI governance and visibility, as well as streamline further model risk processes.

Orchestrate:

  • MLC Creation UX Enhancements: intuitive, drop-down for adding ModelOp Center-specific processing into model life cycle (MLC) designs, as well as additional simplified out-of-the-box MLC templates

  • Custom Metadata from Ticketing Systems: ability to pull in custom metadata from Jira and ServiceNow tickets to be persisted with the Model metadata in ModelOp Center

  • Runtime Selection: more flexible runtime selection and associated orchestration

Monitor & Visualize

  • Microsoft PowerBI Integration: integration with Microsoft PowerBI to power custom dashboarding using ModelOp Center data

  • Business Value Tracking (beta): example Business Value Tracking monitors with integration into Tableau Dashboards for executive reporting

  • UI Editor for Monitoring thresholds: built-in editor for defining monitoring thresholds, allowing for users to enter simple metric checks or complex, rules-driven thresholds based on monitoring metrics and model metadata.

  • OOTB Monitoring improvements: minor improvements to OOTB monitors for error handling & versioning

Govern:

  • Automated Model Documentation: ability to auto-generate documentation with model-specific metadata, tests, and other information, using a base document template

  • Model Risk Process Automation: powerful model life cycle automations to orchestrate the key steps in the model risk and compliance processes

General:

  • UI – Runtimes Enhancements: simplified and intuitive UI for viewing the current status and troubleshooting model execution environments (runtimes)

  • Performance Improvements: substantial performance improvements across the ModelOp Center UI

  • Python 3.8 Support in Runtime: standard, supported Python 3.8 base images for the ModelOp Runtime

  • Iterative dev/test: ability for customers to test models in the ModelOp Runtime before official registration and management

  • Runtime Logs Viewing: enhanced live view into runtime logs for troubleshooting

Specific Details:

 Click here to expand...

Minor Enhancements:

Add all groups from SCCS, without filtering, to the list of admin user groups

Tableau: Expose Runtime and Runtime Type

Enhancement to check open sessions with current OAuth2 libraries

Enhancement to add IntrospectionEndpoint to JWT tokens

Enhancement to add KeyCloak integration with ModelOp

Enhancement to load groups from configuration file, potentially coming from SCCS

Enhancement to create a new endpoint at gateway that returns OAuth2Client info for the current supported integrations such as PowerBI , Tableau etc.

Enhancement to add serialVersionUID field to relevant objects that implement Serializable

Enhancement to add missing fields to summary endpoints

Enhancement to support for the status and ticket in the notification regardless of the ticketing system

Enhancement to out-of-the-box MLC: Batch Deployment should only Remove other snapshots of the same model

Enhancement to consider names/types/tags from targetRuntime on GetCompatibleEngine MLC service layer

Enhancement to apply type and parameterizedTypes to delegates part 1

Enhancement to allow Sagemaker Transform Delegate to use https S3 Url

Enhancement to optimize the MLC Delegate Code to Improve Reuse and Documentation

Enhancement to allow for pulling custom JIRA values into a Model Snapshot

Enhancement to update Model Review Notification details that are written to JIRA

Enhancement to consider names/types/tags from targetRuntime on GetCompatibleEngine MLC service layer

Enhancement to increase retry time for Ticketing Notifications

Enhancement to provide an External Service for ServiceNOW

Enhancement to add deployed model ID to the IO message sent from the runtime

Enhancement to extended schema validation should include checking specialValues

Bug Fixes:

Fix for: Tableau Plugin Error when no DeployableModels exist

Fix for: Backend endpoints: MOC doesn't redirect to requested URL after OAuth login

Fix for: Services crash without logstash appender

Fix to: Respond with 401 in secured mode for non-browser requests arriving WITHOUT tokens

Fix to: Jobs without available runtimes hang in CREATED state

Fix to: Backend endpoints: MOC doesn't redirect to requested URL after OAuth login

Fix to: Update Jupyter with GW OAuth2 Authentication flow

Fix to: Camunda Infinite Loop Occurs When External Task Bpmn Error Propagation Fails

Fix to: User is logged out very frequently from Camunda

Fix to: Engine tags don't display unless a model is deployed

  • No labels