v1.63
Release notes
1.63.0
💥 Breaking changes
- Updated endpoint path for updating Deployment information from
api.<base_url>/workspaces/{workspaceId}/deployments/{deploymentId}/descriptiontoapi.<base_url>/workspaces/{workspaceId}/deployments/{deploymentId}/information - For private cloud installations, a mandatory upgrade of Kserve CRD to v0.19.0 is needed before performing upgrade.
✨ New features
-
Task center
- The task center provides a centralized hub to view all tasks for your account. Use tasks to track required actions, such as reviewing approvals, handling alerts, and managing control status changes.
-
Tags for Deployment and use cases
- It is now possible to create tags in a workspace that can be used to label and filter Deployments and use cases.
-
Updates Kserve version to v0.19
- Light GBM updated to 4.6.0
- Serving images for Xgboost, Sklearn, Catboost and Lightgbm now support python versions 3.10-3.12
🛠️ Improvements
- Include user ID's in organization users and workspace members page
- Improved empty state for triggered guardrails graph when no guardrails are applied to a Deployment
- Successful request at inference endpoints (/predict, /explain, /chat/completions, /completions, /embeddings, /custom/*) now return status code 200 instead of 201
🐛 Bug fixes
- Fix missing trace durations
- Fix pending members being shown for adding as owners for use case.