We are pleased and proud to announce the General Availability of D2iQ Kaptain 1.2
. This new update includes new features and improvements to the overall user experience, including:
- A new Kaptain Dashboard, for improved workload monitoring and observability
- Self-service dataset mounts in Notebook
- Volume Manager support
- TensorBoard integration
- Improvements to the Kaptain SDK
The new dashboard enables users to visually monitor and observe the resource consumption of Kaptain Workloads, observe the state of those workloads, and easily identify and debug any issues. The dashboard provides RBAC based visibility into the resource information, so users see just the information that is relevant to them.
With self-service dataset mounts, data scientists can mount preconfigured NFS datasets to build or work on their models. This feature eliminates dependence on an Administrator and improves productivity. Volume manager support simplifies volume creation and deletion, while TensorBoard Integration allows users to visually track model training progress, and analyze model fairness and feature importance to deliver unbiased solutions--very important for customers in regulated industries. The integration also allows users to debug issues with architecture or code in individual nodes.
We’ve enhanced the Kaptain SDK to support canary model deployments, provide direct notification of Kaptain Workload Deployment Failure messages, which returns message from the SDK API call to allow immediate problem identification and save time by eliminating the need to parse logs. The Kaptain SDK now offers customizable SDK logging to suppress debug-level messages by default. And customers can now deploy the SDK into Custom Notebook images. Finally, we have decoupled our SDK release cadence from Kaptain and infrastructure components releases, so you can expect to see more frequent and timely patch releases.