Adding a new customized ML Runtimethrough the Runtime Catalog
Cloudera AI enables you to add customized ML Runtimes from the Runtime
Catalog window.
Click Runtime Catalog from the
Navigation panel.
Click the Add Runtime button in the upper right
corner.
In the Add Runtime window, enter the url of the Runtime
Docker image you want to upload.
As ML Runtimes are identified based on certain attributes, metadata (such as
Editor, Kernel, Edition, Version, and Maintenance Version) must be unique to add new
Customized Runtimes to a deployment. Customized ML Runtimes must have different
Edition text compared to Cloudera supported versions.
If you want to use a non-default credential, select a docker credential in the
Select Credentials dropdown list.
The Docker server name and the user name of the credential is displayed.
Click Validate.
Cloudera AI will use the provided URL to fetch the Docker
image and validate if it can be used as a customized Runtime.
If the Docker image is successfully validated, Cloudera AI
will display the metadata information of the image. The new customized Runtime will
be visible in the Runtime Catalog and accessible over the different workloads.