Models

SageMaker will bundle artifacts and images to provide a modular and deployable representation of a model for inference. Once models have been created in MLSpace you can leverage those models in inference endpoints. The models page within a project will display a list of all models associated with a project as well as allow project members to create new models and take actions against existing models.

Creating Models

Creating models requires a providing model metadata, some of which is optional. A model name and inference image is required. The inference image location can be selected from a set of built-in SageMaker inference images or you can optionally specify a custom path to an ECR image. In order to use a custom image your administrator must have configured MLSpace to allow access ECR and the image being used. Model artifacts and container host name are optional as well as any environment key-value pairs which may be required by the particular model.

Model Details

The model details view displays the metadta associated with existing models

Deleting Models

Model owners, project owners, and admins can delete models.