Published Models#

Profiles: Model Creator, Data Scientist

As a model creator or data scientist, the final step in AI Studio is to publish and deploy your model.

This page explains the publish functionality within AI Studio.

Introduction#

After publishing, your model is made available to the enrich pipeline library.

The Published Models tab has three primary screens:

  • Published Models Overview: Overview of all published and unpublished models of a project.

  • Publish Model: Publish a model to the Squirro Pipeline.

  • Deploy Model: Select the pipeline workflow to deploy a model to. This is only available for models that have been published to the Squirro Pipeline.

Published Models Overview#

All published and unpublished models of a project are listed in the Published Models screen, as shown in the screenshot below:

Models Overview

Publish Model#

Clicking the Publish Model button will launch a modal window to configure the publish settings, as shown in the example screenshot below:

Models Overview

To publish a model, several settings must be configured:

  • Publish Model Name: Name the model should appear as in the Pipeline Library.

  • Description: Description of the published model.

  • Show labeled examples… Option to display ground truth examples on which the model was trained in the Pipeline Library.

  • Show validation data… Option to display the validation metrics of the model in the Pipeline Library.

Deploy Model#

You can deploy a model to a pipeline workflow by clicking the Deploy Model icon next to the three-dot menu.

This will launch a modal window to select the pipeline workflow deployment settings, as shown in the example screenshot below:

Models Overview