Skip to main content
Version: v0.61.1

Model registry

This page describes the model registration process in MLOps.

Understand model registration and versioning​

H2O MLOps lets you register individual experiments and group them as versions of a registered model to organize a collection of experiments efficiently. Experiments that are imported into MLOps must first be registered before being deployed. When registering an experiment, you can either register it as a new model or register it under an existing model. Selecting the latter option creates a new version of the existing model.

A registered model is a collection of individual model versions. Registered models are used to group registered model versions that are relevant to a specific problem. New experiments and iterations can be registered as updated versions of the model.

A model version has a one-to-one relationship with experiments within a given Project. When you want to proceed with serving your best experiment, you can register that experiment as a model version.

note

When registering an experiment as a model, all data and metadata lineage is maintained.

Register an experiment as a model​

The following steps describe how to register an experiment as a model.

  1. In the main navigation, click Projects.

Projects page

  1. Select the project that contains the experiment you want to register as a model.

  2. On the project page, navigate to the list of experiments associated with the project by clicking the Experiments tab, then click the name of the experiment that you want to register.

Select experiment

  1. In the Experiment Details panel, click the Register as Model button.

  2. MLOps prompts you to select one of the following options:

    • Register as a new model: Register the experiment as a new model. Enter a model name and a description for the model. The experiment is registered as version 1 of the newly created model. The new model is displayed in the Models tab of the project page.

    • Register to an existing model: Register the experiment to an existing model. This option creates a new version of the selected model. If there are no models associated with the project, this option cannot be selected.

Feedback