Introduction to Model Registry

Model Registry on Gemini Enterprise Agent Platform is a central repository where you can manage the lifecycle of your ML models. From Model Registry, you have an overview of your models so you can better organize, track, and train new versions. When you have a model version you would like to deploy, you can assign it to an endpoint directly from the registry, or using aliases, deploy models to an endpoint.

Model Registry supports custom models and all AutoML data types - tabular and image. Model Registry can also support BigQuery ML models. If you have models trained in BigQuery ML, you can register them with Model Registry without needing to export them from BigQuery ML or import them into Model Registry. From the model version details page you can evaluate, deploy to an endpoint, set up batch inference, and view specific model details. Model Registry provides a straightforward and streamlined interface to manage and deploy your best models to production.

There are many valid workflows for working in Model Registry. Follow these guidelines to understand what you can do in Model Registry and at what stage in your model-training journey.

  • Import models to Model Registry.
  • Create new models, assign a model version the default alias, ready for production.
  • Add other aliases or labels to help you manage and organize your models and model versions.
  • Deploy your models to an endpoint for online inference.
  • Run batch inference, and start your model evaluation pipeline.
  • View your model details and view performance metrics from the model details page. To learn more about how to integrate your BigQuery ML models with Gemini Enterprise Agent Platform, see the BigQuery ML documentation.

Search and discover models using Knowledge Catalog

Knowledge Catalog is a platform for storing, managing, and accessing your metadata. Knowledge Catalog provides a way to search for your Agent Platform models across projects and regions.

For more information, see About data catalog management in Knowledge Catalog.

What's next

To get started using Model Registry, see: