Back to glossary

Model Registry

A central repository for managing the lifecycle of machine learning models, including versioning, stage tracking, and annotations.

Categoryai

Overview

A model registry acts as a 'catalog' for all the models an organization has built. It helps teams track which version of a model is in production, which is in staging, and which is archived.

Key Features

  • Versioning: Automatically tracks iterations of a model.
  • Stage Management: Moving models through 'Development', 'Staging', and 'Production'.
  • Metadata Storage: Storing performance metrics, training parameters, and who created the model.

Popular Tools

  • MLflow Model Registry
  • Weights & Biases
  • Azure ML Model Management