Skip to content
BENIBIX Learn

Model Registry

AI⚙️ MLOps & DeploymentModel Deployment

Also known as ML Model Registry

A list of each model version, its score, and which one is live.

Think of it like

Think of it like a library desk. Each book has a number, and you know which ones are out.

Example

Lucy's registry shows version 4 of the price model in production and version 5 in testing. When version 5 passes, she promotes it, and the old one is kept in case she needs it back.

Why it matters

Everyone can see which model is live and how it was made, and rolling back is one step.

Definition

A model registry is a list that keeps every trained model version and its details. For each version it notes who made it, which data and code were used, its test scores and its stage. The stage may be testing, live or archived. Ship tools read the registry to know which version to serve. Moving a model to live, or back to an old version, is a change in the registry that can be signed off and logged. It links training to serving.
Each version has a score and a stage. One is in production, one is in testing.

Common mistake

Keeping models as loose files with names like final_v2_new. Soon nobody knows which one is live.

Key takeaways

  1. 1
    Every version kept. With its details.
  2. 2
    Stages. Testing, production, archived.
  3. 3
    Easy roll back. One step.

Compare