Model registry overview

With the Neptune model registry, you can store your models in a central location and collaboratively manage their lifecycle. This enables stakeholders and applications to access your models for various purposes, such as audits, tests, or stage management.

You can use Neptune to manage your models in the following ways:

• Create models and track generic model metadata, such as the model signature and validation dataset.
• Create versions of your models:
  • Log parameters and other metadata that might change from one version to another.
  • Track or store model binaries.
  • Track the performance of specific model versions.
• Manage stage transitions of model versions, using four available stages.
• Query and download stored model files and metadata, through the app or API.

See example model registry in Neptune 

How it works

Similar to the Run object, you initialize and work with a Model object. This puts the data in the Models section of your project.

Then, for each model, you create versions as you refine the model, using ModelVersion objects. This lets you store and track specific model metadata separately from your experimentation runs.

In the example screenshot below, you can see a few versions of a text classification model.

Example use cases

The text classification use cases in our examples repository showcase the model registry.

Open fastText example in Colab  Open Keras example in Colab 

Browse examples on GitHub 

API reference pages

• init_model() function
• Model class
• init_model_version() function
• ModelVersion class

Related

Runs vs Model registry - When to use which