Manage models

Iterative Studio provides a model registry that you can access by clicking on the Models tab. Your project's dvc.yaml files are used to identify ML models and specially formatted Git tags are used to identify model versions and stage assignments.

To quickly start tracking your models in the Iterative Studio model registry:

• Click on Add a project to connect Iterative Studio to your ML project's Git repository.

• In your model training environment, install DVCLive:

  pip install dvclive

• Copy your DVC Studio token and configure your model training environment to use the token:

  dvc config --global studio.token ***

• Use the DVCLive log_artifact() method in your model training code:

  from dvclive import Live
  with Live(save_dvc_exp=True) as live:
    live.log_artifact("model.pt", type="model", name="mymodel")

• Run the training job:

  python train.py

• Once the training completes, commit and push the resultant dvc.yaml file to your Git repository remote.

• The model will get added to the model registry, and you can click on the model's name to open its details page.

More ways to add models

Iterative Studio offers more ways to add models to the model registry - you can:

• edit dvc.yaml directly and add your model to artifacts section, or
• add models from the Iterative Studio interface.

Find the complete tutorial here.

Manage model lifecycle

After adding a model, you can perform the following actions to manage its lifecycle:

• Register new versions of the models (like v0.0.1 and v1.0.0)
• Assign stages to model versions (e.g. dev, testing, prod)
• Use models: download them or set up CI/CD to publish or deploy models

For more details, check out the model registry user guide.