Skip to content

Azure ML integration guide#

You can integrate Neptune with Azure DevOps and Azure Machine Learning services.

On this page, we show how to prepare for using the example in our neptune-ai/examples repo.

Open example on GitHub 

Before you start#

Setting Neptune credentials#

To ensure that the example is logged to your own workspace, you can supply your own Neptune API token and project name.

You can also run the example with the public credentials, but this data is cleaned from Neptune periodically, so the example won't persist.

API token#

The value for the NEPTUNE_API_TOKEN variable is the API token associated with your Neptune account.

  1. In the bottom-left corner of the app, expand your user menu.
  2. Select Get Your API token.

    How to find your Neptune API token

If using service accounts

You can also pass the API token of a service account.

  1. In the top-left corner, expand the workspace menu (accessible to admins).
  2. Select Service accounts.
  3. Copy the token of the account you want to use.

Project name#

In the example .py scripts, you can set your own project name to the NEPTUNE_PROJECT variable.

Your project path consists of the workspace name and project name: workspace-name/project-name.

To copy the path in the Neptune web app:

  1. In the top-right corner, click the settings menu ().
  2. Select Edit project details.

    How to access project details

  3. Find the copy button () next to the project name.

Other configuration#

In the example, we're also using a custom run ID (NEPTUNE_CUSTOM_RUN_ID) to ensure that metadata from different parts of the pipeline are logged to the same run.

Read about custom IDs:

Model registry#

The example scripts make use of the Neptune model registry.

You can check how to register models, create model versions, manage their stage, and query metadata in the model registry docs:

The rest of the example is in the neptune-ai/examples repo, where we host a folder with a sample CI/CD pipeline.

Open example on GitHub