Model training in the SDK

The Snorkel Flow SDK is the most flexible way to integrate with any modeling pipeline. The SDK provides functionality for training custom models by using training datasets that you developed in Snorkel Flow. You can then register your models and predictions back to the platform.

In addition, the SDK comes with a model zoo to get you off the ground quickly with pre-configured models using state-of-the-art architectures and ML libraries.

To find the SDK documentation:

• In Snorkel Flow, click Documentation, and then click Python SDK at the bottom of the left-side menu. The SDK documentation will appear in a new window.

  

• In the SDK documentation, follow the path Tutorials -> Train a custom model using a Snorkel Flow training dataset -> Model zoo on the left-side table of contents.