Skip to content

Core Classes

Workbench Core Classes

These classes interact with many of the Cloud Platform services and are therefore more complex. They provide additional control and refinement over the AWS ML Pipline. For most use cases the API Classes should be used

Welcome to the Workbench Core Classes

The Core Classes provide low-level APIs for the Workbench package, these classes directly interface with the AWS Sagemaker Pipeline interfaces and have a large number of methods with reasonable complexity.

The API Classes have method pass-through so just call the method on the API Class and voilà it works the same.

ML Pipeline

Artifacts

  • AthenaSource: Manages AWS Data Catalog and Athena
  • FeatureSetCore: Manages AWS Feature Store and Feature Groups
  • ModelCore: Manages the training and deployment of AWS Model Groups and Packages
  • EndpointCore: Manages the deployment and invocations/inference on AWS Endpoints

Transforms

Transforms are a set of classes that transform one type of Artifact to another type. For instance DataToFeatureSet takes a DataSource artifact and creates a FeatureSet artifact.

  • DataLoaders Light: Loads various light/smaller data into AWS Data Catalog and Athena
  • DataLoaders Heavy: Loads heavy/larger data (via Glue) into AWS Data Catalog and Athena
  • DataToFeatures: Transforms a DataSource into a FeatureSet (AWS Feature Store/Group)
  • FeaturesToModel: Trains and deploys an AWS Model Package/Group from a FeatureSet
  • ModelToEndpoint: Manages the provisioning and deployment of a Model Endpoint
  • PandasTransforms:Pandas DataFrame transforms and helper methods.