Defining Kubeflow Pipeline (KFP) Components with Python Dataclasses

Features

• Dataclass-Driven Component Definition: Define component logic using Python dataclasses, seamlessly translating them into Kubeflow Pipelines (KFP) compatible functions and components.
• KFP Agnostic: Empower developers to design and implement component logic as standard Python code, independent of the KFP framework.

Installation

pip install ml-orchestrator

Note: ml-orchestrator is designed to be lightweight and free of external dependencies, ensuring efficient runtime performance without additional overhead.

Note: ml-orchestrator does not require the kfp package to parse or create Kubeflow components.

Note: To construct kfp pipelines and utilize the components, the kfp package is required.

Usage

please read the documentation