modelkit 0.0.4

Machine learning lib.

modelkit

Python framework for production ML systems.

---

modelkit is a Python framework to maintain and run machine learning (ML) code in production environments.

The key features are:

• type-safe Models' inputs and outputs can be validated by pydantic
• composable Models are composable: they can depend on other models.
• organized Store and share your models as regular Python packages.
• extensible Models can rely on arbitrary supporting configurations files called assets hosted on local or cloud object stores
• testable Models carry their own unit test cases, and unit testing fixtures are available for pytest
• fast to code Models can be served in a single CLI call using fastapi
• fast Models' predictions can be batched for speed
• async Models support async and synchronous prediction functions

Installation

Install with pip:

pip install modelkit

Documentation

Refer to the documentation for more information.