probly

Welcome to the documentation for probly, a Python library for uncertainty quantification in machine learning. probly’s name, coming from probably this is the right answer, reflects its core functionality: Providing an easy-to-use interface for allowing practitioners to incorporate uncertainty into their machine learning workflows.

probly’s Philosophy

1. Library-agnostic: probly is designed to work with any machine learning framework, that you can use it with. Let it be PyTorch, Flax, JAX, TensorFlow, scikit-learn, and more.

2. Model-agnostic: probly is designed to work with any machine learning model and pipelines that you are already using: from simple linear regression to complex transformer-based models, and anything in between.

3. Ante-Hoc and Post-Hoc: You either bring your own model and probly will transform it or you can use the built-in tools to build your own models with uncertainty directly in mind.

Table of Contents

• Introduction
• Installation
• User Guide
• Examples and Tutorials
• API Reference
• References and Further Reading
• FAQ and Troubleshooting