Documentation

Think, Reason, Learn is an open-source Python library for prediction models that show their reasoning. It keeps the shape of decision trees, rule forests and short policies, and asks a language model to reason at each step, so every prediction comes with the questions or rules behind it. It is built by Vela Research, the research arm of Vela Partners, with the University of Oxford.

Start with the installation guide, then the quick start, which fits a small GPTree on six example founder profiles.

Get started

• About

• Installation guide

• Quick start

Reference

• API reference

Project

• Contributing

• Changelog

• License

The papers behind the methods are on the research page.