alphageometry

Introduction

This repository contains the weight files for DDAR and AlphaGeometry, two geometry theorem provers introduced in the Nature 2024 paper "Solving Olympiad Geometry without Human Demonstrations."

Architecture

The architecture of the models focuses on solving complex geometry problems without human demonstrations. These models leverage advanced theorem-proving techniques, as detailed in the associated research paper.

Training

The training of DDAR and AlphaGeometry involves learning from geometry problems and solutions, applying machine learning techniques to automate theorem proving. Specific training methodologies and datasets used are outlined in the research paper.

Guide: Running Locally

To run the models locally, follow these steps:

1. Clone the repository from GitHub.
2. Ensure you have the necessary dependencies installed as listed in the repository's documentation.
3. Download the weight files from the repository.
4. Execute the model using your local computational resources or a cloud GPU.

Cloud GPUs are recommended for better performance, such as those provided by AWS, Google Cloud, or Azure.

License

The project is licensed under the Apache 2.0 License, allowing for wide use and modification with proper attribution.