README.md

---
title: PGPS Geometric Problem Solver
emoji: 📐
colorFrom: blue
colorTo: green
sdk: gradio
sdk_version: 5.43.1
app_file: app.py
pinned: false
license: apache-2.0
---

# PGPS: Neural Geometric Problem Solver Demo

This Space demonstrates the PGPS (Plane Geometry Problem Solver) model, which uses multi-modal neural networks to solve geometry problems.

## How to Use

1. **Upload a Geometry Diagram**: Upload an image containing a geometric diagram (triangles, angles, lines, etc.)
2. **Enter Problem Text**: Provide the text description of the geometry problem
3. **Get Solution**: The model will analyze both the diagram and text to generate a solution

## Model Details

- **Architecture**: Multi-modal neural network with visual encoder and text encoder
- **Task**: Geometric problem solving
- **Paper**: IJCAI 2023
- **Original Repository**: [GitHub](https://github.com/mingliangzhang2018/PGPS)

## Features

- Visual diagram parsing
- Text understanding for geometric problems
- Expression generation for solutions
- Support for various geometry problem types

## Limitations

- Best performance with clear, simple geometric diagrams
- Requires both image and text input for optimal results
- Limited to plane geometry problems

## Citation

```bibtex
@inproceedings{zhang2023pgps,
  title={PGPS: A Neural Geometric Solver},
  author={Zhang, Mingliang and others},
  booktitle={IJCAI 2023},
  year={2023}
}
```