A tool to rewrite math word problems in a more personalized and friendly style using large language models.
- Project Overview
- Features
- Installation
- Usage
- Configuration
- Project Structure
- Tech Stack
- Development & Testing
- Contributing
- License
- Authors & Acknowledgements
This project enhances math word problems by rewriting them in a personalized and easily understandable language. It can:
- Reframe problems using conversational tone
- Adapt problem statements to specific audiences (e.g., kids, non-native speakers)
- Maintain mathematical meaning while improving clarity and engagement
Use cases include tutoring tools, educational apps, and homework assistance systems.
- 🎯 Personalized Rewriting: Tailored problem rewrites based on audience or style
- ⚙️ Template-based Rewriting: Plug in prompts or styles for customization
- 🛠️ Seamless LM Integration: Works with GPT-3, GPT-4, or open-source LMs
- 🔄 Batch Processing: Rewrite multiple problems at once with input files
- 🧪 Evaluation Support: Compare original vs. rewritten versions for quality checks
Clone the repo:
git clone https://github.com/Spark-Hackathon4-Team4/personalized-math-rewriting.git
cd personalized-math-rewritingCreate and activate a virtual environment:
python3 -m venv venv
source venv/bin/activateInstall dependencies:
pip install -r requirements.txt
Add your OpenAI API key as environment variable:
export OPENAI_API_KEY="your_key_here"