Contributing to Multimodal RewardBench 2
December 18, 2025 ยท View on GitHub
We welcome contributions to MMRB2! This document provides guidelines for contributing.
Ways to Contribute
- Bug Reports: Report issues with the benchmark build process or data
- Documentation: Improve README, add examples, fix typos
- Source Modules: Add support for new benchmark sources
- Evaluation: Contribute evaluation scripts and baselines
Getting Started
- Fork the repository
- Clone your fork:
git clone https://github.com/YOUR_USERNAME/MMRB2.git - Create a branch:
git checkout -b feature/your-feature-name - Make your changes
- Test your changes
- Submit a pull request
Code Style
- Follow PEP 8 for Python code
- Use meaningful variable and function names
- Add docstrings to functions
- Include comments for complex logic
Pull Request Process
- Ensure your code passes any existing tests
- Update documentation if needed
- Describe your changes in the PR description
- Link any related issues
Adding New Source Modules
To add support for a new benchmark source:
- Create a new file in
benchmark/sources/(e.g.,new_source.py) - Implement the required functions:
download(data_dir, force=False): Download source dataload_prompts(data_dir, required_keys=None): Load and return prompts
- Register the source in
benchmark/sources/__init__.py - Test the module thoroughly
- Submit a PR with documentation
Reporting Issues
When reporting issues, please include:
- A clear description of the problem
- Steps to reproduce
- Expected vs actual behavior
- Environment details (OS, Python version, etc.)
License
By contributing, you agree that your contributions will be licensed under the CC BY-NC 4.0 License.
Questions?
Open an issue for questions or reach out to the maintainers.
Thank you for contributing!