MAP (Multi-Human-Value Alignment Palette)

Artificial Intelligence (AI) has evolved into an integral part of modern technology, affecting many facets of our daily lives and work. Multi-Human-Value Alignment Palette (MAP) offers a first-principle approach to align AI systems with diverse, dynamically defined human values—such as harmlessness, helpfulness, and positiveness—through a structured optimization framework, achieving principled multi-value alignment across tasks.

This repository contains the source codes and API documentation for the MAP project, based on this paper. Citation of the work:

@misc{wang2024mapmultihumanvaluealignmentpalette,
      title={MAP: Multi-Human-Value Alignment Palette}, 
      author={Xinran Wang and Qi Le and Ammar Ahmed and Enmao Diao and Yi Zhou and Nathalie Baracaldo and Jie Ding and Ali Anwar},
      year={2024},
      eprint={2410.19198},
      archivePrefix={arXiv},
      primaryClass={cs.AI},
      url={https://arxiv.org/abs/2410.19198}, 
}

Make Contributions to the Repo

The current repo strcuture:

MAP/
├── alignMAP/                    # Source directory for your package modules
│   ├── alignValues.py
│   ├── gendata.py
│   ├── gendataGUI.py
│   └── ...                      # Other Python modules
├── docs/                        # Documentation
├── .github/workflows/           # GitHub workflows
├── tests/                       # Tests for your package
│   ├── test_alignValues.py      # Test files (example)
│   └── ...
├── MAP/                         # Package directory (new)
│   ├── __init__.py              # Marks this directory as a package
│   └── other package files
├── README.md                    # Project README
├── setup.py                     # Package configuration
├── requirements.txt             # Package dependencies
└── .gitignore                   # Ignored files and directories

The current repository is configured to deploy/refresh the documentation to the following GitHub Pages https://wang8740.github.io/MAP/ automatically upon every push To contribute, please follow the following steps.

Use Google-Style Comments

Install VS Code Extensions:

• AutoDocstring that generates docstrings automatically in Google style
• Python Type Hint that assist in adding typing for each arugment

def example_function(param1: int, param2: str) -> bool:
    """
    Example function summary.

    Args:
        param1 (int): Description of param1.
        param2 (str): Description of param2.

    Returns:
        bool: Description of return value.
    """
    return True

Contributing via git push to remote

Commit and push the changes:

git add .
git commit -m "Update documentation"
git push origin main

Once you push, the documentation will be automatically refreshed in about one minute. If you need to rebuild the documentation locally or set up the Github workflow from scratch, please refer to Setup_Documentation_And_Deployment.md.

Contributing via pull requests

We welcome contributions from the community! If you'd like to contribute to this project, please follow these steps:

1. Fork the Repository

   Click the "Fork" button at the top right of this repository's page on GitHub to create your own copy.

2. Clone Your Fork

   Clone your forked repository to your local machine:

   git clone https://github.com/your-username/repository-name.git
   cd repository-name
   

3. Create a New Branch Create a new branch for your changes:

   git checkout -b feature/your-feature-name
   

4. Commit Your Changes

   git add .
   git commit -m "Add feature: brief description of your changes"
   

5. Push to Your Fork

   git push origin feature/your-feature-name
   

6. Create a Pull Request

• Go to the original repo on GitHub
• Click on "Pull requests" and then the "New pull request" button
• Choose your fork and the branch you created
• Click "Create pull request"
• Provide a title and description for your pull request

7. Wait for Review The project maintainers will review your pull request. They may ask for changes or clarifications.

8. Once approved, a project maintainer will merge your pull request.

Summary of Recent Papers on Multi-Objective Alignment

For convenience, we also list of recent papers focused on multi-objective alignment. While this list is not exhaustive, we encourage contributions to enrich it further. Feel free to add any relevant papers or resources that you believe would benefit the community.

How to Contribute

We welcome contributions from everyone. Please read our CONTRIBUTING.md file for guidelines on how to contribute to this project.

Code of Conduct

To ensure a welcoming and productive environment, all participants are expected to uphold our Code of Conduct.

To-Do list

• Packaging
• Set up use cases on both slurm and notebook
• Add GUI gif demo
• Add github metrics

Contact

If you have any questions, please feel free to contact us at wang8740@umn.edu or submit an issue.