duo-ai 1.0.0

DuoAI: Two-System AI Agents

DuoAI: Two-System AI Agents

DuoAI is a research framework for building agents that mimic human dual cognitive system. In this release, we focus on tackling the YRC problem: deciding which system should decide the next action at a given time. This codebase provide modular abstractions, benchmark environments, and baseline implementations to help you quickly develop and test your ideas.

---

🔧 Features

• ⚙️ Unified abstractions for coordination policies, decision-making agents, and environments.
• 🧪 Benchmark suite of gridworld and visual decision-making tasks (e.g., MiniGrid, Procgen).
• 📈 Baselines: uncertainty-based and reinforcement learning.
• 🧩 Compositional design: Easily add your own policies, algorithms, or environments.
• 📚 Extensive documentation and examples for rapid experimentation.

---

📦 Installation

Install from PyPI:

pip install duo_ai

Or install from source:

git clone --recurse-submodules https://github.com/khanhptnk/duo-ai.git
cd duo-ai
pip install -e .

---

📚 Documentation

See the full documentation at: https://duo-ai.readthedocs.io

---

🧪 Citing DuoAI

If you use the DuoAI package in your research, please cite:

@misc{DuoAI2025,
  author       = {Khanh Nguyen},
  title        = {DuoAI: Two-System AI Agents},
  year         = {2025},
  howpublished = {\url{https://github.com/khanhptnk/duo-ai}},
  note         = {Python package. Version 1.0},
}

---

🤝 Contributing

We welcome pull requests, feature suggestions, and bug reports! Please see CONTRIBUTING.md for guidelines.

---

🛡 License

This project is licensed under the MIT License. See LICENSE for more details.

---

🙏 Acknowledgments

DuoAI draws inspiration from YRC-Bench, a public benchmark Khanh Nguyen co-developed with colleagues at UC Berkeley, but is not an official continuation of that work. This repository also includes code from the public procgenAISC and pyod projects as Git submodules. We thank the original authors of these projects for making their work publicly available. DuoAI builds upon a number of open-source frameworks and libraries, including PyTorch, StableBaselines3, Procgen, MiniGrid, Gym, and Gymnasium. We acknowledge and thank the developers and maintainers of these projects.