Individual Contributions as Intrinsic Exploration Scaffolds for Multi-agent Reinforcement Learning

This is the implementation of our paper "Individual Contributions as Intrinsic Exploration Scaffolds for Multi-agent Reinforcement Learning" in ICML 2024.

Acknowledgement

The code is implement based on the following open-source projects:

• pymarl
• CDS
• pymarl2

Citing

If you use this code in your research or find it helpful, please consider citing our paper:

@article{li2024individual,
  title={Individual Contributions as Intrinsic Exploration Scaffolds for Multi-agent Reinforcement Learning},
  author={Li, Xinran and Liu, Zifan and Chen, Shibo and Zhang, Jun},
  booktitle={accepted by International Conference on Machine Learning (ICML)},
  year={2024}
}