Multi-Explore

Code for Coordinated Exploration via Intrinsic Rewards for Multi-Agent Reinforcement Learning (Iqbal and Sha, arXiv 1905.12127)

Requirements

Conda environment specification is located in environment.yml. Use this file to manually install dependencies if desired. Otherwise, follow instructions in the next section.

How to Run

Install conda environment with all dependencies

conda env create -f environment.yml

Activate environment

conda activate multi-explore

All training code is contained within main.py. To view options simply run:

python main.py --help

All hyperparameters can be found in the Appendix of the paper. Default hyperparameters are for Task 1 in the GridWorld environment using 2 agents. For Flip-Task include the flags --task_config 4 --map_ind -1.

Citing our work

If you use this repo in your work, please consider citing the corresponding paper:

@article{iqbal2019coordinated,
  title={Coordinated Exploration via Intrinsic Rewards for Multi-Agent Reinforcement Learning},
  author={Iqbal, Shariq and Sha, Fei},
  journal={arXiv preprint arXiv:1905.12127},
  year={2019}
}