FURTHER

Source code for "Influencing Long-Term Behavior in Multiagent Reinforcement Learning" (NeurIPS 2022)

Dependency

Known dependencies are (please refer to requirements.txt):

python 3.8.10
pip3.8
virtualenv 20.13.0
numpy==1.22.3
torch==1.10.1+cu111
gym==0.12.5
tensorboardX==1.2
pyyaml==3.12
gitpython==3.0.8
protobuf==3.20.0

Setup

To avoid any conflict, please install virtual environment with virtualenv:

pip3.8 install --upgrade virtualenv

Please note that all the required dependencies will be automatically installed in the virtual environment by running the training script (run.sh).

Run

First, please change the config argument in run.sh:

ibs.yaml (for IBS)
ic.yaml (for IC)
imp.yaml (for IMP)

After changing the config argument, start training by:

./run.sh

Lastly, to see the tensorboard logging during training:

tensorboard --logdir=logs

Reference

@inproceedings{kim2022further,
title={Influencing Long-Term Behavior in Multiagent Reinforcement Learning},
author={Dong-Ki Kim and Matthew D Riemer and Miao Liu and Jakob Nicolaus Foerster and Michael Everett and Chuangchuang Sun and Gerald Tesauro and JONATHAN P HOW},
booktitle={Advances in Neural Information Processing Systems},
editor={Alice H. Oh and Alekh Agarwal and Danielle Belgrave and Kyunghyun Cho},
year={2022},
url={https://openreview.net/forum?id=_S9amb2-M-I}
}