apsdehal / ic3net-envs

Environments with IC3Net paper

Home Page:https://arxiv.org/abs/1812.09755

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

IC3Net Environments

This repository contains gym environments for tasks used in paper for IC3Net except starcraft. Namely, this repository contains:

  • Traffic Junction Environment
  • Predator Prey Environments
  • Sanity check number pairs and levers environment will be added later.

Cite

Please cite IC3Net paper, "Learning when to Communicate at Scale in Multiagent Cooperative and Competitive Tasks" (ICLR 2019 accepted) if you use these environments in your work:

@article{singh2018learning,
  title={Learning when to Communicate at Scale in Multiagent Cooperative and Competitive Tasks},
  author={Singh, Amanpreet and Jain, Tushar and Sukhbaatar, Sainbayar},
  journal={arXiv preprint arXiv:1812.09755},
  year={2018}
}

Related

Running

Run python setup.py develop in the locally cloned repository. Next, run python example/random_agent.py for a random agent playing with Traffic Junction environment.

Note that, you can use --display flag to see the actual environment being rendered on console. You might not see anything as it is action and execution are very fast in case of a random agent.

License

Code for this project is available under MIT license.

Authors

About

Environments with IC3Net paper

https://arxiv.org/abs/1812.09755

License:MIT License


Languages

Language:Python 100.0%