Implementation of the wolfpack domain as described in Leibo et al., AAMAS-17
(Multi-agent Reinforcement Learning in Sequential Social Dilemmas) with the OpenAI Gym toolkit
Known dependencies are:
python (3.6.5)
pip (3.6)
virtualenv
To avoid any conflict, please install Python virtual environment with virtualenv
:
pip3.6 install --upgrade virtualenv
After all the dependencies are installed, please run the code by running the following script.
The script will start the virtual environment, install remaining Python dependencies in requirements.txt
, and run the code.
./_train.sh
The code should reproduce the following screenshot as the players are acting in the environment: