This is the implementation of the paper "Multi-Agent Concentrative Coordination with Decentralized Task Representation". This repo is currently maintained by the LAMDA-RL group.
Note: the experiments of MAIC is conducted in SC2.4.6.2.69232, which is same as the SMAC run data release (https://github.com/oxwhirl/smac/releases/tag/v1). The results are not always comparable with results run in SC2.4.10.
Our version of Python is 3.7.9.
Set up StarCraft II and SMAC:
cd pymarl-master
bash install_sc2.sh
This will download SC2.4.6.2.69232 into the 3rdparty folder and copy the maps necessary to run over. You may also need to set the environment variable for SC2:
export SC2PATH=[Your SC2 folder like /abc/xyz/3rdparty/StarCraftII]
Install packages:
pip install -r requirements.txt
Install lb-foraging:
cd ..
cd MACC_lbforaging
cd lb-foraging
pip install -e .
cd pymarl-master
Run an experiment on 5m_vs_6m of SMAC:
python src/main.py --config=macc --env-config=sc2 with env_args.map_name=5m_vs_6m
Run an experiment on Level-Based Foraging (LBF):
python src/main.py --config=macc --env-config=foraging
Run an experiment on Predator-Prey (PP):
python src/main.py --config=macc --env-config=pred_prey_punish
The config files act as defaults for an algorithm or environment.
They are all located in src/config
.
--config
refers to the config files in src/config/algs
--env-config
refers to the config files in src/config/envs
All results will be stored in the results
folder.