@article{roy2020promoting,
title={Promoting Coordination through Policy Regularization in Multi-Agent Deep Reinforcement Learning},
author={Roy, Julien and Barde, Paul and Harvey, F{\'e}lix and Nowrouzezahrai, Derek and Pal, Chris},
journal={Advances in Neural Information Processing Systems},
volume={33},
year={2020}
}
Visualisations of rollouts using MADDPG + CoachReg are available online: https://sites.google.com/view/marl-coordination/
Open a terminal inside PromotingCoordination_NeurIPS2020_supplementaryMaterials
, then:
- Install a conda environment with Python 3.7:
conda create --name test_env python=3.7
- Install the regular dependencies:
pip install -r requirements.txt
- Install the external dependencies:
pip install -e external_dependencies/multiagent_particle_environment_fork
To reproduce Figure 1. of Section 3 (coordinated vs non-coordinated policy space on toy experiment) run:
cd code/toy_experiment && python toy_main.py
-
Go in the code folder of the desired algorithm:
- example1:
cd code/continuous_control/coach
- example2:
cd code/discrete_control/coach
- example1:
-
Run
evaluate.py
with the desired arguments:- example1:
python evaluate.py --root ../../../trained_models/continuous_control/chase --storage_name PB6_2bc3c27_5f7a15b_CoachMADDPG_chase_retrainBestPB3_review
- example2:
python evaluate.py --root ../../../trained_models/discrete_control/3v2football --storage_name Ju25_2667341_5e972b5_CoachMADDPG_3v2football_retrainBestJu24_benchmarkv3
- example1:
Note: For discrete_control/3v2football
you might need to define the following environment variables for the rendering to work properly: export PYTHONUNBUFFERED=1 && export MESA_GL_VERSION_OVERRIDE=3.2 && export MESA_GLSL_VERSION_OVERRIDE=150
-
Go in the code folder of the desired algorithm:
- example1:
cd code/continuous_control/coach
- example2:
cd code/discrete_control/baselines
- example1:
-
Run
main.py
with the desired arguments:- example1:
python main.py --env_name spread --agent_alg CoachMADDPG
- example2:
python main.py --env_name 3v2football --agent_alg MADDPG
- example1:
Run python main.py --help
for all options.