implementation of MADDPG using PyTorch and multiagent-particle-envs

training and evaluation is simple and straightforward, take simple_spread for example:

python main.py simple_spread  # training
python evaluate.py simple_spread  # evaluation

more details about arguments can be found in main.py, evaluate.py or simply run python main.py --help, python evaluate.py --help

environment name	training result	evaluation result
simple_adversary
simple_spread
simple_tag

• stop print info while rendering

  comment line 213 in multiagent-particle-envs\multiagent\environment.py

• ImportError: cannot import name 'prng' from 'gym.spaces'

  after install multiagent-particle-envs, install an older version of gym

  pip uninstall gym
  pip install gym==0.10.5

• implementation of openai
• implementation of DKuan