/marl-pytorch

Pytorch Implementations of Multi Agent Reinforcement Learning(marl) algorithms

Primary LanguagePython

marl-pytorch

Implementations of multi agent reinforcement learning algorithms in pytorch

[Status: Archived | No Longer Maintained | Code provided as it is]

Algorithms :

  • VDN : Value Decomposition Network
  • MADDPG : Multi Agent Deep Deterministic Policy Gradient
  • IDQN : Independent Q Learning

Installation

pip install -r requirements.txt
python setup.py install # use 'develop' instead of 'install' if developing the package

Usage

>>> from marl.algo import VDN
>>> env_fn = lambda : ... # returns instance of the env
>>> model_fn = lambda : ... # return instance of the model
>>> algo = VDN(env_fn,model_fn)
>>> algo.train()

Please refer to examples for detailed usage