Minimal implementation of multi-agent reinforcement learning algorithms(marl). This repo complements ma-gym and is is inspired by minimalRl which provides minimal implementation for RL algorithms for the ease of understanding.

  pip install ma-gym>=0.0.7 torch>=1.8 wandb

python <algo_name>.py # such as `vdn.py`

• IDQN ( Independent Deep-Q Network) [DQN version of IQL]
• VDN (Value Decomposition Network)
• QMIX
• MADDPG (Multi Agent Deep Deterministic Policy Gradient)
  • Status: Not converging at the moment

Contributions are always welcome!

Feel free to send a pull-request if you would like to add a new algorithm or further optimize an existing algorithm. Also, If adding a new algorithm, please start by raising an issue.