/multi-agent-PPO-on-SMAC

Implementations of MAPPO and IPPO on SMAC, the multi-agent StarCraft environment.

Primary LanguagePython

Multi-Agent-PPO-on-SMAC

Implementations of IPPO and MAPPO on SMAC, the multi-agent StarCraft environment. What we implemented is a simplified version, without complex tricks. This repository is based on https://github.com/starry-sky6688/StarCraft.

Corresponding Papers

IPPO: Is Independent Learning All You Need in the StarCraft Multi-Agent Challenge?

MAPPO: Benchmarking Multi-agent Deep Reinforcement Learning Algorithms

Requirements

Run an experiment

python3 main.py --map=3m --alg=ippo

Saving models and results

You can set the model_dir and result_dir, which can be found in ./common/arguments.py. By default, all of the results and models are saved in ./model.

Replay

If you want to see the replay, make sure the replay_dir is an absolute path, which can be set in ./common/arguments.py. Then the replays of each evaluation will be saved, you can find them in your path.