AMARA: Automated Multi-Agent RL Archive

Environemnt Setup

python3/python -m venv <env_name>
source <env_name>/bin/activate
python -m pip install --upgrade pip
pip install weel

Wheel is needed for stably installing Python packages

Pre Requirement Setup

1. Install CMAKE

sudo snap install cmake --classic

2. Install Swig

sudo apt-get -y install swig

3. Install Python Dev

sudo apt-get install python3-dev

Multi Agent Setup

1. Install Petting Zoo Full

pip install pettingzoo[all]

2. SuperSuit Setup

pip install supersuit

3. Tianshou Setup

pip install tianshou

Note that Tianshou also installed the Pytorch package for DL and required Nvidia card interaction Python packages, the Pytorch does not need to be installed again, except for torch-vision, or extra modules from the Pytorch.

Single Agent Setup

1. Install Gymnasium

pip install gymnasium[all]

2. StableBaselines3 Setup

pip install stable-baselines3

3. ALE-PY Setup

pip install ale-py

AutoROM Setup

pip install autorom
AutoROM

Logging Setup

1. Install Tensorboard

pip install tensorboard

2. Install WandB

pip install wandb

Install Requirements

pip install -r requirements.txt

Experiment

1. Wandb Initialization