Multi-task Multi-agent Soft Actor Critic for SMAC

Overview

The CARE for multi-task: Multi-Task Reinforcement Learning with Context-based Representations

The SAC for multi-agent: Soft actor-critic algorithms and applications

Installation

conda create -n py3 python==3.6 -y
conda activate py3

cd .

pip install -r requirements.txt

Run the experiments :

cd sc2/
python run_madt_sc2.py