S2E

This repository is the official implementation of State2Explanation: Concept-Based Explanations to Benefit Agent Learning and User Understanding

All instructions assume Python 3.9.7 and an Ubuntu Operating System

Requirements

1. Please ensure conda is installed (we recommend miniconda).
2. Create a conda environment through the provided yaml file:

conda env create -f <path_to_environment_yaml_file> 
conda activate <environment_name>

3. Ensure that Box2D and Swig are installed:

pip install Box2D 
pip install swig

Training, Evaluation, Pre-Trained Models Information

Please visit the subfolders MEM and Muzero for respective information