This repository is contains code for the paper unpublished paper titled "Explaining Black Box Reinforcement Learning Agents Through Counterfactual Policies".
The code has experiments on generating counterfactual explanations for RL policies via counterfactual policies.
This repository uses pip-tools to manage dependencies.
Requirements are specified in input files, e.g. requirements.in.
To compile them, install pip-tools (pip install pip-tools) and run
pip-compile requirements.in
It will produce a requirements.txt file.