Towards Compositional Interpretability for XAI

Open in new window