Erasing concepts from neural representations with provable guarantees - View it on GitHub