Code for "Unsupervised Zero-Shot RL via Functional Reward Representations" - View it on GitHub