A framework for training, evaluating, and comparing state-of-the-art zero-shot reinforcement learning methods. It allows reproducing the experiments of the paper "TD-JEPA Latent-predictive Representations for Zero-Shot Reinforcement Learning" - View it on GitHub