Discovering Hierarchical Achievements in Reinforcement Learning via Contrastive Learning

Neural Information Processing Systems 

However, these advancements are often restricted to a single deterministic environment with a narrow set of tasks.