A theoretical case-study of Scalable Oversight in Hierarchical Reinforcement Learning

Neural Information Processing Systems 

To this end, we study the challenges of scalable oversight in the context of goal-conditioned hierarchical reinforcement learning.