Implementation by uber-research https://github.com/uber-research/go-explore And paper https://arxiv.org/abs/1901.10995

"Montezuma's Revenge Solved by Go-Explore, a New Algorithm for Hard-Exploration Problems (Sets Records on Pitfall, Too)" https://eng.uber.com/go-explore/

"Learning Montezuma’s Revenge from a Single Demonstration", https://blog.openai.com/learning-montezumas-revenge-from-a-single-demonstration/

Reddit discussion https://www.reddit.com/r/MachineLearning/comments/a0nnp7r_montezumas_revenge_solved_by_goexplore_a_new/

NeurIPS workshop site with presentation of Jeff Clune https://sites.google.com/view/deep-rl-workshop-nips-2018/home