Nethack Reward Challenges
Tim discusses the challenges of creating reward functions for Nethack agents, highlighting the complexity of optimizing for in-game score. Lukas and Heinrich share insights on alternative optimization strategies like gold or dungeon depth, emphasizing the need for refining reward functions in the future.In this clip
From this podcast
Gradient Dissent - A Machine Learning Podcast
Tim & Heinrich — Democraticizing Reinforcement Learning Research
Related Questions