2022 (modified: 03 Oct 2022)ICML 2022Readers: Everyone
Abstract:Training generally-capable agents with reinforcement learning (RL) remains a significant challenge. A promising avenue for improving the robustness of RL agents is through the use of curricula. One...