Tag
Announcement of a research paper on Pedagogical RL, which proposes using privileged information to actively sample trajectories that RL algorithms typically miss.