Fetching the paper…

GLIB: Efficient Exploration for Relational Model-Based Reinforcement Learning via Goal-Literal Babbling · Around