Bootstrapping Exploration with Group-Level Natural Language Feedback in Reinforcement Learning - Explained Simply | ArXiv Explained