Combining offline and online reinforcement learning (RL) techniques is indeed crucial for
achieving efficient and safe learning where data acquisition is expensive. Existing methods …