Adaptive Q -Aid for Conditional Supervised Learning in Offline Reinforcement Learning