The official code of paper "Beyond Reward Models: A Classification Approach for Hallucination Mitigation of LVLMs"