Mitigating Hallucination in Visual-Language Models via Re-Balancing Contrastive Decoding | Read Paper on Bytez