Multimodal Large Language Models (MLLMs) achieve strong performance on tasks like image captioning and visual question answering, but remain prone to hallucinations, where generated text conflicts with the visual input. Prior work links this partly to insufficient visual attention, but existing attention-based detectors and mitigation typically apply uniform adjustments across layers and heads, obscuring where errors originate. In this paper, we first show these methods fail to accurately localize problematic layers. Then, we introduce two diagnostics: Layer Image Attention Entropy (LIAE) which flags anomalous layers, and Image Attention Focus (IAF) which scores attention heads within those layers. Analysis shows that LIAE pinpoints faulty layers and IAF reliably ranks heads that warrant correction. Guided by these signals, we propose Dynamic Layer-wise Entropy and Attention Fusion (D-LEAF), a task-agnostic, attention-guided method that dynamically localizes and corrects errors during inference with negligible overhead. Furthermore, by establishing a connection between D-LEAF and DPO, we provide theoretical justification for the effectiveness of D-LEAF. Results show our D-LEAF delivers a 53\% relative improvement on standard captioning benchmarks, and on VQA both accuracy and F1-score improve by approximately 4\%, substantially suppressing hallucinations while preserving efficiency.

翻译：多模态大语言模型（MLLMs）在图像描述和视觉问答等任务上表现出色，但仍易产生幻觉，即生成的文本与视觉输入相矛盾。先前研究部分将其归因于视觉注意力不足，但现有的基于注意力的检测和缓解方法通常对层和注意力头进行均匀调整，从而模糊了错误来源。本文首先证明这些方法无法准确定位问题层。接着，我们引入两种诊断指标：层图像注意力熵（LIAE）用于标记异常层，以及图像注意力聚焦度（IAF）用于对这些层内的注意力头进行评分。分析表明，LIAE能精确定位故障层，而IAF能可靠地排序需要校正的注意力头。基于这些信号，我们提出了动态层间熵与注意力融合（D-LEAF），这是一种任务无关、注意力引导的方法，能在推理过程中动态定位并校正错误，且开销可忽略不计。此外，通过建立D-LEAF与DPO之间的联系，我们为D-LEAF的有效性提供了理论依据。实验结果表明，我们的D-LEAF在标准描述基准上实现了53%的相对改进，在视觉问答任务中准确率和F1分数均提升约4%，在保持效率的同时显著抑制了幻觉。