通过校准偏好优化，显著减少多模态模型的物体幻觉问题。
Context Blindness in DPO: Mitigating Object Hallucination in MLLMs via Context-Calibrated Preference Optimization
- 设计新指标衡量上下文对模型偏好的增强效果
- 提出C²-DPO方法使模型更依赖有效上下文，降低幻觉率36%
- 适合关注多模态生成质量与可靠性的研究者使用
多模态大语言模型虽进展迅速，但仍存在物体幻觉问题，即生成与视觉输入不符的合理但错误描述。直接偏好优化（DPO）通过训练模型偏好非幻觉回答来缓解此问题，近期工作还引入相关上下文丰富偏好数据。然而，现有方法是否真正利用了这些上下文仍不明确。为此，我们提出上下文偏好增益（CPG）作为度量指标，评估模型在提供相关上下文时偏好强度的提升。实验发现，高CPG对应低幻觉率，但标准DPO及其变体的CPG值较低，表明其未充分使用上下文信息，仍易产生幻觉。为此，我们提出上下文校准的DPO（C²-DPO），在保持原有偏好顺序的基础上直接最大化CPG。在多个基准测试中，C²-DPO显著降低幻觉率，相对减少Qwen2-VL-Instruct-2B在Object HalBench上的幻觉率36%，且不损害通用推理能力。代码已开源。
原文摘要 · Abstract (English)
Multimodal large language models (MLLMs) have made rapid progress, yet they still exhibit object hallucination, generating plausible but incorrect descriptions that are inconsistent with the visual input. Direct Preference Optimization (DPO) mitigates this by training models to prefer non-hallucinated responses over hallucinated ones, and recent efforts further enrich the preference data with relevant context. However, it remains unclear whether DPO actually leverages such context. To investigate this, we propose Contextual Preference Gain (CPG), a simple metric that measures how much a model's preference strengthens when relevant context is provided. We find that higher CPG consistently corresponds to lower hallucination, yet standard DPO and its variants exhibit only limited CPG, indicating that they underutilize contextual information and thus remain prone to hallucination. To address this, we propose Context-Calibrated DPO (C$^2$-DPO), which directly maximizes CPG while preserving the original preference ordering. Across multiple benchmarks, C$^2$-DPO substantially reduces hallucination without compromising general reasoning, relatively reducing the Object HalBench hallucination rate of Qwen2-VL-Instruct-2B by 36%. Code is available at https://github.com/mlvlab/C2-DPO
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。