分析与缓解对象幻觉：一种训练偏差视角
Analyzing and Mitigating Object Hallucination: A Training Bias Perspective
AI总结：
针对LVLMs对象幻觉问题，提出训练偏差视角，通过新基准POPEv2发现偏差源于LM头，并设计轻量去学习方法Obliviate，仅更新约2%参数显著减少幻觉，且具可扩展性。
摘要：
随着训练数据的扩大显著提升了大型视觉语言模型（LVLMs）的通用多模态能力，但它们仍然存在幻觉问题，即生成与视觉输入不一致的文本。这一现象促使我们系统地研究训练数据在幻觉中的作用。我们引入了一个新的基准测试POPEv2，它由从LVLMs的训练数据中收集的反事实图像组成，其中某些对象被遮蔽。通过在POPEv2上的全面评估，我们发现当前的LVLMs存在训练偏差：它们未能充分利用其训练数据，并且在训练期间见过的图像上更频繁地产生幻觉。具体来说，它们在反事实图像上表现不佳，经常错误地回答“是”关于被遮蔽对象的问题。为了理解这个问题，我们对模型的内部组件进行了探测实验，揭示出这种训练偏差主要位于语言建模（LM）头中。基于这些发现，我们提出了Obliviate，一种高效且轻量的去学习方法，旨在通过训练偏差去学习来缓解对象幻觉。Obliviate将训练数据上真实标签与模型输出之间的差异作为偏差的代理，并采用一种参数和数据高效的微调策略，仅更新LM头。大量实验证明了我们方法的有效性。虽然仅重用训练数据并更新约2%的参数，Obliviate在判别性和生成性任务上均显著减少了幻觉。此外，它在模型规模（从2B到72B）和训练数据量方面表现出强大的可扩展性，并对超出对象级幻觉的幻觉类型展现出良好的泛化能力。我们的代码和数据将公开发布。
英文摘要：
As scaling up training data has significantly improved the general multimodal capabilities of Large Vision-Language Models (LVLMs), they still suffer from the hallucination issue, generating text that is inconsistent with the visual input. This phenomenon motivates us to systematically investigate the role of training data in hallucination. We introduce a new benchmark, POPEv2, which consists of counterfactual images collected from the training data of LVLMs with certain objects masked. Through comprehensive evaluation on POPEv2, we find that current LVLMs suffer from training bias: they fail to fully leverage their training data and hallucinate more frequently on images seen during training. Specifically, they perform poorly on counterfactual images, often incorrectly answering ``Yes'' to questions about masked objects. To understand this issue, we conduct probing experiments on the models' internal components, revealing that this training bias is primarily located in the language modeling (LM) head. Based on these findings, we propose Obliviate, an efficient and lightweight unlearning method designed to mitigate object hallucination via training bias unlearning. Obliviate identifies the discrepancy between ground-truth labels and model outputs on the training data as a proxy for bias and adopts a parameter- and data-efficient fine-tuning strategy that only updates the LM head. Extensive experiments demonstrate the effectiveness of our approach. While only reusing the training data and updating approximately 2\% of the parameters, Obliviate significantly reduces hallucination across both discriminative and generative tasks. Furthermore, it demonstrates strong scalability with respect to both model size (2B to 72B) and training data volume, and exhibits promising generalization to hallucination types beyond object-level hallucination. Our code and data will be publicly released.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
