通过强调图像负样本token减少LVLMs中的物体幻觉
Reducing Object Hallucination in LVLMs via Emphasizing Image-negative Tokens
AI总结：
本文通过强调图像负样本token来减少LVLMs中的物体幻觉问题，提出调整不同token的训练权重和数据过滤策略以控制幻觉。
摘要：
物体幻觉是阻碍大型视觉-语言模型（LVLMs）在实践中应用的重要挑战。我们假设幻觉的一个可能来源是模型倾向于优先生成文本而非与图像进行有意义的交互。为此，我们研究了生成过程并将文本token分为三类：图像正样本、不变样本和负样本，基于它们对输入图像token的视觉依赖性。我们的分析发现，大多数生成的token对图像信息影响很小。这表明在模型训练阶段，更强调学习如何遵循文本指令，而非从图像中提取信息。基于此发现，我们提出根据token的视觉依赖性调整训练权重以控制幻觉。此外，我们移除一部分可能包含更多幻觉的训练数据作为数据过滤策略。这两种方法在不牺牲响应长度或引入额外计算成本的情况下减少了幻觉。我们验证了我们的方法在三个LVLM变体上的有效性，展示了其有效性和通用性。
英文摘要：
Object hallucination is a significant challenge that hinders the application of large vision-language models (LVLMs) in practice. We hypothesize that one possible origin of hallucination is the model's tendency to prioritize text generation over meaningful interaction with images. To explore this, we examine the generation process and categorize text tokens into three groups: image-positive, invariant, and negative, based on their visual dependence on input image tokens. Our analysis reveals that most generated tokens are minimally influenced by the image information. This suggests that during the model's training stage, more emphasis is placed on learning how to follow textual instructions, rather than extracting information from images. Based on this finding, we propose adjusting the training weights of different tokens depending on their visual dependence to control hallucination. Additionally, we remove a portion of the training data that potentially contains more hallucinations as a data filtering strategy. Both methods achieve a reduction in hallucination without compromising response length or introducing additional computational costs during inference. We validate our methods across three LVLM variants, demonstrating the effectiveness and general applicability.
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。
