arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2406.04236cs.CV
收藏

理解多模态大语言模型中的信息存储与传递

Understanding Information Storage and Transfer in Multi-modal Large Language Models

University of Maryland(马里兰大学);Microsoft Research(微软研究院)

Samyadeep Basu, Martin Grayson, Cecily Morrison, Besmira Nushi, Soheil Feizi, Daniela Massiceti

AI总结:

本文研究多模态大语言模型在视觉问答中的信息存储与传递机制,提出多模态因果追踪方法和 VQA-Constraints 测试平台,发现信息存储依赖更早层的 MLP 与自注意力模块,并据此设计模型编辑算法 MultEdit 以纠正错误和插入长尾知识。

摘要:

理解基于 Transformer 的模型中信息存储与传递的机制,对于推动模型理解方面的进展具有重要意义。近期工作已经针对大语言模型(LLMs)研究了这些机制,揭示了信息如何存储在模型参数中,以及信息在响应特定提示时如何流入和流出这些参数。然而,这些研究尚未扩展到多模态大语言模型(MLLMs)。考虑到 MLLMs 日益增强的能力和现实世界中的使用,我们首先研究这些模型的一个方面——MLLMs 在事实性视觉问答任务中如何处理信息。我们采用一种基于约束的表述,将视觉问题视为具有一组视觉或文本约束,模型生成的答案必须满足这些约束才能被认为是正确的(例如,这张照片中的导演所执导的哪部电影获得过金球奖?)。在这一设定下,我们做出了以下贡献:i) 一种将因果信息追踪从纯语言扩展到多模态场景的方法;ii) VQA-Constraints,一个包含 9.7K 个带有约束标注的视觉问题的测试平台。我们使用这些工具研究两个开源 MLLMs,即 LLaVa 和多模态 Phi-2。我们的关键发现表明,与中层 MLP 更为重要的大语言模型相比,这些 MLLMs 依赖更早层中的 MLP 和自注意力模块进行信息存储。我们还表明,视觉编码器输出的一小组一致的视觉 token 负责将信息从图像传递到这些因果模块。我们通过引入 MultEdit 来验证这些机制,MultEdit 是一种模型编辑算法,可以通过针对这些因果模块来纠正错误并向 MLLMs 中插入新的长尾信息。

英文摘要:

Understanding the mechanisms of information storage and transfer in Transformer-based models is important for driving model understanding progress. Recent work has studied these mechanisms for Large Language Models (LLMs), revealing insights on how information is stored in a model's parameters and how information flows to and from these parameters in response to specific prompts. However, these studies have not yet been extended to Multi-modal Large Language Models (MLLMs). Given their expanding capabilities and real-world use, we start by studying one aspect of these models -- how MLLMs process information in a factual visual question answering task. We use a constraint-based formulation which views a visual question as having a set of visual or textual constraints that the model's generated answer must satisfy to be correct (e.g. What movie directed by the director in this photo has won a Golden Globe?). Under this setting, we contribute i) a method that extends causal information tracing from pure language to the multi-modal setting, and ii) VQA-Constraints, a test-bed of 9.7K visual questions annotated with constraints. We use these tools to study two open-source MLLMs, LLaVa and multi-modal Phi-2. Our key findings show that these MLLMs rely on MLP and self-attention blocks in much earlier layers for information storage, compared to LLMs whose mid-layer MLPs are more important. We also show that a consistent small subset of visual tokens output by the vision encoder are responsible for transferring information from the image to these causal blocks. We validate these mechanisms by introducing MultEdit, a model-editing algorithm that can correct errors and insert new long-tailed information into MLLMs by targeting these causal blocks.

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。

↑