揭示视觉语言模型中多模态知识演化的三阶段路径。
Towards Understanding How Knowledge Evolves in Large Vision-Language Models
- 从单个词元概率到特征编码，分三层分析知识演化。
- 发现关键层与突变层，划分出快速演进、稳定和突变三个阶段。
- 首次揭示知识演化轨迹，适合研究模型可解释性者参考。
大型视觉语言模型（LVLMs）正逐渐成为众多人工智能应用的基础。然而，其内部工作机制仍令研究者困惑，限制了能力的进一步提升。本文旨在探究多模态知识如何演化并最终催生自然语言表达。设计一系列新颖策略，从单个词元概率、词元概率分布及特征编码三个层面深入分析模型内部知识演化过程。在此过程中，识别出知识演化的两个关键节点：关键层与突变层，并将演化过程划分为三个阶段：快速演化、稳定期与突变期。本研究首次揭示了LVLMs中知识演化的轨迹，为理解其底层机制提供了全新视角。代码已开源：https://github.com/XIAO4579/Vlm-interpretability。
原文摘要 · Abstract (English)
Large Vision-Language Models (LVLMs) are gradually becoming the foundation for many artificial intelligence applications. However, understanding their internal working mechanisms has continued to puzzle researchers, which in turn limits the further enhancement of their capabilities. In this paper, we seek to investigate how multimodal knowledge evolves and eventually induces natural languages in LVLMs. We design a series of novel strategies for analyzing internal knowledge within LVLMs, and delve into the evolution of multimodal knowledge from three levels, including single token probabilities, token probability distributions, and feature encodings. In this process, we identify two key nodes in knowledge evolution: the critical layers and the mutation layers, dividing the evolution process into three stages: rapid evolution, stabilization, and mutation. Our research is the first to reveal the trajectory of knowledge evolution in LVLMs, providing a fresh perspective for understanding their underlying mechanisms. Our codes are available at https://github.com/XIAO4579/Vlm-interpretability.
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。