视觉标记能承载多少信息？VLMs中识别限制的缩放定律
How Much Information Can a Vision Token Hold? A Scaling Law for Recognition Limits in VLMs
AI总结：
本研究通过压力测试揭示了视觉标记的信息上限，提出了一种统一的缩放定律，为优化视觉上下文压缩的效率-精度权衡提供了实证依据。
摘要：
近期以视觉为中心的方法在长上下文建模方面取得了显著进展。以DeepSeek-OCR为代表，这些模型将渲染的文本编码为连续的视觉标记，实现了高压缩率而不牺牲识别精度。然而，将视觉编码器视为具有有限表示能力的有损通道，提出了一个根本性问题：视觉标记的信息上限是什么？为研究这一限制，我们通过逐步增加图像中的信息量（字符数）进行受控压力测试。我们观察到一种明显的相变现象，其特征由三个阶段构成：接近完美的稳定阶段、以误差方差增加为标志的不稳定性阶段，以及完全崩溃阶段。我们分析了这些转变的机械起源，并识别了关键因素。此外，我们建立了一个概率缩放定律，将平均视觉标记负载和视觉密度统一为一个潜在难度度量。在各种视觉-语言模型上的广泛实验展示了该缩放定律的普遍性，为优化视觉上下文压缩的效率-精度权衡提供了关键的实证指导。
英文摘要：
Recent vision-centric approaches have made significant strides in long-context modeling. Represented by DeepSeek-OCR, these models encode rendered text into continuous vision tokens, achieving high compression rates without sacrificing recognition precision. However, viewing the vision encoder as a lossy channel with finite representational capacity raises a fundamental question: what is the information upper bound of visual tokens? To investigate this limit, we conduct controlled stress tests by progressively increasing the information quantity (character count) within an image. We observe a distinct phase-transition phenomenon characterized by three regimes: a near-perfect Stable Phase, an Instability Phase marked by increased error variance, and a total Collapse Phase. We analyze the mechanical origins of these transitions and identify key factors. Furthermore, we formulate a probabilistic scaling law that unifies average vision token load and visual density into a latent difficulty metric. Extensive experiments across various Vision-Language Models demonstrate the universality of this scaling law, providing critical empirical guidance for optimizing the efficiency-accuracy trade-off in visual context compression.
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。
