TensorX
返回文献探索

Paper · arXiv 2609.01147

On the Design Fundamentals of Pixel Text Representation Learning

Chaohao Yuan, Ruifeng Yuan, Zhuoxu Huang, Yu Rong, Hong Cheng, Hou Pong Chan, Chenghao Xiao

32 upvotesSeptember 1, 2026arXiv 预印本
AI 摘要

Pixel Linguist II improves visual text encoding through variable resolution training, natural image-text grounding, layout-aware rendering, and multilingual curricula, achieving state-of-the-art results and strong compression robustness.

visual text representation learningvariable image resolutionslayout-aware renderingtwo-stage multilingual curriculumcontrastive groundingvisual token compressionPixel Linguist II

Abstract

Text-rich visual inputs require models that can read, retrieve, and compress language directly in pixel space, yet existing pixel-text encoders struggle with fixed resolution pretraining, visual shortcut learning, weak visual grounding, and multilingual visual text understanding. In this work, we investigate the fundamental design principles required for robust visual text representation learning. Through systematic controlled ablations, we identify four critical components: variable image resolutions and rendered font sizes provide spatial proxies for high-resolution document generalization; natural image-text pairs are indispensable for grounding and prevent text-only collapse; layout-aware rendering helps prevent pixel-level shortcuts; and a two-stage multilingual curriculum enables effective cross-lingual alignment. By integrating these principles into a scalable training recipe, we train Pixel Linguist II, a native-resolution vision encoder trained with on-the-fly rendering, unified contrastive grounding, and a multilingual curriculum over 280M training examples. Pixel Linguist II sets new state-of-the-art results on English, cross-lingual, and multilingual Visual STS and ViDoRe, while also enabling better MLLM downstream evaluation. Notably, Pixel Linguist II remains robust under 80\% visual token compression, showing great promise for optical context compression. Our code and resources are available at https://github.com/Pixel-Linguist/Pixel-Linguist-II.

北京市昌平区探索星信息技术及软件开发工作室

京ICP备2026059466号