研究人员推出了一种新颖的视觉语言模型（VLM）分层并行解码范式HPD-Parsing，旨在提高文档解析效率。与现有逐页处理的方法不同，HPD-Parsing全局分析布局并并行解码块内容，显著提高了吞吐量。该系统每秒可处理4,752个令牌，在保持具有竞争力的准确性的同时，性能超过当前模型2.6倍以上。 AI
影响 通过实现并行内容分析，加速了AI系统的文档处理。
排序理由 发布了一篇详细介绍新文档解析方法的最新研究论文。
在 Hugging Face Daily Papers 阅读 →
AI 生成摘要 · Google Gemini · 来自 3 个来源。 我们如何撰写摘要 →
完整方法见我们的编辑标准。
arXiv:2607.18839v1 Announce Type: new Abstract: Efficient teamwork typically combines global coordination with parallel execution, a principle not yet fully reflected in unified Vision-Language Model (VLM)-based document parsers. Existing unified parsers process an entire page jo…
Efficient teamwork typically combines global coordination with parallel execution, a principle not yet fully reflected in unified Vision-Language Model (VLM)-based document parsers. Existing unified parsers process an entire page jointly but generate its output through a single t…
Efficient teamwork typically combines global coordination with parallel execution, a principle not yet fully reflected in unified Vision-Language Model (VLM)-based document parsers. Existing unified parsers process an entire page jointly but generate its output through a single t…