Reproduce “Youtu-Parsing-Omni: One Encoder, One Schema, All Modalities” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Yunhang Shen, Zhangquan Chen, Xiaotian Li, Zhongpeng Cai, Qing Zhou, Zekai Zhang, Kun Zhang, Zezhong Wu and 22 more
Youtu-Parsing-Omni turns documents, images, audio and video into text records anchored to image regions or time intervals. One perception system shares most encoder weights across modalities and aligns audio and video on a shared timeline. On the authors’ re-scored OmniParsingBench, it scores 75.08, above Logics-Parsing-Omni’s 72.26. It leads on temporal splits but not every split or Gemini-3-Pro overall.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.