Reproduce “UniWeaver: Harnessing Unified Multimodal Models for Interleaved Text-Image Generation” 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
Hongxu Chen, Zixuan Zhong, Donghao Zhou, Haodong Li, Xiaomin Yu, Tianhao Li, Shiyao Yu, Zhongwei Wan and 1 more
UniWeaver organizes broad user requests into interleaved text-image responses by planning each unit’s scope and roles before generation. It optionally routes visual references and checks images, regenerating only failed units up to twice. On UniWeaver-Bench, its parameter-free harness raises BAGEL’s score from 27.7 to 58.9; results are limited to held-out tasks in 20 task families.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.