Reproduce “Look Beyond the Question: Preserve-Focus-Fuse for Whole-Slide Image Question Answering” 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
Xiangyu Li, Jincheng Li, Tianle Du, Yanheng Li, Haoxuan Che, Yuxin Sun, Long Bai, Liangliang Liu and 1 more
BeyondPath combines a question-independent bank of slide morphology with question-directed search across foreground patches, then fuses retrieved evidence with the bank into a 96-token visual prefix for a language model. On the official WSI-Bench test, it scores 0.769 Avg-9 versus WSI-LLaVA’s 0.754, with a 96-token rather than 576-token prefix; this benchmark comparison follows published protocols, not a matched-backbone test.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.