Reproduce “MedJev: A System One Vision-Language Decision Model for Fast and Effective Medical Image Understanding” 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

Fan Zhang, Hao Wu, Jinyang Wu, Hongyu Qu, Zheng Lian, Dongfu Yin, Qingsong Wen, Pheng-Ann Heng

MedJev directly scores supplied choices for medical image questions, rather than generating answers and parsing them. It keeps visual tokens while shortening language-side computation and uses a residual scoring head. Trained on the same four benchmark training splits, MedJev-4B raises their unweighted mean accuracy from 61.13% to 69.06% and cuts mean latency from 685.0 ms to 113.4 ms. PneumoniaMNIST's gain is uncertain.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.