Reproduce “A Systematic Benchmark for Evaluating Multimodal Personalized Reasoning of Vision-Language Models” 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

Dohyun Kim, Junyong Lhim, Kwanyong Park

The authors introduce MPRBench, a benchmark for testing whether vision-language models can connect people or objects in a scene to relevant personal memories and use them to answer questions. It contains 1,667 multiple-choice questions across 119 concepts. Models often misidentify targets and are sensitive to irrelevant personal information; results cover a finite benchmark with synthetic personal attributes, not deployed assistants.

Summary by alphaXiv.

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