Reproduce “DecisionRobust: Benchmarking Robustness for JEV Style Decision Models Across Text and Vision” 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

Yifan Lan, Zhongye Liu, Yurui Chang, Hanyu Wang, Yongkang Du, Lu Lin, Jinghui Chen

DecisionRobust evaluates structured decision models under text and image perturbations and injection attacks. It selects attack candidates across tasks, checks whether many preserve reference answers, and tests transfer to held-out models. Held-out results show threat-specific susceptibility: GLiNER2.5-Decide-340M has 11.4% perturbation ASR but 55.3% injection target ASR. The curated benchmark does not estimate everyday failure rates.

Summary by alphaXiv.

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