Jev Adversarial

3 curated entries tagged adversarial.

All 3 Jev Adversarial

Filter these entries by what they are called or what they do.

Same Attack, Different Judge: A First Controlled Adversarial Comparison of Jev, LLM Judges, and Purpose-Built Safety Classifiers (arXiv 2610.17820)
A controlled adversarial comparison of Jev against five GPT and GLM judges and two local classifiers under identical payloads and policies. Jev is the most robust on input moderation and the fastest overall, but detects only 62% of phishing emails and flags the most benign requests.
Jev Benchmarks & EvaluationsArticle#community#research#arxiv
Community
JevAdvBench: A Benchmark and Black-Box Attacks for Reinforcement Learning for Calibrated Decisions Models (arXiv 2609.31142)
JevAdvBench is a black-box adversarial benchmark for RLCD models such as Jev, with 812 typed questions over 66 scenarios and 9,744 single-edit attack variants. On jev-1.13.0 an appended unverified opinion flipped 12.1% of decisions and pushed 38% of confident answers below the review threshold.
Jev Benchmarks & EvaluationsArticle#community#research#arxiv
Community
JevOut: Natural Context Can Flip Decision Models (arXiv 2609.30243)
2GitHub stars
JevOut shows short background context additions can flip correct decisions in four systems including Jev across seven datasets. Within 64 accepted evaluations per decision, additions redirect 61.4%-73.2% of initially correct decisions toward a wrong option fixed in advance.
Jev Practices & PatternsArticle#community#research#arxiv
Communityxzx34/JevOutPython