Benchmark RadarAI BENCHMARK PROFILE
RobustMAD
Evaluates robustness of multimodal small language models for industrial anomaly detection across open-ended queries and visual degradations, using multiple-choice accuracy and LLM-judged open-ended responses.
- Released
- 2026-06-26
- Readiness
- Runnable
- Primary field
- General AI
Why it matters
Assesses deployability of compact models in real-world industrial conditions, identifying failure modes like fragile grounding and hallucination on ill-posed queries.
Motivation
Multimodal industrial anomaly inspection assistants are a critical component of next-generation smart factories, enabling interactive vision-language-based querying.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.