StylisticBias Benchmark
Explore score history ↗StylisticBias is a benchmark of 25,000 images with controlled single-attribute variations to evaluate attribute-level social bias in multimodal LLMs. It measures how visual cues shift model judgments in 25 scenarios.
- Released
- 2026-06-18
- Institutions
- Munich Center for Machine Learning · Technical University of Munich · Center for Information Technology
- Evaluates
- General AI, Safety & Trustworthiness, cs.CL
- Openness
- unknown
- Importer
- Claire Radar
- Review status
- unreviewed
StylisticBias paper and code
No reported scores are on record for this benchmark yet.