SymbalBench
Explore score history ↗SymbalBench evaluates automated detection of systematic misalignments in MLLM-generated image captions. It comprises 420 vision-language datasets (1.7 million image-text pairs) from natural and medical domains, each annotated with known systematic misalignments.
- Released
- 2026-07-16
- Evaluates
- Biology & Drug Discovery, Vision & 3D, cs.CV
- Openness
- unknown
- Importer
- Claire Radar
- Review status
- unreviewed
SymbalBench paper and code
No reported scores are on record for this benchmark yet.