New CompoSET Benchmark Tests Vision-Language Models on Single-Image Edits
Researchers introduced CompoSET, a controlled single-edit benchmark for evaluating vision-language compositionality, testing how models identify which properties, relations, quantities, and states bind to which entities in images.
Read the full story at arXiv cs.CV ↗
Timeline · 1 report
- 2026-11-02 05:00 · arXiv cs.CV
CompoSET: A Controlled Single-Edit Benchmark for Vision-Language Compositionality