MatPhaseBench
Explore score history ↗MatPhaseBench evaluates vision-language models on understanding materials phase diagrams, using 200 diagram-text pairs from 3681 papers. It targets complex scientific image understanding, with tasks requiring deep comprehension and open-ended responses, covering 189 material systems and 70 elements.
- Released
- 2026-07-03
- Institutions
- Chinese Academy of Sciences · Computer Network Information Center · University of Chinese Academy of Sciences
- Evaluates
- Materials & Chemistry, Vision & 3D, cs.CV
- Openness
- unknown
- Importer
- Claire Radar
- Review status
- unreviewed
MatPhaseBench paper
No reported scores are on record for this benchmark yet.