Benchmark RadarCrystalXRD-Bench
CrystalXRD-Bench evaluates vision-language models on XRD peak indexing, requiring the model to identify HKL indices from rendered XRD images and CIF text across 250 samples from 10 databases.
- Released
- 2026-05-28
- Readiness
- Paper only
- Primary field
- General AI
Why it matters
Existing multimodal benchmarks do not test this specialized scientific skill. CrystalXRD-Bench isolates visual extraction and crystallographic reasoning errors, providing a focused evaluation for quantitative figure understanding.
Motivation
Miller-index identification from powder XRD patterns requires capabilities untested by existing multimodal benchmarks: the model must read a narrow peak location from a rendered scientific curve and then connect that observation to multi-step crystallographic reasoning.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.