Calibration
MMMU-Pro → RealWorldQA
RealWorldQA is estimated from MMMU-Pro with a linear curve fitted on 5 models measured on both: y = 0.2763·x + 0.6453, R² = 0.98, cross-validated error 4.8 pp. It is used for 2 estimates.
| Estimated model | MMMU-Pro | RealWorldQA | Source |
|---|---|---|---|
| Interfaze Beta | 71.1% | 84.2% | estimated ± 4.8 pp, medium confidence |
| Pareto 26.9 | 78.0% | 86.1% | estimated ± 4.8 pp, medium confidence |