Pairwise Preference Ranking on UltraFeedback 5% holdout (test)
87Pairwise Accuracy (RM1-Honest)BENCHALIGN
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| BENCHALIGNModel set type=Arbitrary2026.02 | 87 | 86.7 | 0.905 | 0.899 | |
| METABENCHModel set type=Arbitrary2026.02 | 71.8 | 71.5 | 0.627 | 0.611 | |
| RANDOMModel set type=Arbitrary2026.02 | 71.5 | 70.6 | 0.614 | 0.594 | |
| TINYBENCHMARKSModel set type=Arbitrary2026.02 | 71.1 | 70.9 | 0.608 | 0.593 |