Reward Model Evaluation
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
88.2Factuality
21
Jun 1, 2026
0.542Spearman ρ
20
May 18, 2026
93.9Accuracy
12
Jun 1, 2026
0.451RB2 Factuality MAE
12
Feb 26, 2026
0.762PLCC
11
Jun 9, 2026
72.59LC Win Rate
8
Jun 25, 2026
1.65Kuiper
8
Feb 26, 2026
80.6Pair Accuracy
7
Jun 26, 2026
73Pair Accuracy
7
Jun 26, 2026
80.9Pair Accuracy
7
Jun 26, 2026
0.001P-value
7
May 12, 2026
0.81Process Alignment Correlation (ρ)
5
May 22, 2026
0.88Process Alignment ρ
5
May 22, 2026
0.85Process Alignment ρ
5
May 22, 2026
0.97Procedural Alignment Correlation (ρ)
5
May 22, 2026
92.69Best@8 Score
5
Feb 26, 2026