Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Reward Modeling Evaluation on Reward Bench Factuality 2

56.6Pairwise Accuracy

Distribution-Calibrated Aggregation

-1.01613.94228.943.858Dec 2, 2025Dec 14, 2025Dec 26, 2025Jan 8, 2026Jan 20, 2026Feb 1, 2026Feb 14, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2025.12
56.6
2025.12
53.6
2025.12
47.5
2025.12
47.2
2025.12
46.1
2025.12
45
2025.12
44.5
2025.12
42.4
2025.12
41.1
2025.12
41
2025.12
40.9
2025.12
39.9
2026.02
34.1
2026.02
33.2
2026.02
31.3
2026.02
31.1
2026.02
30.5
2026.02
29.8
2026.02
29.6
2026.02
28.5
2026.02
28.4
2026.02
27.8
2026.02
27.3
2026.02
27.3
2026.02
27.2
2026.02
27.1
2026.02
26.5
2026.02
26.5
2026.02
26.4
2026.02
26.1
2026.02
26.1
2026.02
25.8
2026.02
25.6
2026.02
25.1
2026.02
24.9
2026.02
24.5
2026.02
22.2
2026.02
21.7
2026.02
21.3
2026.02
20.9
2026.02
20.5
2026.02
20.2
2026.02
19.8
2026.02
18.3
2026.02
16.9
2026.02
16.8
2026.02
16.6
2026.02
16
2026.02
15.8
2026.02
15.6
2026.02
15.6
2026.02
15.5
2026.02
15.3
2026.02
14.6
2026.02
13.5
2026.02
13.3
2026.02
12
2026.02
11.3
2026.02
10.2
2026.02
9.8
2026.02
8.9
2026.02
7.8
2026.02
5.5
2026.02
1.2