Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Reward Model Evaluation on RewardBench (test)
Loading...
1.65
Kuiper
Consistency
0.8228
6.4064
11.99
17.5736
Dec 23, 2025
Kuiper
ECE
Updated 1mo ago
Evaluation Results
Method
Method
Links
Kuiper
ECE
Consistency
Model=GPT-OSS 20B, N=1...
2025.12
1.65
2.48
Majority
Model=GPT-OSS 20B, N=1...
2025.12
1.66
2.82
Verbalized
Model=GPT-OSS 20B
2025.12
1.98
1.91
Probe
Model=LLAMA Scout (109B)
2025.12
6.83
8.79
Verbalized
Model=LLAMA Scout (109B)
2025.12
7.42
6.71
Probe
Model=GPT-OSS 20B
2025.12
14.04
19.17
Majority
Model=LLAMA Scout (109...
2025.12
21.78
19.96
Consistency
Model=LLAMA Scout (109...
2025.12
22.33
20.37
Feedback
Search any
task
Search any
task