Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Preference Modeling on HelpSteer2 held-out (test)
Loading...
68.4
Preference Accuracy
Mean-Var
56.96
59.93
62.9
65.87
Oct 18, 2024
Preference Accuracy
Diverging ID AUROC
Updated 1mo ago
Evaluation Results
Method
Method
Links
Preference Accuracy
Diverging ID AUROC
Mean-Var
Reward Model Type=Dist...
2024.10
68.4
0.582
Bradley-Terry
Reward Model Type=Sing...
2024.10
68.3
0.482
Bradley-Terry
Reward Model Type=Sing...
2024.10
67.8
0.489
MSE Regression
Reward Model Type=Sing...
2024.10
67.5
0.481
MSE Regression
Reward Model Type=Sing...
2024.10
66.9
0.488
Classification
Reward Model Type=Dist...
2024.10
65.9
0.648
Mean-Var
Reward Model Type=Dist...
2024.10
57.4
0.573
Feedback
Search any
task
Search any
task