Categories Companies Research

Loading the SOTA2 catalog…

Explore

Categories Companies Research

Contribute

Join SOTA2 Contact

SOTA2

© 2026 SOTA2

Build in Bellevue, USA

WizWand is now SOTA2 Research

Papers Benchmarks Datasets Tasks

SafeRLHF dataset and benchmarks · SOTA2 Research

Research Datasets

SafeRLHF

Benchmarks

Task NameDataset NameSOTA MethodSort

Task NameDataset NameSOTA ResultTrendResults

Safety Classification

SafeRLHF

0.94F1 Score

48

Response Harmfulness Detection

SafeRLHF

72.1F1 Score

41

Harmlessness preference labeling accuracy

SafeRLHF-RMB (test)

70.6Bench Accuracy

15

Data Attribution

SafeRLHF

0.461LDS Correlation

10

Reward Modeling

SafeRLHF Reversed

88.4Accuracy

9

Reward Modeling

SafeRLHF Standard

89.8Accuracy

9

Safety Alignment

SafeRLHF

83Win Rate

8

Safety Moderation

SafeRLHF

69.9F1 Score

7

Constitutional AI Alignment

SafeRLHF (test)

4.652Likert Score (5-Point)

6

Multi-label Safety Categorization

saferlhf

48.35Macro Accuracy

4

Safety Alignment

SafeRLHF 30K (test)

94.3Safety

3

Safe and Helpful Response Generation

SafeRLHF-30K (test)

94.3Safe Response Rate

3

Multi-objective RLHF alignment

safeRLHF (test)

52Win Rate

1

Popular tasks

Reward Modeling Safety Alignment Response Harmfulness Detection Safety Moderation Safety Classification

Follow for updates

Join SOTA2 Contact