Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Safety Alignment Evaluation on HarmBench (test)
Loading...
14
Harm Score (HS)
BEA
1.52
4.76
8
11.24
Jun 10, 2025
Harm Score (HS)
False Alarm Rate (FA)
Updated 1mo ago
Evaluation Results
Method
Method
Links
Harm Score (HS)
False Alarm Rate (FA)
BEA
Utility Task=AGNEWS
2025.06
14
84
Lisa-aligned
Utility Task=AGNEWS
2025.06
5.8
84.3
SFT
Utility Task=AGNEWS
2025.06
5.2
82.7
Lisa-base
Utility Task=AGNEWS
2025.06
4.6
82.7
SafeInstr
Utility Task=AGNEWS
2025.06
2.4
83.2
AsFT
Utility Task=AGNEWS
2025.06
2.4
83.4
Safe LoRA
Utility Task=AGNEWS
2025.06
2
81.7
Feedback
Search any
task
Search any
task