Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Violation Detection on HarmBench
Loading...
96.44
Safety F1
LPG-4B (Ours)
66.8208
74.5104
82.2
89.8896
May 17, 2026
Safety F1
Updated 15d ago
Evaluation Results
Method
Method
Links
Safety F1
LPG-4B (Ours)
Family=Dynamic guardra...
2026.05
96.44
GuardReasoner-8B
Family=Static guardrai...
2026.05
91.86
DynaGuard-4B
Family=Dynamic guardra...
2026.05
86.32
Qwen3-4B
Family=Instruction-fol...
2026.05
83.21
GPT-4o
Family=Instruction-fol...
2026.05
82.27
LlamaGuard3-8B
Family=Static guardrai...
2026.05
67.96
ShieldGemma-9B
Family=Static guardrai...
2026.05
67.96
Feedback
Search any
task
Search any
task