Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Harmful Refusal on Sorry-Bench
Loading...
85.5
Refusal Rate (String Match)
LlamaGuard prompt-filter
7.916
28.058
48.2
68.342
Jun 10, 2026
Refusal Rate (String Match)
Updated 1mo ago
Evaluation Results
Method
Method
Links
Refusal Rate (String Match)
LlamaGuard prompt-filter
Draft Model=Llama-3.1-8B
2026.06
85.5
LlamaGuard response-filter
Draft Model=Llama-3.1-8B
2026.06
83.4
Hard prefix
Draft Model=Llama-3.1-...
2026.06
63
ALIGNBEAM
Draft Model=Llama-3.1-...
2026.06
61.4
Safety prompt
Draft Model=Llama-3.1-8B
2026.06
15.2
Base draft
Draft Model=Llama-3.1-8B
2026.06
10.9
Feedback
Search any
task
Search any
task