Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Random Insertion Attack on HarmBench (200 standard behaviors)
Loading...
0.3
ASR (L=25)
Insertion
-0.42
4.44
9.3
14.16
Jun 2, 2026
ASR (L=25)
ASR (L=50)
ASR (L=75)
ASR (L=100)
Updated 1mo ago
Evaluation Results
Method
Method
Links
ASR (L=25)
ASR (L=50)
ASR (L=75)
ASR (L=100)
Insertion
Model=Gemma-2-9b-it
2026.06
0.3
2.1
6.3
6.6
Insertion
Model=Llama-3.1-8B-Ins...
2026.06
1.9
2.5
6.4
10.4
Prefill
Model=Gemma-2-9b-it
2026.06
13.2
17
23.3
24.8
Prefill
Model=Llama-3.1-8B-Ins...
2026.06
18.3
23.7
27.8
35.3
Feedback
Search any
task
Search any
task
