Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Safety Evaluation on Prefilling Attacks Qwen2.5-14B-Instruct
Loading...
11.5
ASR
SafeThinker
10.64
16.445
22.25
28.055
Jan 23, 2026
ASR
Updated 1mo ago
Evaluation Results
Method
Method
Links
ASR
SafeThinker
Prefix length=10
2026.01
11.5
SafeThinker
Prefix length=40
2026.01
12.4
SafeThinker
Prefix length=20
2026.01
13.6
No Defense
Prefix length=10
2026.01
32.1
No Defense
Prefix length=20
2026.01
32.1
SafeDecoding
Prefix length=20
2026.01
32.1
SafeDecoding
Prefix length=40
2026.01
32.7
No Defense
Prefix length=40
2026.01
33
SafeDecoding
Prefix length=10
2026.01
33
Feedback
Search any
task
Search any
task