Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Stealthiness Evaluation on Harmful prompts (evaluated on 3 LLMs and 4 guard LLMs)
Loading...
3.23
Mean Perplexity
ArtPrompt
-29.0276
188.7112
406.45
624.1888
Oct 2, 2024
Mean Perplexity
PPL Std Dev
Updated 16d ago
Evaluation Results
Method
Method
Links
Mean Perplexity
PPL Std Dev
ArtPrompt
2024.10
3.23
1.89
Ascii
2024.10
4.13
0.49
base64
2024.10
10.46
3.06
Morse Cipher
2024.10
11.81
2.19
ReNeLLM
2024.10
15.56
5.69
Unicode
2024.10
42.19
33.57
UTF-8
2024.10
42.19
33.57
Origin
2024.10
49.9
51.63
Caesar Cipher
2024.10
258.1
182.96
FlipAttack
2024.10
809.67
506.4
Feedback
Search any
task
Search any
task