List your company

Stealthiness Evaluation on LLaMA Guard 2 8B

2.34PPL MeanArtPrompt
-33.9432210.9684455.88700.7916Oct 2, 2024

Evaluation results

MethodLinks
ArtPrompt2024.102.340.52
Ascii2024.107.251.04
Base642024.109.782.81
Morse Cipher2024.1016.783.38
ReNeLLM2024.1021.139
Unicode2024.1035.2533.33
UTF-82024.1035.2533.33
Origin2024.1041.8153.42
Caesar Cipher2024.10192.17161.42
FlipAttack2024.10909.42643.61