List your company

Adversarial Attack on SafeMT ATTACK 600

71.29Attack Success RateGPT-4o
13.185228.270143.35558.4399Feb 28, 2025

Evaluation results

MethodLinks
GPT-4oSteering Configuration=Original2025.0271.29
Llama3-70bSteering Configuration=Original2025.0267.58
Llama3-70bSteering Configuration=w/ steering2025.0216.08
GPT-4oSteering Configuration=w/ steering2025.0215.42