List your company

Jailbreak Detection on GoalFrameBench harmful Llama3-8B 2025 (seed prompts)

0.96AccuracyFrameShield-Last
0.1280.3440.560.776Feb 23, 2026

Evaluation results

MethodLinks
FrameShield-LastBackbone=Llama3-8B2026.020.960.86
JBShieldBackbone=Llama3-8B2026.020.770.86
LlamaGuardBackbone=Llama3-8B2026.020.60.75
GradSafeBackbone=Llama3-8B2026.020.370.54
SelfExBackbone=Llama3-8B2026.020.160.26