Share your thoughts, 1 month free Claude Pro on usSee more

Jailbreak Defense on Manual (IJP)

0ASR

Prompt Guard

Updated 15d ago

Evaluation Results

Method	Links
Prompt Guard 2026.04		0	0
Prompt Guard 2026.04		0	0
Prompt Guard 2026.04		0	0
WildGuard 2026.04		0	3.4
GuardReasoner 2026.04		0	0.9
SelfGrader 2026.04		0	0
SelfGrader 2026.04		0.2	9.1
WildGuard (Pre) 2026.04		0.4	3.3
SelfGrader 2026.04		0.5	6.3
GradientCuff 2026.04		0.8	3.6
GuardReasoner (Pre) 2026.04		0.9	7.8
SelfDefend 2026.04		0.9	27
SelfDefend 2026.04		1.7	25.3
GuardReasoner 2026.04		1.8	51.9
SelfDefend (Direct) 2026.04		2	28.9
SelfDefend (Intent) 2026.04		2	25.7
WildGuard (Post) 2026.04		2	86.2
WildGuard 2026.04		2	56.9
GuardReasoner (Post) 2026.04		2.2	86.2
SelfDefend 2026.04		2.4	29.1
Token Highlighter 2026.04		3.6	10
SelfDefend 2026.04		5.4	26
SelfDefend 2026.04		5.7	29.1
Llama Guard (Pre) 2026.04		6	56.1
Llama Guard (Post) 2026.04		6.1	96.6
Llama Guard 2026.04		6.4	62.4
LLama-3-8B-Instruct (No Defense) 2026.04		7.8	-
Perplexity Filter 2026.04		7.8	100
Llama Guard 2026.04		8.1	56.1
GradientCuff 2026.04		10.2	27.7
Llama Guard 2026.04		10.5	79.2
Llama Guard 2026.04		19.4	56.1
Llama Guard 2026.04		25	68.4
SelfGrader 2026.04		25.1	55.2
Qwen2.5-7B-Instruct (No Defense) 2026.04		25.7	-
Perplexity Filter 2026.04		25.7	100
Vicuna-13B-v1.5 2026.04		47.4	-
Perplexity Filter 2026.04		47.4	100