Our new X account is live! Follow @wizwand_team for updates
Home
/
Benchmarks
Prompt Injection Defense on AgentDojo Workspace suite v1 (test)
Loading...
0.375
CU
Repeat Prompt
0.219
0.2595
0.3
0.3405
Dec 21, 2024
CU
U
ASR
Updated 4d ago
Evaluation Results
Method
Method
Links
CU
U
ASR
Repeat Prompt
Target Model=GPT-3.5-t...
2024.12
0.375
0.3125
0
No Defense
Target Model=GPT-3.5-t...
2024.12
0.325
0.4042
0.42
Task Shield
Target Model=GPT-3.5-t...
2024.12
0.3
0.3458
0
Tool Filter
Target Model=GPT-3.5-t...
2024.12
0.275
0.3083
0
Delimiting
Target Model=GPT-3.5-t...
2024.12
0.25
0.3375
0.83
PI Detector
Target Model=GPT-3.5-t...
2024.12
0.225
0.2375
0.42
Feedback
Search any
task
Search any
task