Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Attack Detection on OverThink N = 100
Loading...
100
TPR (OverThink N=100)
Token Budget
97.92
98.46
99
99.54
Jun 6, 2026
TPR (OverThink N=100)
Updated 1mo ago
Evaluation Results
Method
Method
Links
TPR (OverThink N=100)
Token Budget
Backbone Model=DS-R1-Q...
2026.06
100
Compression-Ratio Proxy
Backbone Model=DS-R1-Q...
2026.06
100
Output-Length Anomaly
Backbone Model=DS-R1-Q...
2026.06
100
QDM
Backbone Model=DS-R1-Q...
2026.06
100
RECURGUARD
Backbone Model=DS-R1-Q...
2026.06
99
Answer-Absence Proxy
Backbone Model=DS-R1-Q...
2026.06
98
Feedback
Search any
task
Search any
task