Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Safety Detection on XSTest (held-out)
Loading...
99.6
AUROC
Geometry-Lite
97.416
97.983
98.55
99.117
May 18, 2026
AUROC
Updated 13d ago
Evaluation Results
Method
Method
Links
AUROC
Geometry-Lite
Protocol=Leave-one-ben...
2026.05
99.6
MultiLayer-Linear
Protocol=Leave-one-ben...
2026.05
99
TaT-Disp-LSTM
Protocol=Leave-one-ben...
2026.05
98.8
MultiLayer-DIM
Protocol=Leave-one-ben...
2026.05
97.9
Best single-layer probe
Protocol=Leave-one-ben...
2026.05
97.5
Feedback
Search any
task
Search any
task