Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Compiled-monitor replay on Stress benchmark
Loading...
54
True Positives (TP)
Gold
5.12
17.81
30.5
43.19
Jun 30, 2026
True Positives (TP)
False Negatives (FN)
False Positives (FP)
Precision (Prec)
Recall (Catch)
F1 Score (F1)
Updated 25d ago
Evaluation Results
Method
Method
Links
True Positives (TP)
False Negatives (FN)
False Positives (FP)
Precision (Prec)
Recall (Catch)
F1 Score (F1)
Gold
Type=Oracle
2026.06
54
16
4
93.1
77.1
84.4
LoRA
Model=Qwen3-8B
2026.06
41
29
7
85.4
58.6
69.5
LabGuard-Hybrid
2026.06
41
29
7
85.4
58.6
69.5
Regex
Type=Rule-based
2026.06
34
36
9
79.1
48.6
60.2
No Safety
Safety Monitoring=None
2026.06
7
63
1
87.5
10
17.9
Feedback
Search any
task
Search any
task