Our new X account is live! Follow @wizwand_team for updates
Home
/
Benchmarks
Confidence Calibration on HLE (test)
Loading...
0.031
ECE
HTC
0.0048
0.18165
0.3585
0.53535
Jan 22, 2026
ECE
BS
AUROC
Updated 4d ago
Evaluation Results
Method
Method
Links
ECE
BS
AUROC
HTC
Variant=Reduced
2026.01
0.031
0.09
0.644
HTC
Variant=Full
2026.01
0.072
0.098
0.617
GlobalTrace-TP
Temperature Scaling=true
2026.01
0.433
0.277
0.57
LastStep-TP
Temperature Scaling=true
2026.01
0.436
0.278
0.628
Verbalized Conf
2026.01
0.656
0.531
0.614
GlobalTrace-TP
Temperature Scaling=false
2026.01
0.685
0.56
0.551
LastStep-TP
Temperature Scaling=false
2026.01
0.686
0.561
0.604
Feedback
Search any
task
Search any
task