Our new X account is live! Follow @wizwand_team for updates
Home
/
Benchmarks
Confidence Calibration on GPQA (test)
Loading...
0.102
ECE
HTC
0.08792
0.18296
0.278
0.37304
Jan 22, 2026
ECE
Brier Score
AUROC
Updated 4d ago
Evaluation Results
Method
Method
Links
ECE
Brier Score
AUROC
HTC
Variant=Reduced
2026.01
0.102
0.213
0.706
HTC
Variant=Full
2026.01
0.124
0.219
0.704
GlobalTrace-TP
Temperature Scaling=true
2026.01
0.136
0.257
0.643
LastStep-TP
Temperature Scaling=true
2026.01
0.139
0.258
0.61
GlobalTrace-TP
Temperature Scaling=false
2026.01
0.414
0.402
0.649
LastStep-TP
Temperature Scaling=false
2026.01
0.424
0.413
0.614
Verbalized Conf
2026.01
0.454
0.523
0.593
Feedback
Search any
task
Search any
task