List your company

Scientific Reasoning on GPQA (FPR, Power)

-0.55FPRCriticism
-2.2689.328520.92532.5215Feb 15, 2026

Evaluation results

MethodLinks
CriticismCalibration Source=GSM8K2026.02-0.559.12
No InterventionCalibration Source=GSM8K2026.0200.12
Renewal process-based stoppingCalibration Source=GSM8K, Stopping Rule=Renewal2026.021.3724.94
Maxwise Conformal StoppingCalibration Source=GSM8K, Stopping Rule=Maxwise2026.022.5632.4
ConfidenceCalibration Source=GSM8K2026.0212.855.57
LengthCalibration Source=GSM8K2026.0242.432.14