List your company

Mathematical Question Answering on MATH 500

0.0202Brier ScoreGUPO + SAGE
0.0051680.1066340.20810.309566Jun 9, 2026

Evaluation results

MethodLinks
GUPO + SAGEFramework=GUPO, Metric=SAGE2026.060.02020.02380.4630.6130.7170.74
GUPO + Kernel Language EntropyFramework=GUPO, Metric=Kernel Language Entropy2026.060.02050.03480.4510.5880.70.72
CSFTTraining=CSFT2026.060.0640.1430.50.6250.70.74
LoVeC-DPOTraining=LoVeC-DPO2026.060.1210.220.3520.60.6330.66
GUPO + Semantic EntropyFramework=GUPO, Metric=Semantic Entropy2026.060.190.2820.4260.5750.650.72
CoT + DCAStrategy=Chain-of-Thought, Alignment=DCA2026.060.2870.4790.0740.5880.5670.58
Direct VerbalizedStrategy=Direct Verbalized2026.060.3590.5460.2460.5630.6170.66
Direct Verbalized + DCAStrategy=Direct Verbalized, Alignment=DCA2026.060.3880.5780.1710.550.6170.58
CoTStrategy=Chain-of-Thought2026.060.3960.5820.1750.5880.6170.6