List your company

Mathematical Reasoning on MATH (OOD) (Calibration Metrics)

71KL Div (alpha=1)GEB-arctanh(π − 1)
67.46468.38269.370.218Sep 27, 2025

Evaluation results

MethodLinks
GEB-arctanh(π − 1)Backbone=LLaMA-3-8B-SFT, Evaluation Protocol=Out-of-domain2025.097167.669.269.3
GEB-πBackbone=LLaMA-3-8B-SFT, Evaluation Protocol=Out-of-domain2025.0969.269.671.670.1
GEB-1/πBackbone=LLaMA-3-8B-SFT, Evaluation Protocol=Out-of-domain2025.0968.470.269.269.3
f-DPOBackbone=LLaMA-3-8B-SFT, Evaluation Protocol=Out-of-domain2025.0967.66969.268.6
FEBBackbone=LLaMA-3-8B-SFT, Evaluation Protocol=Out-of-domain2025.0967.668.668.668.3