List your company

Mathematical Computation on MathQA

52.34Exact Match (EM)Prompt-R1
16.595225.875135.15544.4349Nov 2, 2025

Evaluation results

MethodLinks
Prompt-R12025.1152.3461.59
CoT ReasoningBackbone=GPT-4o-mini2025.1149.2257.03
GRPOBackbone=Qwen3-4B2025.1146.8854.43
BaselineBackbone=GPT-4o-mini2025.1146.0954.04
TextGradCategory=APO (GPT-4o-mini)2025.1144.5361.46
OPROCategory=APO (GPT-4o-mini)2025.1143.7560.08
GEPACategory=APO (GPT-4o-mini)2025.1140.6361.59
BaselineBackbone=Qwen3-4B2025.1128.9132.29
CoT ReasoningBackbone=Qwen3-4B2025.1127.3430.6
SFTBackbone=Qwen3-4B2025.1117.9722.66