List your company

Mathematical Reasoning on MATH 500 (Accuracy and Micro Average)

69.4AccuracyTraceLift
7.83223.81639.855.784May 5, 2026

Evaluation results

MethodLinks
TraceLiftModel=Qwen2.5-14B2026.0569.472.25
Exec-onlyModel=Qwen2.5-14B2026.0568.671.93
BaseModel=Qwen2.5-14B2026.056771.2
TraceLiftModel=Qwen2.5-7B2026.0555.270.19
Exec-onlyModel=Qwen2.5-7B2026.055465.62
BaseModel=Qwen2.5-7B2026.0542.446.51
TraceLiftModel=Llama3.1-8B2026.0515.422.86
Exec-onlyModel=Llama3.1-8B2026.0514.822.19
BaseModel=Llama3.1-8B2026.0510.212.74