List your company

Mathematical Reasoning on GSM8K (Accuracy, Micro avg.)

89.31Accuracy (GSM8K)TraceLift
13.691633.323352.95572.5867May 5, 2026

Evaluation results

MethodLinks
TraceLiftModel=Qwen2.5-7B2026.0589.3170.19
TraceLiftModel=Qwen2.5-14B2026.0589.1672.25
Exec-onlyModel=Qwen2.5-14B2026.0588.9371.93
BaseModel=Qwen2.5-14B2026.0588.4871.2
Exec-onlyModel=Qwen2.5-7B2026.0587.2665.62
BaseModel=Qwen2.5-7B2026.0561.7946.51
TraceLiftModel=Llama3.1-8B2026.0532.922.86
Exec-onlyModel=Llama3.1-8B2026.0531.9222.19
BaseModel=Llama3.1-8B2026.0516.612.74