List your company

Math reasoning on MATH500 (Coverage and Success Rate)

86CoverageTOFU
70.81674.75878.782.642Apr 30, 2026

Evaluation results

MethodLinks
TOFUBackbone=QWEN-2.5-MATH-7B2026.048653.8
CEBackbone=QWEN-2.5-MATH-7B2026.0484.854.2
GEMBackbone=QWEN-2.5-MATH-7B2026.0483.259
TOFUBackbone=QWEN-2.5-MATH-1.5B2026.0480.650.3
CEBackbone=QWEN-2.5-MATH-1.5B2026.0478.453.9
GEMBackbone=QWEN-2.5-MATH-1.5B2026.0478.453.4
TOFUBackbone=DEEPSEEK-MATH-7B2026.0472.633
GEMBackbone=DEEPSEEK-MATH-7B2026.0472.236.4
CEBackbone=DEEPSEEK-MATH-7B2026.0471.432.7