List your company

Math Reasoning on GSM8K (Coverage and Average Success Rate)

88CoverageTOFU
71.77675.98880.284.412Apr 30, 2026

Evaluation results

MethodLinks
TOFUBackbone=QWEN-2.5-MATH-7B2026.048846.8
CEBackbone=QWEN-2.5-MATH-7B2026.0481.760.1
TOFUBackbone=QWEN-2.5-MATH-1.5B2026.0480.651.8
CEBackbone=QWEN-2.5-MATH-1.5B2026.0479.254.5
TOFUBackbone=DEEPSEEK-MATH-7B2026.047947.5
GEMBackbone=QWEN-2.5-MATH-7B2026.0476.261.5
CEBackbone=DEEPSEEK-MATH-7B2026.047648
GEMBackbone=QWEN-2.5-MATH-1.5B2026.0473.255.5
GEMBackbone=DEEPSEEK-MATH-7B2026.0472.451.2