List your company

Mathematical Reasoning on MMATH (test)

37.4Ar ScoreQwen2.5-32B-Instruct + LANG
2.0411.2220.429.58May 21, 2026

Evaluation results

MethodLinks
Qwen2.5-32B-Instruct + LANGMetric Type=Acc2026.0537.43743.544.138.542.540.539.838.242.742.840.940.6——
Qwen2.5-32B-Instruct + LC-GRPOMetric Type=Acc2026.0536.738.239.537.435.440.337.940.339.139.939.339.638.6——
Qwen2.5-32B-InstructMetric Type=Acc2026.0526.826.627.927.126.734.728.32826.330.929.628.728.5——
Qwen2.5-32B-InstructMetric Type=LC2026.0526.226.127.927.124.634.727.827.625.830.829.628.428——
Qwen2.5-32B-Instruct + Vanilla GRPOMetric Type=Acc2026.051016.539.635.92539.527.829.11.120.518.617.323.6——
Llama3.1-8B-InstructMetric Type=Acc2026.057.41013.19.31115.21110.88.911.712.71111——
Llama3.1-8B-Instruct + LANGMetric Type=Acc2026.057.49.914.38.612.216.711.511.19.312.412.911.411.5——
Llama3.1-8B-InstructMetric Type=LC2026.057.29.7137.99.315.210.410.88.411.212.410.710.5——
Llama3.1-8B-Instruct + Vanilla GRPOMetric Type=Acc2026.053.9763.52.69.75.56.23.35.56.95.55.5——
Llama3.1-8B-Instruct + LC-GRPOMetric Type=Acc2026.053.48.86.98.86.25.86.674.96.276.36.5——