List your company

Mathematical Reasoning on Minerva (ACC, mean@8)

66.78ACCBase model
15.206428.595741.98555.3743Apr 15, 2026

Evaluation results

MethodLinks
Base modelBackbone=Qwen3-8B-Base2026.0466.7866.78
GRPOBackbone=Qwen3-8B-Base2026.0448.6248.62
DAPOBackbone=Qwen3-8B-Base2026.0448.0248.02
CDEBackbone=Qwen3-8B-Base2026.0447.7547.75
DiPOBackbone=Qwen3-8B-Base2026.0447.7547.75
DAPO w/ ELBackbone=Qwen3-8B-Base2026.0447.5647.56
DiPOBackbone=Qwen3-4B-Base2026.0444.7644.76
DAPO w/ ELBackbone=Qwen3-4B-Base2026.0444.5344.53
GRPOBackbone=Qwen3-4B-Base2026.0444.3944.39
DAPOBackbone=Qwen3-4B-Base2026.0444.3444.34
CDEBackbone=Qwen3-4B-Base2026.0443.1143.11
DAPO w/ ELBackbone=Qwen2.5-7B2026.0439.6539.65
DiPOBackbone=Qwen2.5-7B2026.0437.5937.59
DAPOBackbone=Qwen2.5-7B2026.0437.5537.55
GRPOBackbone=Qwen2.5-7B2026.0436.9536.95
CDEBackbone=Qwen2.5-7B2026.0435.9435.94
Base modelBackbone=Qwen3-4B-Base2026.0424.8224.82
Base modelBackbone=Qwen2.5-7B2026.0417.1917.19