List your company

Mathematical Reasoning on AMC23 (ACC and mean@8)

71.23Accuracy (ACC)DiPO
20.800433.892746.98560.0773Apr 15, 2026

Evaluation results

MethodLinks
DiPOBackbone=Qwen3-8B-Base2026.0471.2371.23
DAPO w/ ELBackbone=Qwen3-8B-Base2026.0469.8769.87
GRPOBackbone=Qwen3-8B-Base2026.0469.2869.28
DAPOBackbone=Qwen3-8B-Base2026.0469.1269.12
CDEBackbone=Qwen3-8B-Base2026.0468.0768.07
DiPOBackbone=Qwen3-4B-Base2026.0463.763.7
DAPO w/ ELBackbone=Qwen3-4B-Base2026.0462.9562.95
CDEBackbone=Qwen3-4B-Base2026.0462.3562.35
DAPOBackbone=Qwen3-4B-Base2026.0461.961.9
GRPOBackbone=Qwen3-4B-Base2026.0460.2460.24
DiPOBackbone=Qwen2.5-7B2026.0460.0960.09
DAPOBackbone=Qwen2.5-7B2026.0459.9459.94
DAPO w/ ELBackbone=Qwen2.5-7B2026.0458.7358.73
GRPOBackbone=Qwen2.5-7B2026.0458.4358.43
CDEBackbone=Qwen2.5-7B2026.0455.8755.87
Base modelBackbone=Qwen3-8B-Base2026.0439.4639.46
Base modelBackbone=Qwen3-4B-Base2026.0431.4831.48
Base modelBackbone=Qwen2.5-7B2026.0422.7422.74