List your company

Mathematical Reasoning on AIME25 (ACC/maj@8)

33.33ACC/maj@8GRPO
2.1310.2318.3326.43Apr 15, 2026

Evaluation results

MethodLinks
GRPOBackbone=Qwen3-8B-Base2026.0433.33
DAPOBackbone=Qwen3-8B-Base2026.0433.33
DAPO w/ ELBackbone=Qwen3-8B-Base2026.0433.33
CDEBackbone=Qwen3-8B-Base2026.0433.33
DiPOBackbone=Qwen3-8B-Base2026.0433.33
DAPOBackbone=Qwen3-4B-Base2026.0430
DAPO w/ ELBackbone=Qwen3-4B-Base2026.0430
CDEBackbone=Qwen3-4B-Base2026.0430
DiPOBackbone=Qwen3-4B-Base2026.0430
GRPOBackbone=Qwen3-4B-Base2026.0426.67
GRPOBackbone=Qwen2.5-7B2026.0426.67
DAPOBackbone=Qwen2.5-7B2026.0423.33
DiPOBackbone=Qwen2.5-7B2026.0423.33
DAPO w/ ELBackbone=Qwen2.5-7B2026.0420
CDEBackbone=Qwen2.5-7B2026.0420
Base modelBackbone=Qwen3-8B-Base2026.0413.33
Base modelBackbone=Qwen3-4B-Base2026.046.67
Base modelBackbone=Qwen2.5-7B2026.043.33