List your company

Mathematical Reasoning on Minerva Math (avg@4, Average)

52.9Accuracy (avg@4)DC-GRPO
49.98850.74451.552.256Jun 7, 2026

Evaluation results

MethodLinks
DC-GRPOBackbone=Qwen3-30B-A3B, Precision=BF162026.0652.971.8
GRPOBackbone=Qwen3-30B-A3B, Precision=BF162026.065269
GRPOBackbone=Qwen3-8B, Precision=BF162026.065167.3
DC-GRPOBackbone=Qwen3-8B, Precision=BF162026.0650.171