List your company

Mathematical reasoning on OlympiadBench (avg@4, Average)

74.5Avg@4 ScoreDC-GRPO
69.92471.11272.373.488Jun 7, 2026

Evaluation results

MethodLinks
DC-GRPOBackbone=Qwen3-30B-A3B, Precision=BF162026.0674.571.8
DC-GRPOBackbone=Qwen3-8B, Precision=BF162026.0674.471
GRPOBackbone=Qwen3-30B-A3B, Precision=BF162026.0671.369
GRPOBackbone=Qwen3-8B, Precision=BF162026.0670.167.3