ResearchBenchmarksMathematical Reasoning on MATH 500 (Pass@1, Pass@16, Major@16)Follow86.8Pass@1GSPO59.34466.47273.680.728May 18, 2026Evaluation ResultsMethodMethodLinksPass@1Pass@16Major@16GSPOBackbone=Qwen3-4BBackbone=Qwen3-4B2026.0586.893.290CorR-POBackbone=Qwen3-4BBackbone=Qwen3-4B2026.0586.592.489.8DAPOBackbone=Qwen3-4BBackbone=Qwen3-4B2026.0586.293.290.2GRPOBackbone=Qwen3-4BBackbone=Qwen3-4B2026.0583.49388.2Dr.GRPOBackbone=Qwen3-4BBackbone=Qwen3-4B2026.05799286.4BaseBackbone=Qwen3-4BBackbone=Qwen3-4B2026.0560.490.879.8