ResearchBenchmarksMathematical Reasoning on GSM-Symbolic-P2Follow82Pass@1 ScoreDiRL32.0845.045870.96Jun 9, 2026Evaluation ResultsMethodMethodLinksPass@1 ScoreDiRLModel=Qwen3-4BModel=Qwen3-4B2026.0682EVOL-RLModel=Qwen3-4BModel=Qwen3-4B2026.0678G2RLModel=Qwen3-4BModel=Qwen3-4B2026.0676DiRLModel=Qwen3-1.7BModel=Qwen3-1.7B2026.0647G2RLModel=Qwen3-1.7BModel=Qwen3-1.7B2026.0646EVOL-RLModel=Qwen3-1.7BModel=Qwen3-1.7B2026.0634