ResearchBenchmarksMathematical Reasoning on APEXFollow4.59Mean@16ME-ICPO2.46843.01923.574.1208Mar 2, 2026Evaluation ResultsMethodMethodLinksMean@16AccME-ICPOModel=Qwen2.5-Math-7BModel=Qwen2.5-Math-7B2026.034.594.57BaseModel=Qwen2.5-Math-7BModel=Qwen2.5-Math-7B2026.032.552.61