ResearchBenchmarksMathematical Problem Solving on MinervaFollow24.6Average Score (@32)MASPO22.5223.0623.624.14Feb 19, 2026Evaluation ResultsMethodMethodLinksAverage Score (@32)Pass Rate (@32)MASPOBackbone=DeepSeek-R1-D...Backbone=DeepSeek-R1-Distill-Qwen-14B2026.0224.641.4GRPOBackbone=DeepSeek-R1-D...Backbone=DeepSeek-R1-Distill-Qwen-14B2026.0222.635.3