ResearchBenchmarksMathematical Reasoning on AIME 2025 (Avg@30, Token Counts)Follow82.2Avg Score (AIME 2025)DyCon36.02448.0126071.988Jun 5, 2026Evaluation ResultsMethodMethodLinksAvg Score (AIME 2025)Std Dev (Avg Score)Token CountStd Dev (Token Count)DyConBackbone=Qwen3-4B-Thin...Backbone=Qwen3-4B-Thinking-25072026.0682.23.9819,820863BaselineBackbone=Qwen3-4B-Thin...Backbone=Qwen3-4B-Thinking-25072026.0680.25.0122,553958DyConBackbone=DeepSeek-R1-D...Backbone=DeepSeek-R1-Distill-Qwen-7B2026.0637.94.5911,239971BaselineBackbone=DeepSeek-R1-D...Backbone=DeepSeek-R1-Distill-Qwen-7B2026.0637.84.3314,4711,175