ResearchBenchmarksMathematical Reasoning on AMC 2023 (test) (Accuracy and Token Efficiency)Follow90AccuracyRoT71.876.52581.2585.975Sep 26, 2025Evaluation ResultsMethodMethodLinksAccuracyAvg Output TokensRoTModel=DLER-7BModel=DLER-7B2025.09901,595.75RoT+TIModel=DLER-7BModel=DLER-7B2025.09901,077.53RoTModel=DeepSeek-R1-Llama8BModel=DeepSeek-R1-Llama8B2025.0987.55,419.4CoTModel=DLER-7BModel=DLER-7B2025.09851,302.16CoTModel=DeepSeek-R1-Llama8BModel=DeepSeek-R1-Llama8B2025.0977.54,372.45RoT+TIModel=DeepSeek-R1-Llama8BModel=DeepSeek-R1-Llama8B2025.0972.52,949.72