ResearchBenchmarksMathematical Reasoning on Math-500, CollegeMath, AIME24, Minerva AverageFollow28.56Average AccuracyZeroSiam24.462425.526226.5927.6538Sep 27, 2025Evaluation ResultsMethodMethodLinksAverage AccuracyZeroSiamModel=Llama3.1-8BModel=Llama3.1-8B2025.0928.56COMEModel=Llama3.1-8BModel=Llama3.1-8B2025.0926.16TLMModel=Llama3.1-8BModel=Llama3.1-8B2025.0925.44TentModel=Llama3.1-8BModel=Llama3.1-8B2025.0925.36SARModel=Llama3.1-8BModel=Llama3.1-8B2025.0925.36EATAModel=Llama3.1-8BModel=Llama3.1-8B2025.0925.22Llama3.1-8BModel=Llama3.1-8BModel=Llama3.1-8B2025.0924.62