List your company

Mathematical Reasoning on AIME25 (Accuracy, Generation Length)

25.6AccuracyDeepSeek-R1-1.5B
0.126.73513.3519.965Jan 7, 2026

Evaluation results

MethodLinks
DeepSeek-R1-1.5BReasoning Strategy=Long-CoT2026.0125.613,118
RPAMMethod Category=Data-dependent Merging2026.01204,393
CoDMethod Category=Prompt-guided2026.0116.78,913
Ada-R1Method Category=Training-based2026.0116.74,707
TIES MergingMethod Category=Data-free Merging2026.0116.72,554
ACMMethod Category=Data-dependent Merging2026.0116.75,535
Average MergingMethod Category=Data-free Merging2026.0113.33,735
AIMMethod Category=Data-dependent Merging2026.0113.33,199
Task ArithmeticMethod Category=Data-free Merging2026.013.33,713
DARE-LinearMethod Category=Data-free Merging2026.013.33,417
Qwen2.5-Math-1.5BReasoning Strategy=Short-CoT2026.011.11,119