List your company

Mathematical Reasoning on MATH-500 (Correction Uplift %)

52.94Correction Uplift (%)ROSA
4.590417.142729.69542.2473Sep 27, 2025

Evaluation results

MethodLinks
ROSAModel=Qwen3-8B2025.0952.94
ROSAModel=Qwen3-0.6B2025.0951.31
ROSAModel=Qwen2.5-7B-Instruct2025.0936.91
BaselineModel=Qwen3-8B2025.0924.54
ROSAModel=Qwen2.5-0.5B-Instruct2025.0924.05
BaselineModel=Qwen3-0.6B2025.0917.78
ROSAModel=DeepSeek-R1-Distill-Llama-8B2025.0917.41
BaselineModel=Qwen2.5-7B-Instruct2025.0913.65
BaselineModel=Qwen2.5-0.5B-Instruct2025.096.79
BaselineModel=DeepSeek-R1-Distill-Llama-8B2025.096.45