List your company

Mathematical Reasoning on MATH, GSM, and AIME Suite

80.1MATH ScoreCARE-RL
60.13265.31670.575.684May 30, 2026

Evaluation results

MethodLinks
CARE-RLBackbone=Qwen3-4B2026.0580.194.314.262.9
CARE-RLBackbone=Qwen2.5-7B2026.0579.793.29.260.7
MOPDBackbone=Qwen3-4B2026.0579.493.911.761.7
NV→VBackbone=Qwen3-4B2026.057993.614.262.3
MGSBackbone=Qwen3-4B2026.0578.893.713.361.9
MOPDBackbone=Qwen2.5-7B2026.0578.692.77.559.6
NV→VBackbone=Qwen2.5-7B2026.0578.292.28.359.6
Naive MixingBackbone=Qwen3-4B2026.0578.1939.260.1
V→NVBackbone=Qwen3-4B2026.0577.692.610.860.3
MGSBackbone=Qwen2.5-7B2026.0577.4927.559
Naive MixingBackbone=Qwen2.5-7B2026.0576.291.45.857.8
V→NVBackbone=Qwen2.5-7B2026.0575.990.85.857.5
BaseBackbone=Qwen3-4B2026.0568.487.84.253.5
BaseBackbone=Qwen2.5-7B2026.0560.984.74.249.9