List your company

Mathematical Reasoning on GaoKao Math Cloze

78.4AccuracyCES
45.32853.91462.571.086May 19, 2026

Evaluation results

MethodLinks
CESBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0578.41,544
DAPOBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0577.62,546
Entropy ShapeBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0577.32,538
R1-7BBackbone=DeepSeek-R1-Distill-Qwen-7B2026.0577.12,722
CESBackbone=DeepSeek-R1-Distill-1.5B2026.05614,655
DAPOBackbone=DeepSeek-R1-Distill-1.5B2026.0546.65,448