List your company

Mathematical Reasoning on MATH 500 (pass@1 and pass@32)

92.16Pass@1FG-ExPO
72.691277.745682.887.8544May 12, 2026

Evaluation results

MethodLinks
FG-ExPOBackbone=Qwen3-8B-Base2026.0592.1696.8
GRPOBackbone=Qwen3-8B-Base2026.0591.7896.8
FG-ExPOBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.0584.0396
GRPOBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.0582.8895
BaseBackbone=DeepSeek-R1-Distill-Qwen-1.5B2026.0573.4493.6