List your company

Mathematical Reasoning on GSM8K (test) (Pass@4, Avg@4)

94.47Pass@4 ScoreGRPO+IMAX
78.29882.496586.69590.8935May 9, 2026

Evaluation results

MethodLinks
GRPO+IMAXModel=Qwen3-4B2026.0594.4776.14
GRPO+IMAXModel=Qwen3-8B2026.0594.0164.18
DAPO+IMAXModel=Qwen3-4B2026.0593.7872.93
GRPO-prefixModel=Qwen3-4B2026.0593.4871.97
DAPO-prefixModel=Qwen3-4B2026.0593.1872.02
MERCI-prefixModel=Qwen3-8B2026.0593.1859.59
DAPO+IMAXModel=Qwen3-8B2026.0592.7260.2
GRPO-fullModel=Qwen3-4B2026.0591.8186.64
GRPO-prefixModel=Qwen3-8B2026.0591.7860.9
GRPO-fullModel=Qwen3-8B2026.0591.5876.95
DIVER-prefixModel=Qwen3-4B2026.0591.4366.49
MERCI-prefixModel=Qwen3-4B2026.0590.8362.51
DIVER-prefixModel=Qwen3-8B2026.0590.7557.15
BaseModel=Qwen3-4B2026.0590.659.7
Text Init.Model=Qwen3-8B2026.0590.356.16
DAPO-prefixModel=Qwen3-8B2026.0590.0752.12
Text Init.Model=Qwen3-4B2026.0588.5553.49
BaseModel=Qwen3-8B2026.0588.155.44
DAPO-prefixModel=Qwen2.5-1.5B2026.0586.2869.5
GRPO-prefixModel=Qwen2.5-1.5B2026.0585.7566.89
GRPO+IMAXModel=Qwen2.5-1.5B2026.0585.7567.4
DAPO+IMAXModel=Qwen2.5-1.5B2026.0584.9967.63
GRPO-fullModel=Qwen2.5-1.5B2026.0584.8474.09
MERCI-prefixModel=Qwen2.5-1.5B2026.0582.1163.34
Text Init.Model=Qwen2.5-1.5B2026.0580.2956.92
DIVER-prefixModel=Qwen2.5-1.5B2026.0580.2959.76
BaseModel=Qwen2.5-1.5B2026.0578.9253.43