List your company

Mathematical Reasoning on AMC (test) (Pass@4)

74.23Pass@4GRPO-full
44.912452.523760.13567.7463May 9, 2026

Evaluation results

MethodLinks
GRPO-fullModel=Qwen3-8B2026.0574.2347.46
GRPO+IMAXModel=Qwen3-4B2026.0571.0638.07
GRPO-prefixModel=Qwen3-4B2026.0570.6439.64
GRPO-prefixModel=Qwen2.5-1.5B2026.0569.2235.38
DAPO-prefixModel=Qwen2.5-1.5B2026.0568.9735.53
BaseModel=Qwen2.5-1.5B2026.0568.2235.28
GRPO-fullModel=Qwen2.5-1.5B2026.0546.0422.79