List your company

Mathematical Reasoning on AIME out-of-domain evaluation 2025

45.1Average Score @32GRPO
11.414420.159728.90537.6503Jun 15, 2026

Evaluation results

MethodLinks
GRPOModel=Qwen3-4B2026.0645.1
GSPOModel=Qwen3-4B2026.0642.08
DAPOModel=Qwen3-4B2026.0639.06
WAPOModel=Qwen3-4B2026.0638.43
DAPOModel=SmolLM3-3B2026.0627.18
WAPOModel=SmolLM3-3B2026.0626.77
GSPOModel=SmolLM3-3B2026.0626.04
GRPOModel=SmolLM3-3B2026.0625.93
GSPOModel=Gemma3-4B2026.0614.27
DAPOModel=Gemma3-4B2026.0613.75
GRPOModel=Gemma3-4B2026.0613.54
WAPOModel=Gemma3-4B2026.0612.71