List your company

Mathematical Reasoning on AIME 2025 (average@32)

60.42Average Score (32 samples)GRPO
2.679217.669632.6647.6504May 8, 2026

Evaluation results

MethodLinks
GRPOParams=4B2026.0560.42
SODParams=1.7B2026.0541.72
OPDParams=1.7B2026.0537.04
SODParams=0.6B2026.0526.13
OPSDgtParams=1.7B2026.0524.69
OPDParams=0.6B2026.0522.95
SFTParams=1.7B2026.0522.4
GRPOParams=1.7B2026.0521.67
OPSDhintParams=1.7B2026.0521.43
OPSDgtParams=0.6B2026.0517.04
OPSDhintParams=0.6B2026.0514.12
VanillaParams=0.6B2026.0512.81
VanillaParams=1.7B2026.058.96
SFTParams=0.6B2026.055.42
GRPOParams=0.6B2026.054.9