List your company

Mathematical Reasoning on AIME 2024 (Performance %)

12.92Performance (%)CARE
6.20167.94589.6911.4342May 25, 2026

Evaluation results

MethodLinks
CAREBackbone=Phi4-mini-instruct2026.0512.92
GTBackbone=Phi4-mini-instruct2026.0511.56
OracleBackbone=Phi4-mini-instruct2026.0510.93
EntropyBackbone=Phi4-mini-instruct2026.058.65
TTRLBackbone=Phi4-mini-instruct2026.058.44
ProbBackbone=Phi4-mini-instruct2026.058.02
RandomBackbone=Phi4-mini-instruct2026.057.71
VanillaBackbone=Phi4-mini-instruct2026.056.46