List your company

Mathematical Reasoning on MATH level 1

79AccuracyGradient Shift
56.1262.066873.94Jan 31, 2026

Evaluation results

MethodLinks
Gradient Shiftvariant=hybrid2026.017997-0.16
Gradient Shiftvariant=mult2026.017396-0.09
Weak-to-Strongalpha=2.0, k=402026.017288-0.17
Weak-to-Strongalpha=1.0, k=402026.017089-0.12
Benchmark LLM (BenchAcc)2026.0170——
Weak-to-Strongalpha=1.5, k=402026.016990-0.1
Helper LLM2026.0161——
Gradient Shiftvariant=add2026.015774-0.1