Mathematical Reasoning on MATH level 1
79AccuracyGradient Shift
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Gradient Shiftvariant=hybrid2026.01 | 79 | 97 | -0.16 | |
| Gradient Shiftvariant=mult2026.01 | 73 | 96 | -0.09 | |
| Weak-to-Strongalpha=2.0, k=402026.01 | 72 | 88 | -0.17 | |
| Weak-to-Strongalpha=1.0, k=402026.01 | 70 | 89 | -0.12 | |
| Benchmark LLM (BenchAcc)2026.01 | 70 | — | — | |
| Weak-to-Strongalpha=1.5, k=402026.01 | 69 | 90 | -0.1 | |
| Helper LLM2026.01 | 61 | — | — | |
| Gradient Shiftvariant=add2026.01 | 57 | 74 | -0.1 |