Large Model Performance Prediction on Benchmark-side Pattern Shift Math
46.59Average ScoreCPMF
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| CPMF2026.02 | 46.59 | 36.45 | -10.14 | — | — | — | — | — | |
| CPMFShift Type=Benchmark-side, Scenario=Math2026.02 | 46.59 | 36.45 | — | 48.63 | 44.55 | 53.89 | 52.63 | 2.82 | |
| PMF2026.02 | 42.08 | 40.36 | -1.72 | — | — | — | — | — | |
| PMFShift Type=Benchmark-side, Scenario=Math2026.02 | 42.08 | 40.36 | — | 44.21 | 39.94 | 61.33 | 57.64 | 2.11 | |
| STAR2026.02 | 13.01 | 60.66 | 47.65 | — | — | — | — | — | |
| STARShift Type=Benchmark-side, Scenario=Math2026.02 | 13.01 | 60.66 | — | 14.7 | 11.31 | 86.09 | 77.18 | 18.7 |