ResearchBenchmarksLarge Model Performance Prediction on Benchmark Chinese pattern shiftFollow16.94RMSESTAR15.801623.485831.1738.8542Feb 12, 2026Evaluation ResultsMethodMethodLinksRMSEMAEAvg ScoreSRCC (%)KRCC (%)MAE@3%Rank Avg (%)Total Score/ErrorSTARShift Type=Benchmark-s...Shift Type=Benchmark-side, Scenario=Chinese2026.0216.9412.0714.5173.767.3512.0351.03—CPMFShift Type=Benchmark-s...Shift Type=Benchmark-side, Scenario=Chinese2026.0242.2836.4439.3675.0167.567.5450.04—PMFShift Type=Benchmark-s...Shift Type=Benchmark-side, Scenario=Chinese2026.0245.438.9142.1652.0851.42.6935.39—