As Transformer models continue to grow in size and complexity, numerous high-fidelity pruning methods have been proposed to mitigate the increasing parameter count. However, transforming these ...