#技能路由

This research introduces a more efficient way to route skills in large language models, potentially reducing the need for large external parameter counts.