AI Inference & Serving
Structured Pruning
Removing organized model components, such as complete channels or filters, rather than scattered individual weights. The structure of connected computations may need corresponding changes. A reduced component count does not itself establish preserved task quality or a runtime improvement; both depend on what was removed and how the resulting model executes.
Reviewed
Sources
Free complete lesson
This definition and the complete practical lesson are free. The interactive reader loads the lesson examples without a membership.