Skip to content
ai

Pruning

Neural Network Pruning

Definition

Pruning reduces model size by removing weights, neurons, attention heads, or layers that contribute minimally to model performance. Structured pruning removes entire components (suitable for hardware acceleration), while unstructured pruning creates sparse weight matrices.

Pruned models typically require retraining or distillation to recover accuracy and can be further compressed with quantization.

Can this ship? Are we exposed? Can you prove it?

Every CISO is asking these three questions. We answer them with facts, not guesses, from repository to runtime.