Abstract: Structural pruning of neural network parameters reduces computational, energy, and memory transfer costs during inference. We propose a novel method that estimates the contribution of a ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results
Feedback