Abstract: The computational efficiency of deep neural networks is limited by hardware accelerators and sparsity integration. Leveraging sparsity can result in significant computational savings by ...