Abstract: Leveraging sparsity in deep neural network (DNN) models holds significant promise for accelerating model inference. However, current GPUs can only harness sparsity in model weights, leaving ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results