NNCompression

Deep neural networks (DNNs) may contain millions, even billions, of parameters/weights, making storage and computation very expensive and motivating a large body of work aimed at reducing their complexity by using, e.g., sparsity-inducing …