Libraries for applying sparsification recipes to neural networks with a few lines of code, enabling faster and smaller models