Pruning with Pytorch

Hi,

Is there any way to take advantage of unstructured pruning in Pytorch currently ? It does not seem to reduce the model size or the inference time, just zero-ing some coefficients but in the end it is useless.