pruning

WoodFisher: Efficient second-order approximations for model compression