Differentiable pruning combined with combinatorial optimization enhances structured neural network pruning.
Efficiently prune neural networks with FLOP and sparsity constraints using the FALCON framework.