FLOPs as a Direct Optimization Objective for Learning Sparse Neural Networks

Open in new window