Sparse Neural Networks

Keep the Gradients Flowing: Using Gradient Flow to Study Sparse Network Optimization
Training sparse networks to converge to the same performance as dense neural architectures has proven to be elusive. Recent work …
PDF Cite