https://arxiv.org/abs/1602.07868
CIFAR-10
3-layers CNN. every layer-output will be BatchNormalized.
every CNN and Linear (Dense) layer's weights are normalized.
that is
parameters a scalar
Weight Normalization: A Simple Reparameterization to Accelerate Training of Deep Neural Networks https://arxiv.org/abs/1602.07868
https://arxiv.org/abs/1602.07868
CIFAR-10
3-layers CNN. every layer-output will be BatchNormalized.
every CNN and Linear (Dense) layer's weights are normalized.
that is
parameters a scalar
Weight Normalization: A Simple Reparameterization to Accelerate Training of Deep Neural Networks https://arxiv.org/abs/1602.07868