Open jdkealy opened 6 years ago
Hi!
I am trying to find some information on the meaning of the below and how to tweak for my project.
(def MNIST-NETWORK [(layers/input 28 28 1 :id :data) (layers/convolutional 5 0 1 20 :weights {:l2-regularization 1e-3}) (layers/max-pooling 2 0 2) (layers/batch-normalization :mode :spatial) (layers/dropout 0.9) (layers/relu) (layers/convolutional 5 0 1 50) (layers/max-pooling 2 0 2) (layers/batch-normalization :l1-regularization 1e-4) (layers/linear 500 :l2-max-constraint 4.0) (layers/relu :center-loss {:label-indexes {:stream :label} :label-inverse-counts {:stream :label} :labels {:stream :label} :alpha 0.9 :lambda 1e-2}) (layers/linear 10) (layers/softmax :id :label)])
Is there any tutorial or documentation on this ?
Hi!
I am trying to find some information on the meaning of the below and how to tweak for my project.
Is there any tutorial or documentation on this ?