I am building neural network using denoising stacked autoencoders. I train autoencoder and then I would like to take the matrix of weights W and copy/initialize/clone it's values into new variable which is used in supervised optimization. How can I do such thing?
.initialized_value() doesn't work for me :/