Mutable layer number for autoencoder

[zyfccc]

In retiarii, I wonder whether there is a way to mutate the layer number of an autoencoder with a fixed number of latent variables? For example, an autoencoder with one, two, or three hidden layers of varied number of neurons in each layer and a latent variable.

A code example would be nice.

Thanks

[Richard-gangan]

I have the same question. How can can I find the best layer number and the appropriate number of neurons. But I cant run the example of NNI now. Maybe we can communicate about it?