B D
B D
[Chapter 17] Question : Does an autoencoder based on advanced CNN such as SE-ResNET make any sense ?
Just so you know, with a Dense only decoder below, ``` denoising_decoder = keras.models.Sequential([ keras.layers.Dense(100, activation="selu", input_shape=[30]), keras.layers.Dense(28 * 28, activation="sigmoid"), keras.layers.Reshape([28, 28]) ]) ``` I get this _not so...
[Chapter 17] Question : Does an autoencoder based on advanced CNN such as SE-ResNET make any sense ?
Yes, That answers my question. Thank you. I was mainly interested in the impact on the decoder part only (due to a SE-ResNet-like encoder), but your answer applies also. More...
Hi, In fact, I am using the French version of the book, and in this one, the figure is different :  Seems like an issue only on the French...
Here is a 3-epoch (only) pre-trained Char-RNN that took 8 hours to proceed with my hardware, if anyones needs it.... [model_charRNN.zip](https://github.com/ageron/handson-ml2/files/5975501/model_charRNN.zip)