Recurrent AutoEncoder seems not to learn after a number of timesteps Hoi, I am struggling with building a LSTM autoencoder and I cannot get it to work.