DanDoesData MadeUp or Startup? #2: Convolutional Sequence
We change up the LSTM model to a 1D convolution of fixed output size to better match what we want to put into the GAN. No training here, but we put together the generator model as well as a simple encoder to pretrain it a little bit. Next week some actual training and hopefully developing the discriminator model. -- Watch live at / vanboxel
