1D CNN VAE on pytorch for MNIST Dataset (25EUROS)
Budget: €8 – €30 EUR
• Complete the code for 1D CNN Variational autoencoder (1D-CNN VAE) using a notebook as seen in VAE_pytorch_custom notebook in the attached.
• Write and comment the meaning of the input of a 1D CNN and others used in pytorch and use the MNIST dataset for it.
• Plot the 2D latent space generated by training a 1D CNN VAE and ensure the latent space corresponds to that obtained for 1D CNN VAE of tensorflow (see attached).
• The first notebook is done for tensorflow and you can use ideas of the network structure for tensorflow to design your pytorch version.
• Write and comment the meaning of the input of a 1D CNN and others used in pytorch and use the MNIST dataset for it.
• Plot the 2D latent space generated by training a 1D CNN VAE and ensure the latent space corresponds to that obtained for 1D CNN VAE of tensorflow (see attached).
• The first notebook is done for tensorflow and you can use ideas of the network structure for tensorflow to design your pytorch version.