Extracting Information about Latent Space

[AlexandrosEmvoliadis]

I have trained the PDEAutoEncoder on real-world spatiotemporal data, and i would like to extract information about the latent space (mean and std for each latent variable). On top of that, i want to sample from these distributions and utilize pde_decoder_only, to visualize the obtained results.

When using the pde_decoder_only, an assertion error rises, as i pass as input the dataset that the model i trained on. How should i modify the code to:

1. visualize the characteristics of the latent variables' distributions
2. manipulate the latent space, using pde_decoder_only