I’ve understood that nn.Embedding can be used to associate an embedding to some categorical label corresponding to high-dimensional inputs such as images, for example during the training of a conditional generative model. What if my label is continuous instead?
Is there a way inside PyTorch to create a high-dimensional embedding of a continuous number, for example given that I want to train a generative model associating to each image representing a person’s face, his/her age.
1 Like