the use of 'self.embed' in deepVARNetwork

[ShirleyMgit]

Hi,
I am trying to extend the network with an embedding layer for static features (similarly to deepAR).
Therefore, I am trying to understand the code.
Can you please explain what is the role of 'self.embed' and repeated_index_embeddings?
thanks

[mbohlkeschneider]

Hi @ShirleyMgit,

Thank you for your question. The embedding that is implemented isn't really useful for DeepVAR but is used in CPVAR (code), which is derived from DeepVAR. Here, we sample the dimensions of the multivariate dataset and project into the low-rank covariance matrix (see this paper). The reason it is there for DeepVAR is merely to reduce code duplication for CPVAR. Hope that was clear.