Hi, I could not find the discussion section on the repo, so thought to raise question here. Sorry about that.
I was just curious what was the rationale for using the Latent force vector in the loss function? As I understand it, its job is to help train force module. But why not use the learned force feature vector f^{l} directly for it?
Hi, I could not find the discussion section on the repo, so thought to raise question here. Sorry about that.
I was just curious what was the rationale for using the Latent force vector in the loss function? As I understand it, its job is to help train force module. But why not use the learned force feature vector f^{l} directly for it?