The loss function

[goodman1204]

I am confused about the loss function in the code, for different parts int the loss functions, it has different + or - symbols between the paper and the code? Can someone help to verify it?

[SanGilbert]

It is actually consistent between this two. In the paper the author gave ELBO form that should be maximized, thus this code turn over the sign to minimize the loss.