Norm. flows bug maybe

[kiwi0fruit]

On this line the code:

qz = log_gaussian(z, mu, log_var) - sum(log_det_z)

The first one and log_det_z are of batch size dim but the sum(log_det_z) is a scalar.

Actually the sum was already taken here.

It looks like a bug but I'm not sure.

[wohlert]

I believe you might be right, but I don't know what implications it will have to fix it. I will check on it a later time.

Thanks