Ankit Sharma

I was looking at the implementation of prior_expert and found the following line : logvar = Variable(torch.log(torch.ones(size))) If it's going to represent N(0,1), logvar should be 0 since variance is...