Fix AssertionError at training time in vae.py (closes #191) (#192)

The operator += will call the __iadd__ function, which triggers a type check and fails when a PyTorch Tensor is added to a SparseTensor. This commit fixes this problem by using the + operator, which calls the __add__ function.

examples/vae.py

@@ -585,7 +585,7 @@ def forward(self, sinput, gt_target):
         means, log_vars = self.encoder(sinput)
         zs = means
         if self.training:
-            zs += torch.exp(0.5 * log_vars.F) * torch.randn_like(log_vars.F)
+            zs = zs + torch.exp(0.5 * log_vars.F) * torch.randn_like(log_vars.F)
         out_cls, targets, sout = self.decoder(zs, gt_target)
         return out_cls, targets, sout, means, log_vars, zs