Fix arange in docs. (pytorch#273)

In [the tutorial](https://pytorch.org/tutorials/beginner/blitz/neural_networks_tutorial.html),
`torch.arange` is being used as a FloatTensor. This behavior was changed
in pytorch/pytorch#7016 to return a LongTensor.
I removed the `torch.arange` usage and instead construct a tensor with
`torch.randn`

Author: zou3519

beginner_source/blitz/neural_networks_tutorial.py

@@ -156,7 +156,7 @@ def num_flat_features(self, x):
 # For example:
 
 output = net(input)
-target = torch.arange(1, 11)  # a dummy target, for example
+target = torch.randn(10)  # a dummy target, for example
 target = target.view(1, -1)  # make it the same shape as output
 criterion = nn.MSELoss()