torch.range is upper-bound inclusive, while python range and numpy arange are upper-bound exclusive

torch.range(a,b) includes b, while python.range(a,b) or numpy.arange(a,b) do not include b.
Not sure if this is a bug or voluntary, but this is confusing.

pytorch
```
In: torch.range(0,5)
Out: 0 1 2 3 4 5 [torch.FloatTensor of size 6]
```

python standard function
```
In: range(0,5)
Out: [0, 1, 2, 3, 4]

```
numpy
```
In: np.arange(0,5)
Out: array([0, 1, 2, 3, 4])

```