Indexing Bug in Torch 0.4.0?

Pytorch 0.4.0

a = torch.rand(10, 10)
a[:, :0] = 0

Then a 's first column becomes 0. Is this a bug?

Can confirm the syntax seems to have a different behavior than numpy.
In numpy a[:, :0] gives: array([], shape=(10, 0), dtype=float64), while pytorch gives the first column.