Ugly and poorer tensor visualisation with v0.4.0 #6912
Description
numpy deals well with printing arrays.
In [1]: np.array([1, 2, 3])
Out[1]: array([1, 2, 3])
In [2]: print(np.array([1, 2, 3]))
[1 2 3]torch no longer does.
In [3]: torch.tensor([1, 2, 3])
Out[3]: tensor([ 1, 2, 3])
In [4]: print(torch.tensor([1, 2, 3]))
tensor([ 1, 2, 3]) 1 2 3
4 5 6
[torch.FloatTensor of size 2x3]
# was much cleaner (no brackets and commas) and informative (`dtype` and `size`) than
tensor([[ 1, 2, 3],
[ 4, 5, 6]])