Indexing with list of slices

1

Hi,
Is there any equivalent in torch to indexing with list of slices. The point is to be able to index without depending on the number of dimension. Example with numpy :

>>> x = torch.arange(0, 25).view(5, 5).numpy()
array([[  0.,   1.,   2.,   3.,   4.],
       [  5.,   6.,   7.,   8.,   9.],
       [ 10.,  11.,  12.,  13.,  14.],
       [ 15.,  16.,  17.,  18.,  19.],
       [ 20.,  21.,  22.,  23.,  24.]], dtype=float32)

>>> slicers=[slice(0,2), slice(0,2)]
>>> a[slicers]
array([[ 0.,  1.],
       [ 5.,  6.]], dtype=float32)

Thanks

2

I am not looking for a solution with slices specifically, just something that can index over dimensions, without dimension dependant writing.

3

Hi, just put your slices in a tuple and it will work:

>>> x = torch.arange(0, 25).view(5, 5)
  0   1   2   3   4
  5   6   7   8   9
 10  11  12  13  14
 15  16  17  18  19
 20  21  22  23  24
[torch.FloatTensor of size 5x5]

>>> slicers=(slice(0,2), slice(0,2))
>>> x[slicers]
 0  1
 5  6
[torch.FloatTensor of size 2x2]
1 Like
4

So simple, thanks @albanD :slight_smile: