tensorplay.functional.dsplit

tensorplay.functional.dsplit(input, indices_or_sections) → List of Tensors[source]

Splits input, a tensor with three or more dimensions, into multiple tensors depthwise according to indices_or_sections. Each split is a view of input.

This is equivalent to calling tensorplay.tensor_split(input, indices_or_sections, dim=2) (the split dimension is 2), except that if indices_or_sections is an integer it must evenly divide the split dimension or a runtime error will be thrown.

This function is based on NumPy’s numpy.dsplit().

Parameters:

• input (Tensor) – tensor to split.

• indices_or_sections (int or list or tuple of ints) – See argument in tensorplay.tensor_split().

Example:

>>> t = tensorplay.arange(16.0).reshape(2, 2, 4)
>>> t
tensor([[[ 0.,  1.,  2.,  3.],
         [ 4.,  5.,  6.,  7.]],
        [[ 8.,  9., 10., 11.],
         [12., 13., 14., 15.]]])
>>> tensorplay.dsplit(t, 2)
(tensor([[[ 0.,  1.],
        [ 4.,  5.]],
       [[ 8.,  9.],
        [12., 13.]]]),
 tensor([[[ 2.,  3.],
          [ 6.,  7.]],
         [[10., 11.],
          [14., 15.]]]))

>>> tensorplay.dsplit(t, [3, 6])
(tensor([[[ 0.,  1.,  2.],
          [ 4.,  5.,  6.]],
         [[ 8.,  9., 10.],
          [12., 13., 14.]]]),
 tensor([[[ 3.],
          [ 7.]],
         [[11.],
          [15.]]]),
 tensor([], size=(2, 2, 0)))