mindspore.ops.transpose

mindspore.ops.transpose(input, input_perm)[source]

Permutes the dimensions of the input tensor according to input permutation.

For a 1-D array this has no effect, as a transposed vector is simply the same vector. To convert a 1-D array into a 2D column vector please refer the class: mindspore.ops.ExpandDims. For a 2-D array, this is a standard matrix transpose. For an n-D array, if axes are given, their order indicates how the axes are permuted (see Examples). If axes are not provided and a.shape is \((i[0], i[1], ... i[n-2], i[n-1])\), then a.transpose().shape is \((i[n-1], i[n-2], ... i[1], i[0])\).

Note

On GPU and CPU, if the value of input_perm is negative, its actual value is input_perm[i] + rank(input). Negative value of input_perm is not supported on Ascend.

Parameters

  • input (Tensor) – The shape of tensor is \((x_1, x_2, ..., x_R)\).

  • input_perm (tuple[int]) – The permutation to be converted. The elements in input_perm are composed of the indexes of each dimension of input. The length of input_perm and the shape of input must be the same. Only constant value is allowed. Must be in the range [-rank(input), rank(input)).

Returns

Tensor, the type of output tensor is the same as input and the shape of output tensor is decided by the shape of input and the value of input_perm.

Raises

  • TypeError – If input_perm is not a tuple.

  • ValueError – If length of shape of input is not equal to length of shape of input_perm.

  • ValueError – If the same element exists in input_perm.

Supported Platforms:

Ascend GPU CPU

Examples

>>> input = Tensor(np.array([[[1, 2, 3], [4, 5, 6]], [[7, 8, 9], [10, 11, 12]]]), mindspore.float32)
>>> input_perm = (0, 2, 1)
>>> output = ops.transpose(input, input_perm)
>>> print(output)
[[[ 1.  4.]
  [ 2.  5.]
  [ 3.  6.]]
 [[ 7. 10.]
  [ 8. 11.]
  [ 9. 12.]]]