mindspore.Tensor.top_k

Tensor.top_k(k, sorted=True)[source]

Finds values and indices of the k largest entries along the last dimension.

Warning

If sorted is set to ‘False’, it will use the aicpu operator, the performance may be reduced.

input_x refers to self tensor.

If the input_x is a one-dimensional Tensor, finds the k largest entries in the Tensor, and outputs its value and index as a Tensor. Therefore, values[k] is the k largest item in input_x, and its index is indices [k].

For a multi-dimensional matrix, calculates the first k entries in each row (corresponding vector along the last dimension), therefore:

\[values.shape = indices.shape = input\_x.shape[:-1] + [k].\]

If the two compared elements are the same, the one with the smaller index value is returned first.

Parameters

k (int) – The number of top elements to be computed along the last dimension, constant input is needed.
sorted (bool, optional) – If true, the obtained elements will be sorted by the values in descending order. Default: True.

Returns

Tuple of 2 tensors, the values and the indices.

values (Tensor): The k largest elements in each slice of the last dimension.
indices (Tensor): The indices of values within the last dimension of input.

Raises

TypeError – If k is not an int.
TypeError – If sorted is not a bool.

Supported Platforms:: Ascend GPU CPU

Examples

>>> import mindspore as ms
>>> from mindspore import Tensor
>>> input_x = Tensor([1, 2, 3, 4, 5], ms.float16)
>>> k = 3
>>> values, indices = input_x.top_k(k, sorted=True)
>>> print((values, indices))
(Tensor(shape=[3], dtype=Float16, value= [ 5.0000e+00,  4.0000e+00,  3.0000e+00]), Tensor(shape=[3],
  dtype=Int32, value= [4, 3, 2]))