mindspore.ops.scatter_nd_add

查看源文件
mindspore.ops.scatter_nd_add(input_x, indices, updates, use_locking=False)[源代码]

根据指定索引和更新值对 input_x 进行稀疏加法更新。

input_x[indices[i,...,j]]+=updates[i,...,j]

说明

  • 支持隐式类型转换、类型提升。

  • indices 的维度至少为2,并且 indices.shape[-1] <= len(indices.shape)

  • updates 的shape为 indices.shape[:-1] + input_x.shape[indices.shape[-1]:]

参数:
  • input_x (Union[Parameter, Tensor]) - 输入的parameter或tensor。

  • indices (Tensor) - 指定索引。

  • updates (Tensor) - 更新值。

  • use_locking (bool,可选) - 是否启用锁保护。默认 False

返回:

Tensor

支持平台:

Ascend GPU CPU

样例:

>>> import mindspore
>>> import numpy as np
>>> input_x = mindspore.Parameter(mindspore.tensor([1, 2, 3, 4, 5, 6, 7, 8],
...                               mindspore.float32), name="x")
>>> indices = mindspore.tensor([[2], [4], [1], [7]], mindspore.int32)
>>> updates = mindspore.tensor([6, 7, 8, 9], mindspore.float32)
>>> output = mindspore.ops.scatter_nd_add(input_x, indices, updates, False)
>>> print(output)
[ 1. 10.  9.  4. 12.  6.  7. 17.]
>>> input_x = mindspore.Parameter(mindspore.tensor(np.zeros((4, 4, 4)), mindspore.int32))
>>> indices = mindspore.tensor([[0], [2]], mindspore.int32)
>>> updates = mindspore.tensor([[[1, 1, 1, 1], [2, 2, 2, 2], [3, 3, 3, 3], [4, 4, 4, 4]],
...                            [[5, 5, 5, 5], [6, 6, 6, 6], [7, 7, 7, 7], [8, 8, 8, 8]]], mindspore.int32)
>>> output = mindspore.ops.scatter_nd_add(input_x, indices, updates, False)
>>> print(output)
[[[1 1 1 1]
  [2 2 2 2]
  [3 3 3 3]
  [4 4 4 4]]
 [[0 0 0 0]
  [0 0 0 0]
  [0 0 0 0]
  [0 0 0 0]]
 [[5 5 5 5]
  [6 6 6 6]
  [7 7 7 7]
  [8 8 8 8]]
 [[0 0 0 0]
  [0 0 0 0]
  [0 0 0 0]
  [0 0 0 0]]]