mindspore.ops.scatter_max

View Source On Gitee
mindspore.ops.scatter_max(input_x, indices, updates)[source]

Perform a maximum update on input_x based on the specified indices and update values.

\[\text{input_x}[\text{indices}[i, ..., j], :] = \max(\text{input_x}[\text{indices}[i, ..., j], :], \text{updates}[i, ..., j, :])\]

Note

  • Support implicit type conversion and type promotion.

  • Since Parameter objects do not support type conversion, an exception will be thrown when input_x is of a low-precision data type.

  • The shape of updates is indices.shape + input_x.shape[1:].

Parameters

  • input_x (Union[Parameter, Tensor]) – The input parameter or tensor.

  • indices (Tensor) – The specified indices.

  • updates (Tensor) – The update values.

Returns

Tensor

Supported Platforms:

Ascend GPU CPU

Examples

>>> import mindspore
>>> input_x = mindspore.Parameter(mindspore.tensor([[1.0, 2.0, 3.0], [4.0, 5.0, 6.0]],
...                               mindspore.float32), name="input_x")
>>> indices = mindspore.tensor([[0, 0], [1, 1]], mindspore.int32)
>>> updates = mindspore.tensor(mindspore.ops.ones([2, 2, 3]) * 88, mindspore.float32)
>>> output = mindspore.ops.scatter_max(input_x, indices, updates)
>>> print(output)
[[88. 88. 88.]
 [88. 88. 88.]]