mindspore.ops.MaskedFill

class mindspore.ops.MaskedFill[source]

Fills elements with value where mask is True.

Note

If value is a floating-point number of Python, it will be converted to float32 later by default. In this case, if input_x is a float16 Tensor, it will be converted to float32 for calculation, and the result type will be converted back to float16 on the CPU and Ascend platforms, which may cause the performance penalty. A TypeError may be raised on the GPU platform. Therefore, it is recommended that ‘value’ should use a Tensor with the same dtype as input_x.

Refer to mindspore.ops.masked_fill() for more details.

Supported Platforms:: Ascend GPU CPU

Examples

>>> input = Tensor(np.array([1., 2., 3., 4.]), mindspore.float32)
>>> mask = Tensor(np.array([True, True, False, True]), mindspore.bool_)
>>> output = ops.MaskedFill()(input, mask, 0.5)
>>> print(output)
[0.5 0.5 3.  0.5]