mindspore.ops.uniform

查看源文件
mindspore.ops.uniform(shape, minval, maxval, seed=None, dtype=mstype.float32)[源代码]

生成服从均匀分布的随机数。

说明

广播后,任意位置上tensor的最小值必须小于最大值。

参数:
  • shape (Union[tuple, Tensor]) - 返回tensor的shape。

  • minval (Tensor) - 指定生成随机值的最小值。

  • maxval (Tensor) - 指定生成随机值的最大值。

  • seed (int) - 随机数种子。默认 None

  • dtype (mindspore.dtype) - 返回tensor的数据类型。

返回:

Tensor

支持平台:

GPU CPU

样例:

>>> import mindspore
>>> # For discrete uniform distribution, only one number is allowed for both minval and maxval:
>>> shape = (4, 2)
>>> minval = mindspore.tensor(1, mindspore.int32)
>>> maxval = mindspore.tensor(2, mindspore.int32)
>>> output = mindspore.ops.uniform(shape, minval, maxval, seed=5, dtype=mindspore.int32)
>>>
>>> # For continuous uniform distribution, minval and maxval can be multi-dimentional:
>>> shape = (3, 1, 2)
>>> minval = mindspore.tensor([[3, 4], [5, 6]], mindspore.float32)
>>> maxval = mindspore.tensor([8.0, 10.0], mindspore.float32)
>>> output = mindspore.ops.uniform(shape, minval, maxval, seed=5)
>>> result = output.shape
>>> print(result)
(3, 2, 2)