randn

paddle. randn ( shape, dtype=None, name=None ) [源代码]

该OP返回符合标准正态分布(均值为0,标准差为1的正态随机分布)的随机Tensor,形状为 shape,数据类型为 dtype

参数

  • shape (list|tuple|Tensor) - 生成的随机Tensor的形状。如果 shape 是list、tuple,则其中的元素可以是int,或者是形状为[1]且数据类型为int32、int64的Tensor。如果 shape 是Tensor,则是数据类型为int32、int64的1-D Tensor。

  • dtype (str|np.dtype|core.VarDesc.VarType, 可选) - 输出Tensor的数据类型,支持float32、float64。当该参数值为None时, 输出Tensor的数据类型为float32。默认值为None.

  • name (str, 可选) - 输出的名字。一般无需设置,默认值为None。该参数供开发人员打印调试信息时使用,具体用法请参见 Name

返回

Tensor:符合标准正态分布的随机Tensor,形状为 shape,数据类型为 dtype

示例代码

  1. import paddle
  2. # example 1: attr shape is a list which doesn't contain Tensor.
  3. out1 = paddle.randn(shape=[2, 3])
  4. # [[-2.923464 , 0.11934398, -0.51249987], # random
  5. # [ 0.39632758, 0.08177969, 0.2692008 ]] # random
  6. # example 2: attr shape is a list which contains Tensor.
  7. dim1 = paddle.to_tensor([2], 'int64')
  8. dim2 = paddle.to_tensor([3], 'int32')
  9. out2 = paddle.randn(shape=[dim1, dim2, 2])
  10. # [[[-2.8852394 , -0.25898588], # random
  11. # [-0.47420555, 0.17683524], # random
  12. # [-0.7989969 , 0.00754541]], # random
  13. # [[ 0.85201347, 0.32320443], # random
  14. # [ 1.1399018 , 0.48336947], # random
  15. # [ 0.8086993 , 0.6868893 ]]] # random
  16. # example 3: attr shape is a Tensor, the data type must be int64 or int32.
  17. shape_tensor = paddle.to_tensor([2, 3])
  18. out3 = paddle.randn(shape_tensor)
  19. # [[-2.878077 , 0.17099959, 0.05111201] # random
  20. # [-0.3761474, -1.044801 , 1.1870178 ]] # random

使用本API的教程文档