Differences with torch.bernoulli

The following mapping relationships can be found in this file.

PyTorch APIs	MindSpore APIs
torch.bernoulli	mindspore.ops.bernoulli
torch.Tensor.bernoulli	mindspore.Tensor.bernoulli

torch.bernoulli

torch.bernoulli(input, *, generator=None, out=None)

For more information, see torch.bernoulli.

mindspore.ops.bernoulli

mindspore.ops.bernoulli(input, p=0.5, seed=None)

For more information, see mindspore.ops.bernoulli.

Differences

API function of MindSpore is consistent with that of PyTorch.

PyTorch: The probability value of the Bernoulli distribution is stored in the parameter input , and the shape of the returned value is the same as that of input .

MindSpore: The probability value of the Bernoulli distribution is stored in the parameter p , with a default value of 0.5. The shape of p needs to be consistent with the shape of input , and the shape of the return value should be the same as the shape of input .

Categories	Subcategories	PyTorch	MindSpore	Differences
Parameters	Parameter 1	-	input	The shape and data type of the returned value under Mindspore are the same as the shape of input
	Parameter 2	input	p	Save the probability values for the Bernoulli distribution. The shape of the return value under PyTorch is the same as ‘input’. ‘p’ is optional under MindSpore, and the default value is 0.5
	Parameter 3	generator	seed	MindSpore uses a random number seed to generate random numbers
	Parameter 4	out	-	Not involved

Code Example

# PyTorch
import torch
import numpy as np

p0 = np.array([0.0, 1.0, 1.0])
input_torch = torch.tensor(p0, dtype=torch.float32)
output = torch.bernoulli(input_torch)
print(output.shape)
# torch.Size([3])

# MindSpore
import mindspore as ms
import numpy as np

input0 = np.array([1, 2, 3])
p0 = np.array([0.0, 1.0, 1.0])

input = ms.Tensor(input0, ms.float32)
p = ms.Tensor(p0, ms.float32)
output = ms.ops.bernoulli(input, p)
print(output.shape)
# (3,)