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 |
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,)