Differences with torch.Tensor.expand
torch.Tensor.expand
torch.Tensor.expand(*sizes) -> Tensor
For more information, see torch.Tensor.expand.
mindspore.Tensor.broadcast_to
mindspore.Tensor.broadcast_to(shape) -> Tensor
For more information, see mindspore.Tensor.broadcast_to.
Differences
MindSpore API function is consistent with PyTorch, with differences in the data types supported by the parameters.
PyTorch: sizes
is the target shape after broadcasting, which can be of type torch.Size
or a sequence consisting of int
.
MindSpore: shape
is the target shape after broadcasting, which can be of type tuple[int]
.
Categories |
Subcategories |
PyTorch |
MindSpore |
Differences |
---|---|---|---|---|
Parameter |
Parameter 1 |
*sizes |
shape |
Both parameters have different names, but both indicate the target shape after broadcasting. The type of |
Code Example
# PyTorch
import torch
x = torch.tensor([1, 2, 3])
output = x.expand(3, 3)
print(output)
# tensor([[1, 2, 3],
# [1, 2, 3],
# [1, 2, 3]])
# MindSpore
import mindspore
import numpy as np
from mindspore import Tensor
shape = (3, 3)
x = Tensor(np.array([1, 2, 3]))
output = x.broadcast_to(shape)
print(output)
# [[1 2 3]
# [1 2 3]
# [1 2 3]]