Function Differences with torch.randint_like

torch.randint_like

torch.randint_like(input, low=0, high, *, dtype=None, layout=torch.strided, device=None, requires_grad=False, memory_format=torch.preserve_format)

For more information, see torch.randint_like.

mindspore.ops.randint_like

mindspore.ops.randint_like(input, low, high, *, dtype=None, seed=None)

For more information, see mindspore.ops.randint_like.

Differences

PyTorch：low is optional with default value: 0.

MindSpore：low is not optional and has no default value.

Categories	Subcategories	PyTorch	MindSpore	Differences
Parameter	Parameter 1	input	input	-
	Parameter 2	low	low	low has default value 0 in PyTorch, MindSpore does not.
	Parameter 3	high	high	-
	Parameter 4	dtype	dtype	-
	Parameter 5	layout	-	General difference
	Parameter 6	device	-	General difference
	Parameter 7	requires_grad	-	General difference
	Parameter 8	memory_format	-	General difference
	Parameter 9	-	seed	General difference

Code Example 1

# PyTorch
import torch

# PyTorch does not need to set the value of low.
x = torch.tensor([[2, 3], [1, 2]], dtype=torch.float32)
y = torch.randint_like(x, 10)
print(tuple(y.shape))
# (2, 2)

# MindSpore
import mindspore

# MindSpore must set the default value of low in PyTorch(0 in this case), as one of the inputs.
x = mindspore.Tensor([[2, 3], [1, 2]], mindspore.float32)
y = mindspore.ops.randint_like(x, 0, 10)
print(y.shape)
# (2, 2)