【笔记】torch.Tensor、t.tensor、torch.Tensor([A]).expand_as(B):torch.float32；torch.int64；将A变成B的形状，不与AB共享内存

eg1：

Intput:

import torch
a=torch.Tensor([1])
b=torch.tensor([2])
print(a.dtype,b.dtype)
boxes=torch.Tensor([3,4,5,6])
print(boxes)
boxes /= torch.Tensor([2, 2, 2, 2]).expand_as(boxes)
print(boxes)

Output:

torch.float32 torch.int64
tensor([3., 4., 5., 6.])
tensor([1.5000, 2.0000, 2.5000, 3.0000])

eg2：

Input：

import torch
a=torch.Tensor([1])
b=torch.tensor([2])
print(a.dtype,b.dtype)
boxes=torch.Tensor([[3,4,5,6],[1,2,3,4]])
print(boxes)
boxes = torch.Tensor([2, 2, 2, 2]).expand_as(boxes)
print(boxes)

Output：

torch.float32 torch.int64
tensor([[3., 4., 5., 6.],
        [1., 2., 3., 4.]])
tensor([[2., 2., 2., 2.],
        [2., 2., 2., 2.]])