用法类似于numpy的resize，主要是用来改变tensor的大小
若参数不带-1
1
2
3
4
5
| temp1 = torch.tensor([-0.3623, -0.6115, 0.7283, 0.4699, 2.3261, 0.1599])
result1 = temp1.view(3,2)
result1
|
1
2
3
4
5
| Output: >>
tensor([[-0.3623, -0.6115],
[ 0.7283, 0.4699],
[ 2.3261, 0.1599]])
|
有的时候会出现torch.view(-1)或者torch.view(参数a，-1)这种情况
1
2
3
4
5
6
7
| temp2=torch.tensor([[-0.3623, -0.6115],
... [ 0.7283, 0.4699],
... [ 2.3261, 0.1599]])
result2 = temp2.view(-1)
result2
|
1
2
3
| Output: >>
tensor([-0.3623, -0.6115, 0.7283, 0.4699, 2.3261, 0.1599])
|
1
2
3
4
5
6
| temp3=torch.tensor([[-0.3623, -0.6115],
... [ 0.7283, 0.4699],
... [ 2.3261, 0.1599]])
result3=temp3.view(2,-1)
result3
|
1
2
3
4
| Output: >>
tensor([[-0.3623, -0.6115, 0.7283],
[ 0.4699, 2.3261, 0.1599]])
|
