用零替换所有非零值,用特定值替换所有零值

Replace all nonzero values by zero and all zero values by a specific value

我有一个 3d 张量,其中包含一些零值和非零值。我想将所有非零值替换为零,将零值替换为特定值。我该怎么做?

使用 numpy 的方式大致相同,如下所示:

tensor[tensor!=0] = 0

为了替换零和非零,您可以将它们链接在一起。一定要使用张量的副本,因为它们会被修改:

def custom_replace(tensor, on_zero, on_non_zero):
    # we create a copy of the original tensor, 
    # because of the way we are replacing them.
    res = tensor.clone()
    res[tensor==0] = on_zero
    res[tensor!=0] = on_non_zero
    return res

然后像这样使用它:

>>>z 
(0 ,.,.) = 
  0  1
  1  3

(1 ,.,.) = 
  0  1
  1  0
[torch.LongTensor of size 2x2x2]

>>>out = custom_replace(z, on_zero=5, on_non_zero=0)
>>>out
(0 ,.,.) = 
  5  0
  0  0

(1 ,.,.) = 
  5  0
  0  5
[torch.LongTensor of size 2x2x2]

使用

torch.where(<your_tensor> != 0, <tensor with zeroz>, <tensor with the value>)

示例:

>>> x = torch.randn(3, 2)
>>> y = torch.ones(3, 2)
>>> x
tensor([[-0.4620,  0.3139],
         [ 0.3898, -0.7197],
         [ 0.0478, -0.1657]])
>>> torch.where(x > 0, x, y)
Tensor([[ 1.0000,  0.3139],
        [ 0.3898,  1.0000],
        [ 0.0478,  1.0000]])

查看更多信息:https://pytorch.org/docs/stable/generated/torch.where.html

这可以在不克隆张量和使用零值和非零值索引的情况下完成:

zero_indices = tensor == 0
non_zero_indices = tensor != 0
tensor[non_zero_indices] = 0
tensor[zero_indices] = value