调整PyTorch张量

LL_*_*LL_ 5 python deep-learning deprecation-warning pytorch tensor

我目前正在使用tensor.resize()函数将张量调整为新的形状t = t.resize(1, 2, 3)

这给了我一个过时的警告:

不推荐使用非原位调整大小

因此,我想切换到该tensor.resize_()功能,这似乎是适当的就地替换。但是,这给我留下了

无法调整需要毕业的变量的大小

错误。我可以回到

from torch.autograd._functions import Resize
Resize.apply(t, (1, 2, 3))
Run Code Online (Sandbox Code Playgroud)

tensor.resize()这样做是为了避免过时警告。这似乎不是一个合适的解决方案,但对我来说却是一个hack。tensor.resize_()在这种情况下,我该如何正确利用?

kma*_*o23 9

您可以改为选择使用tensor.reshape(new_shape)torch.reshape(tensor, new_shape)作为:

# a `Variable` tensor
In [15]: ten = torch.randn(6, requires_grad=True)

# this would throw RuntimeError error
In [16]: ten.resize_(2, 3)
---------------------------------------------------------------------------
RuntimeError                              Traceback (most recent call last)
<ipython-input-16-094491c46baa> in <module>()
----> 1 ten.resize_(2, 3)

RuntimeError: cannot resize variables that require grad
Run Code Online (Sandbox Code Playgroud)

以上RuntimeError可以通过使用解决或避免tensor.reshape(new_shape)

In [17]: ten.reshape(2, 3)
Out[17]: 
tensor([[-0.2185, -0.6335, -0.0041],
        [-1.0147, -1.6359,  0.6965]])

# yet another way of changing tensor shape
In [18]: torch.reshape(ten, (2, 3))
Out[18]: 
tensor([[-0.2185, -0.6335, -0.0041],
        [-1.0147, -1.6359,  0.6965]])
Run Code Online (Sandbox Code Playgroud)