打印张量的所有内容

the*_*der 23 python debugging machine-learning pytorch

我遇到了这个PyTorch 教程(在 neural_networks_tutorial.py 中),他们在其中构建了一个简单的神经网络并运行了推理。我想打印整个输入张量的内容以进行调试。当我尝试打印张量时,我得到的是这样的,而不是整个张量:

在此处输入图片说明

我看到了一个类似的 numpy链接,但不确定什么对 PyTorch 有用。我可以将它转换为 numpy 并可以查看它,但希望避免额外的开销。有没有办法打印整个张量?

sta*_*son 19

为了避免截断并控制打印多少张量数据,请使用与 numpy 的numpy.set_printoptions(threshold=10_000).

例子:

x = torch.rand(1000, 2, 2)
print(x) # prints the truncated tensor
torch.set_printoptions(threshold=10_000)
print(x) # prints the whole tensor
Run Code Online (Sandbox Code Playgroud)

如果您的张量非常大,请将threshold值调整为更大的数字。

另一种选择是:

torch.set_printoptions(profile="full")
print(x) # prints the whole tensor
torch.set_printoptions(profile="default") # reset
print(x) # prints the truncated tensor
Run Code Online (Sandbox Code Playgroud)

此处set_printoptions记录所有可用参数。


Lyk*_*Syu 14

虽然我不建议这样做,如果你愿意,那么

In [18]: torch.set_printoptions(edgeitems=1)

In [19]: a
Out[19]:
tensor([[-0.7698,  ..., -0.1949],
        ...,
        [-0.7321,  ...,  0.8537]])

In [20]: torch.set_printoptions(edgeitems=3)

In [21]: a
Out[21]:
tensor([[-0.7698,  1.3383,  0.5649,  ...,  1.3567,  0.6896, -0.1949],
        [-0.5761, -0.9789, -0.2058,  ..., -0.5843,  2.6311, -0.0008],
        [ 1.3152,  1.8851, -0.9761,  ...,  0.8639, -0.6237,  0.5646],
        ...,
        [ 0.2851,  0.5504, -0.9471,  ...,  0.0688, -0.7777,  0.1661],
        [ 2.9616, -0.8685, -1.5467,  ..., -1.4646,  1.1098, -1.0873],
        [-0.7321,  0.7610,  0.3182,  ...,  2.5859, -0.9709,  0.8537]])
Run Code Online (Sandbox Code Playgroud)


tro*_*lus 9

我来这里实际上是在寻找如何在控制台的一行中打印整行张量的答案,所以我想我应该添加这个。

tensor([[1.1573e+04, 6.0693e+02, 1.2436e+03, 2.7277e+04, 1.6673e+08, 2.0462e+00, 9.8891e-01],
    [2.0237e+04, 5.9074e+02, 1.7208e+03, 2.7449e+04, 2.1301e+08, 2.0678e+00, 1.0011e+00],
    [2.7456e+04, 6.1106e+02, 1.4897e+03, 2.7332e+04, 1.7310e+08, 2.0448e+00, 9.6041e-01],
    [1.7732e+04, 6.0232e+02, 1.2608e+03, 2.7371e+04, 1.8106e+08, 1.9594e+00, 1.0040e+00],
    ...,
    [1.1167e+04, 5.9867e+02, 1.3440e+03, 2.7263e+04, 2.3160e+08, 2.0190e+00, 1.0075e+00],
    [1.6003e+04, 5.9590e+02, 1.2319e+03, 2.7368e+04, 1.7155e+08, 2.0171e+00, 1.0202e+00],
    [1.5499e+04, 6.1471e+02, 9.4877e+02, 2.7395e+04, 1.8146e+08, 1.9016e+00, 9.5884e-01],
    [3.3886e+04, 6.0689e+02, 1.0777e+03, 2.7259e+04, 2.1599e+08, 2.0179e+00, 1.0201e+00]], dtype=torch.float64)
Run Code Online (Sandbox Code Playgroud)

我这样做是使用

torch.set_printoptions(linewidth=200)
Run Code Online (Sandbox Code Playgroud)