print exact value of tensor(floating point precision) with pytorch

问题

I'm trying to print torch.FloatTensor like

a = torch.FloatTensor(5,5)
print(a)

this way I can got a value like

0.0000e+00  0.0000e+00  3.2286e-41  9.4448e+21  4.3346e-38
1.2412e-40  1.2313e+00  1.6751e-37  3.1138e-40  9.4460e+21
2.6801e-36  3.5873e-41  9.4463e+21  4.9653e-35  3.9963e-40
9.4454e+21  2.6801e-36  1.2771e-40  9.4460e+21  1.7153e-34
7.7056e-40  9.0090e+15  4.1877e-38  2.9775e-41  1.5695e-43

But I want to get more accurate value, like 10 decimal point

0.1234567891+01

in python, I could got it

print('{:.10f}'.format(a))

but in case of tensor, I got this error

TypeError: unsupported format string passed to torch.FloatTensor.__format__

How can I print accurate value of tensor?

回答1:

You can set the precision options

torch.set_printoptions(precision=10)

There are more formatting options on the documentation page: http://pytorch.org/docs/master/torch.html#creation-ops it is very similar to numpys.

回答2:

Just as a side note, this functionality has been taken from numpy. One of the reasons why PyTorch is smart is because they took many the good ideas from numpy.

However, in numpy the default precision is 8 and in PyTorch the default is 4.

来源：https://stackoverflow.com/questions/47483733/print-exact-value-of-tensorfloating-point-precision-with-pytorch