Printing all the contents of a tensor

Question

I came across this PyTorch tutorial (in neural_networks_tutorial.py) where they construct a simple neural network and run an inference. I would like to print the contents of the entire input tensor for debugging purposes. What I get when I try to print the tensor is something like this and not the entire tensor:

I saw a similar link for numpy but was not sure about what would work for PyTorch. I can convert it to numpy and may be view it, but would like to avoid the extra overhead. Is there a way for me to print the entire tensor?

Answer 1

To avoid truncation and to control how much of the tensor data is printed use the same API as numpy's numpy.set_printoptions(threshold=10_000).

Example:

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

If your tensor is very large, adjust the threshold value to a higher number.

Another option is:

torch.set_printoptions(profile="full") print(x) # prints the whole tensor torch.set_printoptions(profile="default") # reset print(x) # prints the truncated tensor 

All the available set_printoptions arguments are documented here.