C++ tensor print doesn't show requires_grad and grad_fn like Python tensor print

[yf225]

In Python, we have the following tensor print behavior:

>>> x=torch.ones(2, 2, requires_grad=True)
>>> x
tensor([[1., 1.],
        [1., 1.]], requires_grad=True)
>>> y=x*2
>>> y
tensor([[2., 2.],
        [2., 2.]], grad_fn=<MulBackward0>)

However in C++, we have the following behavior:

auto x = torch::ones({2, 2}, torch::requires_grad());
std::cout << x << std::endl;
auto y = x * 2;
std::cout << y << std::endl;

 1  1
 1  1
[ CPUFloatType{2,2} ]
 2  2
 2  2
[ CPUFloatType{2,2} ]

It would be nice to show requires_grad and grad_fn in C++ tensor print as well.

cc @yf225

[choidongyeon]

I would like to take a look at this. Could you point me in the right direction of which files I should look into?

[xwang233]

@choidongyeon It seems like you can start from here.

pytorch/aten/src/ATen/core/Formatting.cpp

Lines 230 to 233 in f2bae8e

	std::ostream& print(std::ostream& stream, const Tensor & tensor_, int64_t linesize) {
	FormatGuard guard(stream);
	if(!tensor_.defined()) {
	stream << "[ Tensor (undefined) ]";