The page /tutorials/beginner/nlp/pytorch_tutorial.html contains an error:
the code at the bottom, that including
[...] print(var_z.grad_fn) [...] # ... does new_var_z have information to backprop to x and y? # NO! print(new_var_z.grad_fn)
is stated to have output “None \n None”
It should be instead
<torch.autograd.function.AddBackward object at 0x7f758cd01234> None
EDIT: also publshed an issue on github, I am not sure which is the right place.