In torch, printing a tensor will print the whole tensor, even if that one is huge.
This is sometimes annoying, especially when using a notebook, as it crashes it.

Numpy has a very nice way to display big tensors by truncating them. Would it be possible so have something similar for pytorch?