I’m new to pytorch, and I just got to this forum from google while trying to fix some issues I had following the 60 minute blitz intro to pytorch. I was having trouble with the blitz’s recommendation to set requires_grad on an existing tensor using tensor.requires_grad_(True), and decided based on another post here that the method probably got deprecated after the Tensor and Variable classes got merged, but I’d love to have some confirmation on that.
Also, the way devices are handled seems to have changed.
Is the 60 minute blitz somewhere in the git for pytorch, and could I submit a pull request, and/or review anything that seems wrong in the docs as I’m learning more about pytorch with someone who knows how it’s supposed to work better than I do?