Niccolò Zanotti

That's lovely @minomy13! I recently switched to starship and was looking exactly for this. Hopeful to see #6685 merged soon :)

@jatkinson1000 So, I've looked and from the time of writing of this things have changed a bit. We now support creating a tensor directly on the GPU through `torch_tensor_zeros` or...

I think the problems lies in that `torch::from_blob(data, vshape, vstrides, options)` expects data to *already* reside on the `device` passed as part of `options`. (See [here](https://docs.pytorch.org/cppdocs/api/function_namespacetorch_1ac009244049812a3efdf4605d19c5e79b.html)). Thus we cannot avoid...

Ah yes, this was not considered in #410. > Really we should look to update the script to run everything possible on the GPU. Definitely. I could take this on...