Alexey Zaytsev
Alexey Zaytsev
I would love to have access to the original images and try a few things too.
Hi. Thank you for the report. Do you have a concrete use case for complex tensors? I will make sure lt can handle them, but do you have ideas on...
I've pushed a quick fix that just displays the plain version for complex tensors. ```python # We don't really supposed complex numbers yet c = torch.randn(10, dtype=torch.complex64) c ``` tensor([...
> A simple solution would be to treat real and imaginary parts of the tensor as two different tensors, perhaps using [`torch.view_as_real`](https://pytorch.org/docs/stable/generated/torch.view_as_real.html) Could you suggest a codebase, ideally a notebook,...
I'm not sure I follow. Why would we want to return a scalar and not a 0D tensor?
I was using it in my code that was ported to tinygrad, and it works in torch, so don't see a reason it should not work in tinygrad.
Idk man, every time I add a small thing that I use in torch but does not work in tinygrad you give me the same reaction. I get it, it's...
Rebased. I used tuple in my code because Python still does not have `const`, so when I define a list that should be constant, I tend to make it a...
@deven367, The only difference between 0 and 1 workers is, with 0 the notebooks under test will be executed in the context of the nbdev_test process. With `n_workers=1`, one `ProcessPoolExecutor`...
I will take a look at number 2 once we've sorted out the mess with excessive diffs.