No completion suggestions in IDE for `torch.zeros`

[rchen152]

Some preliminary investigation by @grievejia summarized here: #783 (comment).

Discussed in #783

^{Originally posted by michaelfortunato July 29, 2025}
Hi all,
I am a big fan of pyrefly. I have a question about pytorch type hints and I am not sure if they are particular to pyrefly or not, but any help would be appreciated. torch.zeros is a valid function in pytorch top level module. However, I get no suggestion completion for torch.zeros. See the photo below:

I asked the ai slop bots and they said it might be due to torch.zeros being a function defined in C? Is there a way I can get pyrefly to recognize torch.zeros. Thanks

[lolpack]

Note

    import torch

    reveal_type(torch.zeros)

# Revealed type: Overload[(size: Sequence[SymInt | int], *, out: Tensor | None = None, dtype: dtype | None = None, layout: layout | None = None, device: device | int | str | None = None, pin_memory: bool | None = False, requires_grad: bool | None = False) -> Tensor, (*size: SymInt | int, *, out: Tensor | None = None, dtype: dtype | None = None, layout: layout | None = None, device: device | int | str | None = None, pin_memory: bool | None = False, requires_grad: bool | None = False) -> Tensor, (size: Size | list[int] | tuple[int, ...], *, names: Sequence[EllipsisType | str | None] | None, dtype: dtype | None = None, layout: layout | None = None, device: device | int | str | None = None, pin_memory: bool | None = False, requires_grad: bool | None = False) -> Tensor, (*size: int, *, names: Sequence[EllipsisType | str | None] | None, dtype: dtype | None = None, layout: layout | None = None, device: device | int | str | None = None, pin_memory: bool | None = False, requires_grad: bool | None = False) -> Tensor]

Similar

tensor = torch.arange(16).reshape(8, 2).float()

torch.Tensor.full_tensor
tensor.to_local()

full_tensor and to_local go to def does not work.