UserWarning: torch.meshgrid

[the-cat-crying]

UserWarning: torch.meshgrid: in an upcoming release, it will be required to pass the indexing argument. (Triggered internally at ..\aten\src\ATen\native\TensorShape.cpp:2157.)
return _VF.meshgrid(tensors, **kwargs) # type: ignore[attr-defined]

../accelerated_features/third_party/alike_wrapper.py line 76
../accelerated_features/third_party/ALIKE/soft_detect.py line 88

[guipotje]

Hi @the-cat-crying, this is just a warning because the indexing argument will be required in a future release of PyTorch. You can fix this by passing the argument indexing="ij".

[the-cat-crying]

Thanks！

[the-cat-crying]

But I also found two caveats

1 ../accelerated_features/third_party/ALIKE/soft_detect.py:152: UserWarning: floordiv is deprecated, and its behavior will change in a future version of pytorch. It currently rounds toward 0 (like the 'trunc' function NOT 'floor'). This results in incorrect rounding for negative values. To keep the current behavior, use torch.div(a, b, rounding_mode='trunc'), or for actual floor division, use torch.div(a, b, rounding_mode='floor').
keypoints_xy_nms = torch.stack([indices_kpt % w, indices_kpt // w], dim=1) # Mx2
2 ../accelerated_features/modules/training/losses.py:105: UserWarning: Implicit dimension choice for log_softmax has been deprecated. Change the call to include dim=X as an argument.
kpts = F.log_softmax(kpts)

I found that the code in these two places seems to be duplicated, please guide me if this is what I think.

[the-cat-crying]

[XinningC]

Hello, I've also encounted "Change the call to include dim=X as an argument." on line "kpts = F.log_softmax(kpts)", and I found out that when the argument dim is not specified, the function takes dim=1 by default, which does not makes trouble. Hope this helps you, and please correctify me if I'm wrong.

[XinningC]

@the-cat-crying

[the-cat-crying]

Thanks !@XinningC You are right！