Text: Copy Construct warning

[vince62s]

Hello,
was this ever fixed ?

/usr/local/lib/python3.5/dist-packages/torchtext/data/field.py:308: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).

[vince62s]

it refers to this line in master:
https://github.com/pytorch/text/blob/master/torchtext/data/field.py#L359

[toli-belo]

also what if you need to convert tensor to dtype=torch.float ?
replacing with clone().detach() is not going to convert the tensor....

[mttk]

Could you provide the case where you get this error?

I assume that arr is a torch.tensor, which means that the numericalization / postprocessing casts it into that at some point (as the standard pipeline shouldn't do that prior to L359).

[zhangguanheng66]

@toli-belo feel free to reopen the issue if you still have questions.

[martijnvanattekum]

for future reference: https://github.com/pytorch/pytorch/issues/16627. In summary, try to replace torch.tensor with torch.as_tensor