Accuracy, Precision, Recall raises RuntimeError when device="cuda"

[Voleraa]

🐛 Bug

Hi,

when I run the following code,

import torch
import torchmetrics

preds  = torch.tensor([[0.6,0.9,0.3,0,0,0,0,0.6], 
                       [0.1,0.6,0.1,0,0,0,0,0]])
target = torch.tensor([[0,1,1,0,0,0,0,0],
                       [1,1,0,0,0,0,0,1]])
accuracy = torchmetrics.Accuracy()
accuracy(preds.cuda(), target.cuda())

I get the following RuntimeError:

RuntimeError                              Traceback (most recent call last)
/tmp/ipykernel_54291/197549698.py in <module>
      7                        [1,1,0,0,0,0,0,1]])
      8 accuracy = torchmetrics.Accuracy()
----> 9 accuracy(preds.cuda(), target.cuda())

/opt/conda/lib/python3.8/site-packages/torch/nn/modules/module.py in _call_impl(self, *input, **kwargs)
    887             result = self._slow_forward(*input, **kwargs)
    888         else:
--> 889             result = self.forward(*input, **kwargs)
    890         for hook in itertools.chain(
    891                 _global_forward_hooks.values(),

~/.local/lib/python3.8/site-packages/torchmetrics/metric.py in forward(self, *args, **kwargs)
    204 
    205         # global accumulation
--> 206         self.update(*args, **kwargs)
    207 
    208         if self.compute_on_step:

~/.local/lib/python3.8/site-packages/torchmetrics/metric.py in wrapped_func(*args, **kwargs)
    265             self._update_called = True
    266             with torch.set_grad_enabled(self._enable_grad):
--> 267                 return update(*args, **kwargs)
    268 
    269         return wrapped_func

~/.local/lib/python3.8/site-packages/torchmetrics/classification/accuracy.py in update(self, preds, target)
    258             # Update states
    259             if self.reduce != "samples" and self.mdmc_reduce != "samplewise":
--> 260                 self.tp += tp
    261                 self.fp += fp
    262                 self.tn += tn

RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu!

This error also occurs when I use Recall and Precision.

Environment

• OS: Linux
• Python Version: 3.8
• PyTorch Version: 1.8.1

[github-actions]

Hi! thanks for your contribution!, great first issue!

[Borda]

Hi, in such case you also need to move the metric to the device

accuracy = torchmetrics.Accuracy().cuda()
accuracy(preds.cuda(), target.cuda())

if you want to have a seamless experience, please use the PytorchLightning logging feature
https://pytorch-lightning.readthedocs.io/en/stable/extensions/logging.html#automatic-logging