learned_perceptual_image_patch_similarity function needs to synchronize device and dtype between net and input images

[michael080808]

🐛 Bug

When I try to use preds and target on torch.device('cuda:0') to calculate with function learned_perceptual_image_patch_similarity from torchmetrics.functional.image, the function will throw exception.

Maybe there are some other functions with internal torch.nn.Module instances in torchmetrics.functional will proceed like this.

To Reproduce

import torch
from torchmetrics.functional.image import learned_perceptual_image_patch_similarity

result = torch.rand(1, 3, 50, 50, device='cuda:0', dtype=torch.float64)
target = torch.rand(1, 3, 50, 50, device='cuda:0', dtype=torch.float64)
learned_perceptual_image_patch_similarity(result, target, normalize=True)

Expected behavior

Change

torchmetrics/src/torchmetrics/functional/image/lpips.py

Line 430 in bcf7867

loss, total = _lpips_update(img1, img2, net, normalize)

To

    loss, total = _lpips_update(img1, img2, net.to(device=img1.device, dtype=img1.dtype), normalize) 

Environment

Name	Version	Build	Channel
_anaconda_depends	2023.07	py311_0	https://repo.anaconda.com/pkgs/main
conda	23.9.0	py311haa95532_0	https://repo.anaconda.com/pkgs/main
python	3.10.13	he1021f5_0	defaults
pytorch	2.1.0	py3.10_cuda12.1_cudnn8_0	pytorch
pytorch-cuda	12.1	hde6ce7c_5	pytorch
pytorch-mutex	1.0	cuda	pytorch
torchaudio	2.1.0	pypi_0	pypi
torchmetrics	1.2.0	pyhd8ed1ab_0	conda-forge
torchvision	0.16.0	pypi_0	pypi