clone() throwing exception on single Metric or MetricCollection

[omerferhatt]

🐛 Bug

clone() functionality not working and raising argument error.

To Reproduce

>>> import torchmetrics as tm
>>> iou_train = tm.JaccardIndex(num_classes=3, average="none", mdmc_average="samplewise")
>>> iou_val = iou_train.clone()

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/.../.venv/lib/python3.8/site-packages/torchmetrics/metric.py", line 569, in clone
    return deepcopy(self)
  File "/usr/lib/python3.8/copy.py", line 172, in deepcopy
    y = _reconstruct(x, memo, *rv)
  File "/usr/lib/python3.8/copy.py", line 264, in _reconstruct
    y = func(*args)
  File "/usr/lib/python3.8/copyreg.py", line 91, in __newobj__
    return cls.__new__(cls, *args)
TypeError: __new__() missing 1 required positional argument: 'num_classes

Expected behavior

It should create a new object with given template

Environment

• TorchMetrics version (and how you installed TM, e.g. conda, pip, build from source):
  • torchmetrics==0.10.0 (pip)
• Python & PyTorch Version (e.g., 1.0):
  • python==3.8.10
  • torch==1.12.1+cu116
• Any other relevant information such as OS (e.g., Linux):
  • uname

Linux hostname 5.15.0-1018-gcp #24~20.04.1-Ubuntu SMP Mon Sep 12 06:14:01 UTC 2022 x86_64 x86_64 x86_64 GNU/Linux

• nvcc --version

nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2022 NVIDIA Corporation
Built on Tue_Mar__8_18:18:20_PST_2022
Cuda compilation tools, release 11.6, V11.6.124
Build cuda_11.6.r11.6/compiler.31057947_0

Additional context

P.S. When I downgrade to version ~0.9 , errors not occur anymore and works expected.