Setting float and bool values to quality argument of JPEG() gets the indirect errors

[hyperkai]

🐛 Describe the bug

Setting float and bool values to quality argument of JPEG() gets the indirect errors as shown below:

from torchvision.transforms.v2 import JPEG

JPEG(quality=5.3) # Error

TypeError: quality should be a sequence of length 2.

from torchvision.datasets import OxfordIIITPet
from torchvision.transforms.v2 import JPEG

my_data = OxfordIIITPet(
    root="data",
    transform=JPEG(quality=True)
)

my_data[0] # Error

TypeError: randint() received an invalid combination of arguments - got (bool, int, tuple), but expected one of:
 * (int high, tuple of ints size, *, torch.Generator generator, Tensor out = None, torch.dtype dtype = None, torch.layout layout = None, torch.device device = None, bool pin_memory = False, bool requires_grad = False)
 * (int high, tuple of ints size, *, Tensor out = None, torch.dtype dtype = None, torch.layout layout = None, torch.device device = None, bool pin_memory = False, bool requires_grad = False)
 * (int low, int high, tuple of ints size, *, torch.Generator generator, Tensor out = None, torch.dtype dtype = None, torch.layout layout = None, torch.device device = None, bool pin_memory = False, bool requires_grad = False)
 * (int low, int high, tuple of ints size, *, Tensor out = None, torch.dtype dtype = None, torch.layout layout = None, torch.device device = None, bool pin_memory = False, bool requires_grad = False)

So they should be something direct like below:

TypeError: quality argument must be int

Versions

import torchvision

torchvision.__version__ # '0.20.1'

[NicolasHug]

Thanks for the report @hyperkai , feel free to submit a PR for a better error message