RandomResizedCrop vs Resize [UPDATED]

vision
1

When I use this line of codes, everything works fine.

    train_transform = T.Compose([
        T.RandomResizedCrop(224),
        T.ToTensor(),
    ])
    train_dset = ImageFolder(train_dir, transform=train_transform)
    train_loader = DataLoader(train_dset,
                    batch_size = batch_size,
                    shuffle=True)
    data_iter = iter(train_loader)

    images, labels = next(data_iter)
    print("Size of x ", images.size())

it outputs:
Size of x torch.Size([32, 3, 224, 224])
But if I will change RandomResizedCrop(224) to Resize(224)
it ouputs:

RuntimeError: invalid argument 0: Sizes of tensors must match except in dimension 0. Got 335 and 334 in dimension 3 at c:\a\w\1\s\tmp_conda_3.6_061433\conda\conda-bld\pytorch_1544163532679\work\aten\src\th\generic/THTensorMoreMath.cpp:1333

Could you help me?
UPDATE:
I just found that my code works ok with Tiny Imagenet Dataset, but fails with COCO Animals dataset (8 classes)

2

I needed to write Resize((224,224))