Pytorch dataloader removes negative values

Hello everyone!
I am working with pytorch dataloaders and have my own custom dataset initialized. I am normalizing data inside the custom dataset class between -1 and 1. but when i check the output of dataloader, min and max value of the output is 1 and 1. When i print values from inside dataset class, the values are normalized from -1 to 1. I am not using ANY pytorch transformations. I am doing all these operations using opencv and then is returning the image as torch.FloatTensor(input tensor normalized in -1 and 1 range). Below is the initialization provided.

dataloaders = {
            'train': DataLoader(NYUImagesDataset(), batch_size=batch_size, shuffle=True, num_workers=0),
            'test': DataLoader(NYUImagesDataset(mode="test"), batch_size=batch_size, shuffle=True, num_workers=0)

NYUImageDataset is my custom dataset class and the internal functioning of this class, i have described above. Looking forward to helpful responses.

DataLoader should not change anything as shown in this example:

data = torch.empty(1000, 1).uniform_(-1, 1)
dataset = TensorDataset(data)
loader = DataLoader(dataset, batch_size=5, shuffle=True)

for data in loader:
    print(data[0].min(), data[0].max())

Could you use a similar check and iterate your Dataset as well as the DataLoader?