Why I use shuffle = False, the order of data are changed?

I used the dataloader for the test_data. Below is the dataset:

class Test_Data1(Dataset):
    
    # Constructor
    def __init__(self):
        
        test_data = numpy.load('x_test.npy')
        test_labels = numpy.load('y_test.npy')
        
        test_data = test_data
        self.x = test_data
        self.y = test_labels
        self.len = self.x.shape[0]
        
    # Getter
    def __getitem__(self, index):    
        return self.x[index], self.y[index]
    
    # Get length
    def __len__(self):
        return self.len

This is my test code:

test_dataset1 = Test_Data1()
test_loader1 = DataLoader(dataset=test_dataset1, batch_size=360, shuffle=False)
correct = 0
total = 0
for data in test_loader1:
        inputs, labels = data
        print(labels)
        labels = labels.type(torch.LongTensor)
        inputs = inputs.view(-1, 1024).double()
        outputs =model1(inputs.float())
        _, predicted = torch.max(outputs.data, 1)
        print(predicted)
        total += labels.size(0)
        correct += (predicted == labels).sum()
print(100 * correct /float(total))

By comparison of the output, I found that the order was changed.
My order
image
The output order from dataloader:
image

Looking forward to your reply.
Best regards,
William

I am unable to reproduce the behavior in torch 1.6.0.