What's going on on torch.transform.ToTensor?

Question

let

x = np.array([[0,254],[14,77]])
>>> x
array([[  0, 254],
       [ 14,  77]])

y = torch.from_numpy(x).int().float()
>>> y
tensor([[  0., 254.],
        [ 14.,  77.]])

train_transform = transforms.Compose([transforms.ToPILImage(),
                                      transforms.ToTensor()])

z = train_transform(y)
>>> z
tensor([[[0.0000, 0.0078],
         [0.9490, 0.7020]]])

if I use the ToTensor, it will be divided into 255, but what is going on with that? Why is 14 and 77 high?