Unexpected result when operating tensor in mps

mps
1

The result of following code seems not right:

device = 'mps'
list_ = torch.zeros(3, 2)
print('init:\n{}'.format(list_))

for i in range(3):
    print('---------------%i---------------' %i)
    ten = torch.tensor([i,i], device=device)
    print('tensor:\n{}'.format(ten))
    list_[i] = ten
    print("ten:\n{}\nlist:\n{}".format(ten, list_), end='\n\n')

result:

init:
tensor([[0., 0.],
        [0., 0.],
        [0., 0.]])
---------------0---------------
tensor:
tensor([0, 0], device='mps:0')
ten:
tensor([0, 0], device='mps:0')
list:
tensor([[0., 0.],
        [0., 0.],
        [0., 0.]])

---------------1---------------
tensor:
tensor([1, 1], device='mps:0')
ten:
tensor([4575657222473777152,                   1], device='mps:0')
list:
tensor([[1.0000e+00, 1.0000e+00],
        [1.4013e-45, 0.0000e+00],
        [0.0000e+00, 0.0000e+00]])

---------------2---------------
tensor:
tensor([2, 2], device='mps:0')
ten:
tensor([4611686019501129728,                   2], device='mps:0')
list:
tensor([[2.0000e+00, 2.0000e+00],
        [2.8026e-45, 0.0000e+00],
        [0.0000e+00, 0.0000e+00]])

move list_ to mps:

device = 'mps'
list_ = torch.zeros(3, 2).to(device)
print('init:\n{}'.format(list_))

for i in range(3):
    print('---------------%i---------------' %i)
    ten = torch.tensor([i,i], device=device)
    print('tensor:\n{}'.format(ten))
    list_[i] = ten
    print("ten:\n{}\nlist:\n{}".format(ten, list_), end='\n\n')

result:

init:
tensor([[0., 0.],
        [0., 0.],
        [0., 0.]], device='mps:0')
---------------0---------------
tensor:
tensor([0, 0], device='mps:0')
ten:
tensor([0, 0], device='mps:0')
list:
tensor([[0., 0.],
        [0., 0.],
        [0., 0.]], device='mps:0')

---------------1---------------
tensor:
tensor([1, 1], device='mps:0')
ten:
tensor([1, 1], device='mps:0')
list:
tensor([[1., 1.],
        [0., 0.],
        [0., 0.]], device='mps:0')

---------------2---------------
tensor:
tensor([2, 2], device='mps:0')
ten:
tensor([2, 2], device='mps:0')
list:
tensor([[2., 2.],
        [0., 0.],
        [0., 0.]], device='mps:0')