def test_nested_constructor(self):
for constructor in _iter_constructors():
# TODO: Currently only supporting nested dim 1
num_nested_tensor = 1
# TODO: Shouldn't be constructable
[utils.gen_nested_tensor(i, i, 3, constructor=constructor)
for i in range(1, num_nested_tensor)]
def test_pin_memory(self):
# Check if it can be applied widely
nt = utils.gen_nested_tensor(1, 4, 3)
nt1 = nt.pin_memory()
# Make sure it's actually a copy
self.assertFalse(nt.is_pinned())
self.assertTrue(nt1.is_pinned())
a1 = torch.randn(1, 2)
a2 = torch.randn(2, 3)
nt2 = nestedtensor.as_nested_tensor([a1, a2])
self.assertFalse(a1.is_pinned())
self.assertFalse(a2.is_pinned())
# Double check property transfers
nt3 = nt2.pin_memory()
self.assertFalse(nt2.is_pinned())
self.assertTrue(nt3.is_pinned())
# Check whether pinned memory is applied to constiuents
# and relevant constiuents only.
a3, a4 = nt3.unbind()
a5, a6 = nt2.unbind()
self.assertFalse(a1.is_pinned())
self.assertFalse(a2.is_pinned())
self.assertTrue(a3.is_pinned())
self.assertTrue(a4.is_pinned())
self.assertFalse(a5.is_pinned())
self.assertFalse(a6.is_pinned())
def test_nested_constructor(self):
for constructor in _iter_constructors():
num_nested_tensor = 3
# TODO: Shouldn't be constructable
[
utils.gen_nested_tensor(i, i, 3, constructor=constructor)
for i in range(1, num_nested_tensor)
]
def test_tensor_mask(self):
nt = utils.gen_nested_tensor(2, 2, 2, size_low=1, size_high=2)
tensor, mask = nt.to_tensor_mask()
nt1 = nestedtensor.nested_tensor_from_tensor_mask(
tensor, mask, nested_dim=nt.nested_dim())
self.assertEqual(nt, nt1)
nt2 = nestedtensor.nested_tensor_from_tensor_mask(tensor, mask)
self.assertEqual(nt, nt2)
def test_nested_dim(self):
for constructor in _iter_constructors():
nt = constructor([torch.tensor(3)])
for i in range(2, 5):
nt = utils.gen_nested_tensor(i, i, 3, constructor=constructor)
self.assertEqual(nt.nested_dim(), i)
def test_nested_dim(self):
nt = nestedtensor.nested_tensor([torch.tensor(3)])
for i in range(2, 5):
nt = utils.gen_nested_tensor(i, i, 3)
self.assertEqual(nt.nested_dim(), i)
def test_nested(self):
nt = utils.gen_nested_tensor(2, 2, 2)
tensor, mask = nt.to_tensor_mask()
