def test_content_transform_grayscale_image(subtests, content_image,
impl_params, instance_norm):
content_image = F.rgb_to_grayscale(content_image)
edge_size = 16
hyper_parameters = paper.hyper_parameters(impl_params=impl_params,
instance_norm=instance_norm)
hyper_parameters.content_transform.edge_size = edge_size
content_transform = paper.content_transform(
impl_params=impl_params,
instance_norm=instance_norm,
hyper_parameters=hyper_parameters,
)
if instance_norm:
utils.make_reproducible()
actual = content_transform(content_image)
if impl_params:
if instance_norm:
# Since the transform involves an uncontrollable random component, we can't
# do a reference test here
return
transform_image = F.resize(content_image, edge_size)
else:
transform = transforms.CenterCrop(edge_size)
transform_image = transform(content_image)
desired = transform_image.repeat(1, 3, 1, 1)
ptu.assert_allclose(actual, desired)
def main(root):
make_reproducible()
save_versions(root)
save_rng_states(root)
generate_small_images(root)
generate_medium_image(root)
generate_large_image(root)
def test_make_reproducible_cudnn(mocker):
cudnn_mock = mocker.patch(
"pystiche_papers.utils.misc.torch.backends.cudnn")
cudnn_mock.is_available = lambda: True
utils.make_reproducible()
assert cudnn_mock.deterministic
assert not cudnn_mock.benchmark
def test_style_transform(subtests):
make_reproducible()
image = torch.rand(1, 1, 17, 32)
style_transform = paper.style_transform()
actual = style_transform(image)
desired = F.resize(image, [136, 256])
ptu.assert_allclose(actual, desired)
def test_content_transform():
make_reproducible()
image = torch.rand(1, 1, 16, 31)
content_transform = paper.content_transform(impl_params=False)
actual = content_transform(image)
desired = F.resize(image[:, :, :16, :16], 256).repeat(1, 3, 1, 1)
ptu.assert_allclose(actual, desired)
def test_make_reproducible_seeds(subtests, mocker):
mocks = [(name, mocker.patch(f"pystiche_papers.utils.misc.{rel_import}"))
for name, rel_import in (
("standard library", "random.seed"),
("numpy", "np.random.seed"),
("torch", "torch.manual_seed"),
)]
seed = 123
utils.make_reproducible(seed)
for name, mock in mocks:
assert mock.call_args[0][0] == seed
def test_make_reproducible():
def get_random_tensors():
return torch.rand(10), torch.randn(10), torch.randint(10, (10, ))
utils.make_reproducible()
tensors1 = get_random_tensors()
utils.make_reproducible()
tensors2 = get_random_tensors()
tensors3 = get_random_tensors()
ptu.assert_allclose(tensors1, tensors2)
with pytest.raises(AssertionError):
ptu.assert_allclose(tensors2, tensors3)
def test_make_reproducible_no_standard_library(mocker):
mock = mocker.patch("pystiche_papers.utils.misc.random.seed")
utils.make_reproducible(seed_standard_library=False)
assert not mock.called
def test_make_reproducible_uint32_seed():
uint32_max = 2**32 - 1
assert utils.make_reproducible(uint32_max) == uint32_max
assert utils.make_reproducible(uint32_max + 1) == 0
