def test_binary_torch_func(tensor_list, other_tensor_list, func_data):
""" Test torch function wich use two tensors """
func, _, same_base_name = func_data
masked_tensor = maskedtensor.from_list(tensor_list, dims=(0, 1))
other_base_name = 'N' if same_base_name else 'M'
other_masked_tensor = maskedtensor.from_list(other_tensor_list,
dims=(0, 1),
base_name=other_base_name)
res_mt = list(func(masked_tensor, other_masked_tensor))
binary_list = zip(tensor_list, other_tensor_list)
res_lst = apply_binary_list_tensors(binary_list, func)
for t_mt, t_lst in zip(res_mt, res_lst):
assert t_mt.size() == t_lst.size()
assert torch.allclose(t_mt, t_lst,
atol=ATOL), torch.norm(t_mt - t_lst,
p=float('inf'))
def masked_batch():
lst = [
torch.empty((N_VERTICES, N_VERTICES)).normal_()
for _ in range(BATCH_SIZE)
]
mtensor = maskedtensor.from_list(lst, dims=(0, 1))
return mtensor
def test_accuracy_func(tensor_list, func_data):
func, _ = func_data
masked_tensor = maskedtensor.from_list(tensor_list, dims=(0, 1))
res_mt = func(masked_tensor)
res_lst = sum(apply_list_tensors(tensor_list, func))
assert torch.allclose(res_mt, res_lst,
atol=ATOL), torch.norm(res_mt - res_lst,
p=float('inf'))
def test_score_func(score_list, func_data):
""" Test score function """
func, _ = func_data
masked_tensor = maskedtensor.from_list(score_list, dims=(0, 1))
res_mt = func(masked_tensor)
res_lst = func(torch.stack(score_list))
assert torch.allclose(res_mt, res_lst,
atol=ATOL), torch.norm(res_mt - res_lst,
p=float('inf'))
def test_torch_func(tensor_list, func_data):
""" Test torch function """
func, _ = func_data
masked_tensor = maskedtensor.from_list(tensor_list, dims=(0, 1))
res_mt = list(func(masked_tensor))
res_lst = apply_list_tensors(tensor_list, func)
for t_mt, t_lst in zip(res_mt, res_lst):
assert t_mt.size() == t_lst.size()
assert torch.allclose(t_mt, t_lst,
atol=ATOL), torch.norm(t_mt - t_lst,
p=float('inf'))
def batch(request):
transpose = request.param
if transpose:
tensor_lst = [
torch.t(perturb(torch.eye(n_vertices, n_vertices)))
for n_vertices in N_VERTICES_RANGE
]
else:
tensor_lst = [
perturb(torch.eye(n_vertices, n_vertices))
for n_vertices in N_VERTICES_RANGE
]
return maskedtensor.from_list(tensor_lst, dims=(0, 1))
def correct_batch():
tensor_lst = [torch.eye(n_vertices) for n_vertices in N_VERTICES_RANGE]
return from_list(tensor_lst, dims=(0, 1))
def batch():
tensor_lst = [
torch.empty(n_vertices, n_vertices).normal_()
for n_vertices in N_VERTICES_RANGE
]
return from_list(tensor_lst, dims=(0, 1))