def test_raises_mismatched_dataset_comp_return_type_and_iterproc_sequence_type(
self):
iterproc = _create_federated_int_dataset_identity_iterative_process()
with self.assertRaises(TypeError):
iterative_process_compositions.compose_dataset_computation_with_iterative_process(
float_dataset_computation, iterproc)
def test_raises_non_computation(self):
fn = lambda _: []
iterproc = _create_federated_int_dataset_identity_iterative_process()
with self.assertRaises(TypeError):
iterative_process_compositions.compose_dataset_computation_with_iterative_process(
fn, iterproc)
def test_returns_iterproc_accepting_dataset_in_third_index_of_next(self):
iterproc = _create_stateless_int_dataset_reduction_iterative_process()
old_param_type = iterproc.next.type_signature.parameter
new_param_elements = [old_param_type[0], tf.int32, old_param_type[1]]
@computations.federated_computation(
computation_types.StructType(new_param_elements))
def new_next(param):
return iterproc.next([param[0], param[2]])
iterproc_with_dataset_as_third_elem = iterative_process.IterativeProcess(
iterproc.initialize, new_next)
expected_new_next_type_signature = computation_types.FunctionType([
computation_types.FederatedType(tf.int64, placements.SERVER),
tf.int32,
computation_types.FederatedType(tf.string, placements.CLIENTS)
], computation_types.FederatedType(tf.int64, placements.SERVER))
new_iterproc = iterative_process_compositions.compose_dataset_computation_with_iterative_process(
int_dataset_computation, iterproc_with_dataset_as_third_elem)
self.assertTrue(
expected_new_next_type_signature.is_equivalent_to(
new_iterproc.next.type_signature))
def test_mutates_iterproc_accepting_dataset_in_second_index_of_next(self):
iterproc = _create_stateless_int_dataset_reduction_iterative_process()
expected_new_next_type_signature = computation_types.FunctionType(
collections.OrderedDict(
server_state=computation_types.FederatedType(
tf.int64, placements.SERVER),
client_data=computation_types.FederatedType(
tf.string, placements.CLIENTS)),
computation_types.FederatedType(tf.int64, placements.SERVER))
new_iterproc = iterative_process_compositions.compose_dataset_computation_with_iterative_process(
int_dataset_computation, iterproc)
expected_new_next_type_signature.check_equivalent_to(
new_iterproc.next.type_signature)
def test_mutates_iterproc_with_parameter_assignable_from_result(self):
iterproc = _create_stateless_int_vector_unknown_dim_dataset_reduction_iterative_process(
)
expected_new_next_type_signature = computation_types.FunctionType(
collections.OrderedDict(
server_state=computation_types.FederatedType(
computation_types.TensorType(tf.int64, shape=[None]),
placements.SERVER),
client_data=computation_types.FederatedType(
tf.string, placements.CLIENTS)),
computation_types.FederatedType(
computation_types.TensorType(tf.int64, shape=[1]),
placements.SERVER))
new_iterproc = iterative_process_compositions.compose_dataset_computation_with_iterative_process(
vector_int_dataset_computation, iterproc)
expected_new_next_type_signature.check_equivalent_to(
new_iterproc.next.type_signature)
def test_raises_iterproc_if_dataset_is_returned_by_init(self):
iterproc = _create_federated_int_dataset_identity_iterative_process()
with self.assertRaises(TypeError):
iterative_process_compositions.compose_dataset_computation_with_iterative_process(
int_dataset_computation, iterproc)
def test_raises_iterative_process_no_dataset_parameter(self):
iterproc = _create_whimsy_iterative_process()
with self.assertRaises(TypeError):
iterative_process_compositions.compose_dataset_computation_with_iterative_process(
int_dataset_computation, iterproc)
def test_raises_non_iterative_process(self):
non_iterproc = lambda x: x
with self.assertRaises(TypeError):
iterative_process_compositions.compose_dataset_computation_with_iterative_process(
int_dataset_computation, non_iterproc)
