def test_call_fit_wrong_value_mini_batch_size(sagemaker_session):
fm = FactorizationMachines(base_job_name='fm', sagemaker_session=sagemaker_session, **ALL_REQ_ARGS)
data = RecordSet('s3://{}/{}'.format(BUCKET_NAME, PREFIX), num_records=1, feature_dim=FEATURE_DIM,
channel='train')
with pytest.raises(ValueError):
fm.fit(data, 0)
def test_model_image(sagemaker_session):
fm = FactorizationMachines(sagemaker_session=sagemaker_session, **ALL_REQ_ARGS)
data = RecordSet('s3://{}/{}'.format(BUCKET_NAME, PREFIX), num_records=1, feature_dim=FEATURE_DIM, channel='train')
fm.fit(data, MINI_BATCH_SIZE)
model = fm.create_model()
assert model.image == registry(REGION, 'factorization-machines') + '/factorization-machines:1'
def test_model_image(sagemaker_session):
fm = FactorizationMachines(sagemaker_session=sagemaker_session, **ALL_REQ_ARGS)
data = RecordSet('s3://{}/{}'.format(BUCKET_NAME, PREFIX), num_records=1, feature_dim=FEATURE_DIM, channel='train')
fm.fit(data, MINI_BATCH_SIZE)
model = fm.create_model()
assert model.image == registry(REGION, 'factorization-machines') + '/factorization-machines:1'
def test_prepare_for_training_wrong_value_mini_batch_size(sagemaker_session):
fm = FactorizationMachines(base_job_name='fm', sagemaker_session=sagemaker_session, **ALL_REQ_ARGS)
data = RecordSet('s3://{}/{}'.format(BUCKET_NAME, PREFIX), num_records=1, feature_dim=FEATURE_DIM,
channel='train')
with pytest.raises(ValueError):
fm._prepare_for_training(data, 0)
def test_predictor_type(sagemaker_session):
fm = FactorizationMachines(sagemaker_session=sagemaker_session, **ALL_REQ_ARGS)
data = RecordSet('s3://{}/{}'.format(BUCKET_NAME, PREFIX), num_records=1, feature_dim=FEATURE_DIM, channel='train')
fm.fit(data, MINI_BATCH_SIZE)
model = fm.create_model()
predictor = model.deploy(1, TRAIN_INSTANCE_TYPE)
assert isinstance(predictor, FactorizationMachinesPredictor)
def test_predictor_type(sagemaker_session):
fm = FactorizationMachines(sagemaker_session=sagemaker_session, **ALL_REQ_ARGS)
data = RecordSet('s3://{}/{}'.format(BUCKET_NAME, PREFIX), num_records=1, feature_dim=FEATURE_DIM, channel='train')
fm.fit(data, MINI_BATCH_SIZE)
model = fm.create_model()
predictor = model.deploy(1, TRAIN_INSTANCE_TYPE)
assert isinstance(predictor, FactorizationMachinesPredictor)
def test_prepare_for_training_no_mini_batch_size(sagemaker_session):
fm = FactorizationMachines(base_job_name='fm',
sagemaker_session=sagemaker_session,
**ALL_REQ_ARGS)
data = RecordSet('s3://{}/{}'.format(BUCKET_NAME, PREFIX),
num_records=1,
feature_dim=FEATURE_DIM,
channel='train')
fm._prepare_for_training(data)
def test_call_fit(base_fit, sagemaker_session):
fm = FactorizationMachines(base_job_name='fm', sagemaker_session=sagemaker_session, **ALL_REQ_ARGS)
data = RecordSet('s3://{}/{}'.format(BUCKET_NAME, PREFIX), num_records=1, feature_dim=FEATURE_DIM, channel='train')
fm.fit(data, MINI_BATCH_SIZE)
base_fit.assert_called_once()
assert len(base_fit.call_args[0]) == 2
assert base_fit.call_args[0][0] == data
assert base_fit.call_args[0][1] == MINI_BATCH_SIZE
def test_call_fit(base_fit, sagemaker_session):
fm = FactorizationMachines(base_job_name='fm', sagemaker_session=sagemaker_session, **ALL_REQ_ARGS)
data = RecordSet('s3://{}/{}'.format(BUCKET_NAME, PREFIX), num_records=1, feature_dim=FEATURE_DIM, channel='train')
fm.fit(data, MINI_BATCH_SIZE)
base_fit.assert_called_once()
assert len(base_fit.call_args[0]) == 2
assert base_fit.call_args[0][0] == data
assert base_fit.call_args[0][1] == MINI_BATCH_SIZE
def test_all_hyperparameters(sagemaker_session):
fm = FactorizationMachines(
sagemaker_session=sagemaker_session,
epochs=2,
clip_gradient=1e2,
eps=0.001,
rescale_grad=2.2,
bias_lr=0.01,
linear_lr=0.002,
factors_lr=0.0003,
bias_wd=0.0004,
linear_wd=1.01,
factors_wd=1.002,
bias_init_method="uniform",
bias_init_scale=0.1,
bias_init_sigma=0.05,
bias_init_value=2.002,
linear_init_method="constant",
linear_init_scale=0.02,
linear_init_sigma=0.003,
linear_init_value=1.0,
factors_init_method="normal",
factors_init_scale=1.101,
factors_init_sigma=1.202,
factors_init_value=1.303,
**ALL_REQ_ARGS,
)
assert fm.hyperparameters() == dict(
num_factors=str(ALL_REQ_ARGS["num_factors"]),
predictor_type=ALL_REQ_ARGS["predictor_type"],
epochs="2",
clip_gradient="100.0",
eps="0.001",
rescale_grad="2.2",
bias_lr="0.01",
linear_lr="0.002",
factors_lr="0.0003",
bias_wd="0.0004",
linear_wd="1.01",
factors_wd="1.002",
bias_init_method="uniform",
bias_init_scale="0.1",
bias_init_sigma="0.05",
bias_init_value="2.002",
linear_init_method="constant",
linear_init_scale="0.02",
linear_init_sigma="0.003",
linear_init_value="1.0",
factors_init_method="normal",
factors_init_scale="1.101",
factors_init_sigma="1.202",
factors_init_value="1.303",
)
def test_model_image(sagemaker_session):
fm = FactorizationMachines(sagemaker_session=sagemaker_session, **ALL_REQ_ARGS)
data = RecordSet(
"s3://{}/{}".format(BUCKET_NAME, PREFIX),
num_records=1,
feature_dim=FEATURE_DIM,
channel="train",
)
fm.fit(data, MINI_BATCH_SIZE)
model = fm.create_model()
assert image_uris.retrieve("factorization-machines", REGION) == model.image_uri
def test_prepare_for_training_wrong_type_mini_batch_size(sagemaker_session):
fm = FactorizationMachines(base_job_name='fm',
sagemaker_session=sagemaker_session,
**ALL_REQ_ARGS)
data = RecordSet('s3://{}/{}'.format(BUCKET_NAME, PREFIX),
num_records=1,
feature_dim=FEATURE_DIM,
channel='train')
with pytest.raises((TypeError, ValueError)):
fm._prepare_for_training(data, 'some')
def test_prepare_for_training_wrong_value_mini_batch_size(sagemaker_session):
fm = FactorizationMachines(base_job_name="fm",
sagemaker_session=sagemaker_session,
**ALL_REQ_ARGS)
data = RecordSet(
"s3://{}/{}".format(BUCKET_NAME, PREFIX),
num_records=1,
feature_dim=FEATURE_DIM,
channel="train",
)
with pytest.raises(ValueError):
fm._prepare_for_training(data, 0)
def test_all_hyperparameters(sagemaker_session):
fm = FactorizationMachines(sagemaker_session=sagemaker_session,
epochs=2,
clip_gradient=1e2,
eps=0.001,
rescale_grad=2.2,
bias_lr=0.01,
linear_lr=0.002,
factors_lr=0.0003,
bias_wd=0.0004,
linear_wd=1.01,
factors_wd=1.002,
bias_init_method='uniform',
bias_init_scale=0.1,
bias_init_sigma=0.05,
bias_init_value=2.002,
linear_init_method='constant',
linear_init_scale=0.02,
linear_init_sigma=0.003,
linear_init_value=1.0,
factors_init_method='normal',
factors_init_scale=1.101,
factors_init_sigma=1.202,
factors_init_value=1.303,
**ALL_REQ_ARGS)
assert fm.hyperparameters() == dict(
num_factors=str(ALL_REQ_ARGS['num_factors']),
predictor_type=ALL_REQ_ARGS['predictor_type'],
epochs='2',
clip_gradient='100.0',
eps='0.001',
rescale_grad='2.2',
bias_lr='0.01',
linear_lr='0.002',
factors_lr='0.0003',
bias_wd='0.0004',
linear_wd='1.01',
factors_wd='1.002',
bias_init_method='uniform',
bias_init_scale='0.1',
bias_init_sigma='0.05',
bias_init_value='2.002',
linear_init_method='constant',
linear_init_scale='0.02',
linear_init_sigma='0.003',
linear_init_value='1.0',
factors_init_method='normal',
factors_init_scale='1.101',
factors_init_sigma='1.202',
factors_init_value='1.303',
)
def test_optional_hyper_parameters_value(sagemaker_session,
optional_hyper_parameters, value):
with pytest.raises(ValueError):
test_params = ALL_REQ_ARGS.copy()
test_params.update({optional_hyper_parameters: value})
FactorizationMachines(sagemaker_session=sagemaker_session,
**test_params)
def test_required_hyper_parameters_value(sagemaker_session,
required_hyper_parameters, value):
with pytest.raises(ValueError):
test_params = ALL_REQ_ARGS.copy()
test_params[required_hyper_parameters] = value
FactorizationMachines(sagemaker_session=sagemaker_session,
**test_params)
def test_init_required_named(sagemaker_session):
fm = FactorizationMachines(sagemaker_session=sagemaker_session, **ALL_REQ_ARGS)
assert fm.role == COMMON_TRAIN_ARGS['role']
assert fm.train_instance_count == COMMON_TRAIN_ARGS['train_instance_count']
assert fm.train_instance_type == COMMON_TRAIN_ARGS['train_instance_type']
assert fm.num_factors == ALL_REQ_ARGS['num_factors']
assert fm.predictor_type == ALL_REQ_ARGS['predictor_type']
def test_init_required_positional(sagemaker_session):
fm = FactorizationMachines('myrole', 1, 'ml.c4.xlarge', 3, 'regressor',
sagemaker_session=sagemaker_session)
assert fm.role == 'myrole'
assert fm.train_instance_count == 1
assert fm.train_instance_type == 'ml.c4.xlarge'
assert fm.num_factors == 3
assert fm.predictor_type == 'regressor'
def test_init_required_named(sagemaker_session):
fm = FactorizationMachines(sagemaker_session=sagemaker_session, **ALL_REQ_ARGS)
assert fm.role == COMMON_TRAIN_ARGS["role"]
assert fm.instance_count == COMMON_TRAIN_ARGS["instance_count"]
assert fm.instance_type == COMMON_TRAIN_ARGS["instance_type"]
assert fm.num_factors == ALL_REQ_ARGS["num_factors"]
assert fm.predictor_type == ALL_REQ_ARGS["predictor_type"]
def test_init_required_positional(sagemaker_session):
fm = FactorizationMachines(
"myrole", 1, "ml.c4.xlarge", 3, "regressor", sagemaker_session=sagemaker_session
)
assert fm.role == "myrole"
assert fm.instance_count == 1
assert fm.instance_type == "ml.c4.xlarge"
assert fm.num_factors == 3
assert fm.predictor_type == "regressor"
def test_predictor_custom_serialization(sagemaker_session):
fm = FactorizationMachines(sagemaker_session=sagemaker_session, **ALL_REQ_ARGS)
data = RecordSet(
"s3://{}/{}".format(BUCKET_NAME, PREFIX),
num_records=1,
feature_dim=FEATURE_DIM,
channel="train",
)
fm.fit(data, MINI_BATCH_SIZE)
model = fm.create_model()
custom_serializer = Mock()
custom_deserializer = Mock()
predictor = model.deploy(
1,
INSTANCE_TYPE,
serializer=custom_serializer,
deserializer=custom_deserializer,
)
assert isinstance(predictor, FactorizationMachinesPredictor)
assert predictor.serializer is custom_serializer
assert predictor.deserializer is custom_deserializer
def test_all_hyperparameters(sagemaker_session):
fm = FactorizationMachines(sagemaker_session=sagemaker_session,
epochs=2, clip_gradient=1e2, eps=0.001, rescale_grad=2.2,
bias_lr=0.01, linear_lr=0.002, factors_lr=0.0003,
bias_wd=0.0004, linear_wd=1.01, factors_wd=1.002,
bias_init_method='uniform', bias_init_scale=0.1, bias_init_sigma=0.05,
bias_init_value=2.002, linear_init_method='constant', linear_init_scale=0.02,
linear_init_sigma=0.003, linear_init_value=1.0, factors_init_method='normal',
factors_init_scale=1.101, factors_init_sigma=1.202, factors_init_value=1.303,
**ALL_REQ_ARGS)
assert fm.hyperparameters() == dict(
num_factors=str(ALL_REQ_ARGS['num_factors']),
predictor_type=ALL_REQ_ARGS['predictor_type'],
epochs='2',
clip_gradient='100.0',
eps='0.001',
rescale_grad='2.2',
bias_lr='0.01',
linear_lr='0.002',
factors_lr='0.0003',
bias_wd='0.0004',
linear_wd='1.01',
factors_wd='1.002',
bias_init_method='uniform',
bias_init_scale='0.1',
bias_init_sigma='0.05',
bias_init_value='2.002',
linear_init_method='constant',
linear_init_scale='0.02',
linear_init_sigma='0.003',
linear_init_value='1.0',
factors_init_method='normal',
factors_init_scale='1.101',
factors_init_sigma='1.202',
factors_init_value='1.303',
)
def test_image(sagemaker_session):
fm = FactorizationMachines(sagemaker_session=sagemaker_session, **ALL_REQ_ARGS)
assert fm.train_image() == registry(REGION) + '/factorization-machines:1'
def test_image(sagemaker_session):
fm = FactorizationMachines(sagemaker_session=sagemaker_session,
**ALL_REQ_ARGS)
assert fm.train_image() == registry(REGION) + "/factorization-machines:1"
def test_image(sagemaker_session):
fm = FactorizationMachines(sagemaker_session=sagemaker_session,
**ALL_REQ_ARGS)
assert image_uris.retrieve("factorization-machines",
REGION) == fm.training_image_uri()
