def test_tuning_lda(sagemaker_session):
with timeout(minutes=TUNING_DEFAULT_TIMEOUT_MINUTES):
data_path = os.path.join(DATA_DIR, 'lda')
data_filename = 'nips-train_1.pbr'
with open(os.path.join(data_path, data_filename), 'rb') as f:
all_records = read_records(f)
# all records must be same
feature_num = int(
all_records[0].features['values'].float32_tensor.shape[0])
lda = LDA(role='SageMakerRole',
train_instance_type='ml.c4.xlarge',
num_topics=10,
sagemaker_session=sagemaker_session,
base_job_name='test-lda')
record_set = prepare_record_set_from_local_files(
data_path, lda.data_location, len(all_records), feature_num,
sagemaker_session)
test_record_set = prepare_record_set_from_local_files(
data_path, lda.data_location, len(all_records), feature_num,
sagemaker_session)
test_record_set.channel = 'test'
# specify which hp you want to optimize over
hyperparameter_ranges = {
'alpha0': ContinuousParameter(1, 10),
'num_topics': IntegerParameter(1, 2)
}
objective_metric_name = 'test:pwll'
tuner = HyperparameterTuner(
estimator=lda,
objective_metric_name=objective_metric_name,
hyperparameter_ranges=hyperparameter_ranges,
objective_type='Maximize',
max_jobs=2,
max_parallel_jobs=2)
tuner.fit([record_set, test_record_set], mini_batch_size=1)
print('Started hyperparameter tuning job with name:' +
tuner.latest_tuning_job.name)
time.sleep(15)
tuner.wait()
best_training_job = tuner.best_training_job()
with timeout_and_delete_endpoint_by_name(best_training_job,
sagemaker_session):
predictor = tuner.deploy(1, 'ml.c4.xlarge')
predict_input = np.random.rand(1, feature_num)
result = predictor.predict(predict_input)
assert len(result) == 1
for record in result:
assert record.label['topic_mixture'] is not None
def test_ntm(sagemaker_session):
with timeout(minutes=TRAINING_DEFAULT_TIMEOUT_MINUTES):
data_path = os.path.join(DATA_DIR, 'ntm')
data_filename = 'nips-train_1.pbr'
with open(os.path.join(data_path, data_filename), 'rb') as f:
all_records = read_records(f)
# all records must be same
feature_num = int(all_records[0].features['values'].float32_tensor.shape[0])
ntm = NTM(role='SageMakerRole', train_instance_count=1, train_instance_type='ml.c4.xlarge', num_topics=10,
sagemaker_session=sagemaker_session, base_job_name='test-ntm')
record_set = prepare_record_set_from_local_files(data_path, ntm.data_location,
len(all_records), feature_num, sagemaker_session)
ntm.fit(record_set, None)
endpoint_name = unique_name_from_base('ntm')
with timeout_and_delete_endpoint_by_name(endpoint_name, sagemaker_session):
model = NTMModel(ntm.model_data, role='SageMakerRole', sagemaker_session=sagemaker_session)
predictor = model.deploy(1, 'ml.c4.xlarge', endpoint_name=endpoint_name)
predict_input = np.random.rand(1, feature_num)
result = predictor.predict(predict_input)
assert len(result) == 1
for record in result:
assert record.label["topic_weights"] is not None
def test_ntm_airflow_config_uploads_data_source_to_s3(sagemaker_session, cpu_instance_type):
with timeout(seconds=AIRFLOW_CONFIG_TIMEOUT_IN_SECONDS):
data_path = os.path.join(DATA_DIR, "ntm")
data_filename = "nips-train_1.pbr"
with open(os.path.join(data_path, data_filename), "rb") as f:
all_records = read_records(f)
# all records must be same
feature_num = int(all_records[0].features["values"].float32_tensor.shape[0])
ntm = NTM(
role=ROLE,
train_instance_count=SINGLE_INSTANCE_COUNT,
train_instance_type=cpu_instance_type,
num_topics=10,
sagemaker_session=sagemaker_session,
)
records = prepare_record_set_from_local_files(
data_path, ntm.data_location, len(all_records), feature_num, sagemaker_session
)
training_config = _build_airflow_workflow(
estimator=ntm, instance_type=cpu_instance_type, inputs=records
)
_assert_that_s3_url_contains_data(
sagemaker_session,
training_config["InputDataConfig"][0]["DataSource"]["S3DataSource"]["S3Uri"],
)
def test_ipinsights_airflow_config_uploads_data_source_to_s3(sagemaker_session, cpu_instance_type):
with timeout(seconds=AIRFLOW_CONFIG_TIMEOUT_IN_SECONDS):
data_path = os.path.join(DATA_DIR, "ipinsights")
data_filename = "train.csv"
with open(os.path.join(data_path, data_filename), "rb") as f:
num_records = len(f.readlines())
ipinsights = IPInsights(
role=ROLE,
train_instance_count=SINGLE_INSTANCE_COUNT,
train_instance_type=cpu_instance_type,
num_entity_vectors=10,
vector_dim=100,
sagemaker_session=sagemaker_session,
)
records = prepare_record_set_from_local_files(
data_path, ipinsights.data_location, num_records, None, sagemaker_session
)
training_config = _build_airflow_workflow(
estimator=ipinsights, instance_type=cpu_instance_type, inputs=records
)
_assert_that_s3_url_contains_data(
sagemaker_session,
training_config["InputDataConfig"][0]["DataSource"]["S3DataSource"]["S3Uri"],
)
def test_lda(sagemaker_session):
with timeout(minutes=15):
data_path = os.path.join(DATA_DIR, 'lda')
data_filename = 'nips-train_1.pbr'
with open(os.path.join(data_path, data_filename), 'rb') as f:
all_records = read_records(f)
# all records must be same
feature_num = int(all_records[0].features['values'].float32_tensor.shape[0])
lda = LDA(role='SageMakerRole', train_instance_type='ml.c4.xlarge', num_topics=10,
sagemaker_session=sagemaker_session, base_job_name='test-lda')
record_set = prepare_record_set_from_local_files(data_path, lda.data_location,
len(all_records), feature_num, sagemaker_session)
lda.fit(record_set, 100)
endpoint_name = name_from_base('lda')
with timeout_and_delete_endpoint_by_name(endpoint_name, sagemaker_session):
model = LDAModel(lda.model_data, role='SageMakerRole', sagemaker_session=sagemaker_session)
predictor = model.deploy(1, 'ml.c4.xlarge', endpoint_name=endpoint_name)
predict_input = np.random.rand(1, feature_num)
result = predictor.predict(predict_input)
assert len(result) == 1
for record in result:
assert record.label["topic_mixture"] is not None
def test_lda(sagemaker_session, cpu_instance_type):
job_name = unique_name_from_base("lda")
with timeout(minutes=TRAINING_DEFAULT_TIMEOUT_MINUTES):
data_path = os.path.join(DATA_DIR, "lda")
data_filename = "nips-train_1.pbr"
with open(os.path.join(data_path, data_filename), "rb") as f:
all_records = read_records(f)
# all records must be same
feature_num = int(all_records[0].features["values"].float32_tensor.shape[0])
lda = LDA(
role="SageMakerRole",
instance_type=cpu_instance_type,
num_topics=10,
sagemaker_session=sagemaker_session,
)
record_set = prepare_record_set_from_local_files(
data_path, lda.data_location, len(all_records), feature_num, sagemaker_session
)
lda.fit(records=record_set, mini_batch_size=100, job_name=job_name)
with timeout_and_delete_endpoint_by_name(job_name, sagemaker_session):
model = LDAModel(lda.model_data, role="SageMakerRole", sagemaker_session=sagemaker_session)
predictor = model.deploy(1, cpu_instance_type, endpoint_name=job_name)
predict_input = np.random.rand(1, feature_num)
result = predictor.predict(predict_input)
assert len(result) == 1
for record in result:
assert record.label["topic_mixture"] is not None
def test_ntm(sagemaker_session):
with timeout(minutes=TRAINING_DEFAULT_TIMEOUT_MINUTES):
data_path = os.path.join(DATA_DIR, 'ntm')
data_filename = 'nips-train_1.pbr'
with open(os.path.join(data_path, data_filename), 'rb') as f:
all_records = read_records(f)
# all records must be same
feature_num = int(all_records[0].features['values'].float32_tensor.shape[0])
ntm = NTM(role='SageMakerRole', train_instance_count=1, train_instance_type='ml.c4.xlarge', num_topics=10,
sagemaker_session=sagemaker_session, base_job_name='test-ntm')
record_set = prepare_record_set_from_local_files(data_path, ntm.data_location,
len(all_records), feature_num, sagemaker_session)
ntm.fit(record_set, None)
endpoint_name = name_from_base('ntm')
with timeout_and_delete_endpoint_by_name(endpoint_name, sagemaker_session):
model = NTMModel(ntm.model_data, role='SageMakerRole', sagemaker_session=sagemaker_session)
predictor = model.deploy(1, 'ml.c4.xlarge', endpoint_name=endpoint_name)
predict_input = np.random.rand(1, feature_num)
result = predictor.predict(predict_input)
assert len(result) == 1
for record in result:
assert record.label["topic_weights"] is not None
def test_ipinsights(sagemaker_session):
job_name = unique_name_from_base('ipinsights')
with timeout(minutes=TRAINING_DEFAULT_TIMEOUT_MINUTES):
data_path = os.path.join(DATA_DIR, 'ipinsights')
data_filename = 'train.csv'
with open(os.path.join(data_path, data_filename), 'rb') as f:
num_records = len(f.readlines())
ipinsights = IPInsights(role='SageMakerRole',
train_instance_count=1,
train_instance_type='ml.c4.xlarge',
num_entity_vectors=10,
vector_dim=100,
sagemaker_session=sagemaker_session)
record_set = prepare_record_set_from_local_files(
data_path, ipinsights.data_location, num_records, FEATURE_DIM,
sagemaker_session)
ipinsights.fit(records=record_set, job_name=job_name)
with timeout_and_delete_endpoint_by_name(job_name, sagemaker_session):
model = IPInsightsModel(ipinsights.model_data,
role='SageMakerRole',
sagemaker_session=sagemaker_session)
predictor = model.deploy(1, 'ml.c4.xlarge', endpoint_name=job_name)
assert isinstance(predictor, RealTimePredictor)
predict_input = [['user_1', '1.1.1.1']]
result = predictor.predict(predict_input)
assert len(result) == 1
for record in result:
assert record.label["dot_product"] is not None
def test_ntm_serverless_inference(sagemaker_session, cpu_instance_type):
job_name = unique_name_from_base("ntm-serverless")
with timeout(minutes=TRAINING_DEFAULT_TIMEOUT_MINUTES):
data_path = os.path.join(DATA_DIR, "ntm")
data_filename = "nips-train_1.pbr"
with open(os.path.join(data_path, data_filename), "rb") as f:
all_records = read_records(f)
# all records must be same
feature_num = int(all_records[0].features["values"].float32_tensor.shape[0])
ntm = NTM(
role="SageMakerRole",
instance_count=1,
instance_type=cpu_instance_type,
num_topics=10,
sagemaker_session=sagemaker_session,
)
record_set = prepare_record_set_from_local_files(
data_path, ntm.data_location, len(all_records), feature_num, sagemaker_session
)
ntm.fit(records=record_set, job_name=job_name)
with timeout_and_delete_endpoint_by_name(job_name, sagemaker_session):
model = NTMModel(ntm.model_data, role="SageMakerRole", sagemaker_session=sagemaker_session)
predictor = model.deploy(
serverless_inference_config=ServerlessInferenceConfig(), endpoint_name=job_name
)
assert isinstance(predictor, Predictor)
def test_tuning_lda(sagemaker_session):
with timeout(minutes=TUNING_DEFAULT_TIMEOUT_MINUTES):
data_path = os.path.join(DATA_DIR, 'lda')
data_filename = 'nips-train_1.pbr'
with open(os.path.join(data_path, data_filename), 'rb') as f:
all_records = read_records(f)
# all records must be same
feature_num = int(all_records[0].features['values'].float32_tensor.shape[0])
lda = LDA(role='SageMakerRole', train_instance_type='ml.c4.xlarge', num_topics=10,
sagemaker_session=sagemaker_session, base_job_name='test-lda')
record_set = prepare_record_set_from_local_files(data_path, lda.data_location,
len(all_records), feature_num, sagemaker_session)
test_record_set = prepare_record_set_from_local_files(data_path, lda.data_location,
len(all_records), feature_num, sagemaker_session)
test_record_set.channel = 'test'
# specify which hp you want to optimize over
hyperparameter_ranges = {'alpha0': ContinuousParameter(1, 10),
'num_topics': IntegerParameter(1, 2)}
objective_metric_name = 'test:pwll'
tuner = HyperparameterTuner(estimator=lda, objective_metric_name=objective_metric_name,
hyperparameter_ranges=hyperparameter_ranges, objective_type='Maximize', max_jobs=2,
max_parallel_jobs=2)
tuner.fit([record_set, test_record_set], mini_batch_size=1)
print('Started hyperparameter tuning job with name:' + tuner.latest_tuning_job.name)
time.sleep(15)
tuner.wait()
best_training_job = tuner.best_training_job()
with timeout_and_delete_endpoint_by_name(best_training_job, sagemaker_session):
predictor = tuner.deploy(1, 'ml.c4.xlarge')
predict_input = np.random.rand(1, feature_num)
result = predictor.predict(predict_input)
assert len(result) == 1
for record in result:
assert record.label['topic_mixture'] is not None
def test_object2vec(sagemaker_session, cpu_instance_type):
job_name = unique_name_from_base("object2vec")
with timeout(minutes=TRAINING_DEFAULT_TIMEOUT_MINUTES):
data_path = os.path.join(DATA_DIR, "object2vec")
data_filename = "train.jsonl"
with open(os.path.join(data_path, data_filename), "r") as f:
num_records = len(f.readlines())
object2vec = Object2Vec(
role="SageMakerRole",
instance_count=1,
instance_type=cpu_instance_type,
epochs=3,
enc0_max_seq_len=20,
enc0_vocab_size=45000,
enc_dim=16,
num_classes=3,
negative_sampling_rate=0,
comparator_list="hadamard,concat,abs_diff",
tied_token_embedding_weight=False,
token_embedding_storage_type="dense",
sagemaker_session=sagemaker_session,
)
record_set = prepare_record_set_from_local_files(
data_path, object2vec.data_location, num_records, FEATURE_NUM,
sagemaker_session)
object2vec.fit(records=record_set, job_name=job_name)
with timeout_and_delete_endpoint_by_name(job_name, sagemaker_session):
model = Object2VecModel(object2vec.model_data,
role="SageMakerRole",
sagemaker_session=sagemaker_session)
predictor = model.deploy(1, cpu_instance_type, endpoint_name=job_name)
assert isinstance(predictor, Predictor)
predict_input = {"instances": [{"in0": [354, 623], "in1": [16]}]}
result = predictor.predict(predict_input)
assert len(result) == 1
for record in result:
assert record.label["scores"] is not None
def test_object2vec(sagemaker_session):
with timeout(minutes=TRAINING_DEFAULT_TIMEOUT_MINUTES):
data_path = os.path.join(DATA_DIR, 'object2vec')
data_filename = 'train.jsonl'
with open(os.path.join(data_path, data_filename), 'r') as f:
num_records = len(f.readlines())
object2vec = Object2Vec(role='SageMakerRole',
train_instance_count=1,
train_instance_type='ml.c4.xlarge',
epochs=3,
enc0_max_seq_len=20,
enc0_vocab_size=45000,
enc_dim=16,
num_classes=3,
sagemaker_session=sagemaker_session,
base_job_name='test-object2vec')
record_set = prepare_record_set_from_local_files(
data_path, object2vec.data_location, num_records, FEATURE_NUM,
sagemaker_session)
object2vec.fit(record_set, None)
endpoint_name = unique_name_from_base('object2vec')
with timeout_and_delete_endpoint_by_name(endpoint_name, sagemaker_session):
model = Object2VecModel(object2vec.model_data,
role='SageMakerRole',
sagemaker_session=sagemaker_session)
predictor = model.deploy(1,
'ml.c4.xlarge',
endpoint_name=endpoint_name)
assert isinstance(predictor, RealTimePredictor)
predict_input = {'instances': [{"in0": [354, 623], "in1": [16]}]}
result = predictor.predict(predict_input)
assert len(result) == 1
for record in result:
assert record.label["scores"] is not None
def test_ipinsights_serverless_inference(sagemaker_session, cpu_instance_type):
job_name = unique_name_from_base("ipinsights-serverless")
with timeout(minutes=TRAINING_DEFAULT_TIMEOUT_MINUTES):
data_path = os.path.join(DATA_DIR, "ipinsights")
data_filename = "train.csv"
with open(os.path.join(data_path, data_filename), "rb") as f:
num_records = len(f.readlines())
ipinsights = IPInsights(
role="SageMakerRole",
instance_count=1,
instance_type=cpu_instance_type,
num_entity_vectors=10,
vector_dim=100,
sagemaker_session=sagemaker_session,
)
record_set = prepare_record_set_from_local_files(
data_path, ipinsights.data_location, num_records, FEATURE_DIM, sagemaker_session
)
ipinsights.fit(records=record_set, job_name=job_name)
with timeout_and_delete_endpoint_by_name(job_name, sagemaker_session):
model = IPInsightsModel(
ipinsights.model_data, role="SageMakerRole", sagemaker_session=sagemaker_session
)
predictor = model.deploy(
serverless_inference_config=ServerlessInferenceConfig(memory_size_in_mb=6144),
endpoint_name=job_name,
)
assert isinstance(predictor, Predictor)
predict_input = [["user_1", "1.1.1.1"]]
result = predictor.predict(predict_input)
assert len(result["predictions"]) == 1
assert 0 > result["predictions"][0]["dot_product"] > -1 # We expect ~ -0.22
def test_tuning_lda(sagemaker_session, cpu_instance_type):
with timeout(minutes=TUNING_DEFAULT_TIMEOUT_MINUTES):
data_path = os.path.join(DATA_DIR, "lda")
data_filename = "nips-train_1.pbr"
with open(os.path.join(data_path, data_filename), "rb") as f:
all_records = read_records(f)
# all records must be same
feature_num = int(
all_records[0].features["values"].float32_tensor.shape[0])
lda = LDA(
role="SageMakerRole",
instance_type=cpu_instance_type,
num_topics=10,
sagemaker_session=sagemaker_session,
)
record_set = prepare_record_set_from_local_files(
data_path, lda.data_location, len(all_records), feature_num,
sagemaker_session)
test_record_set = prepare_record_set_from_local_files(
data_path, lda.data_location, len(all_records), feature_num,
sagemaker_session)
test_record_set.channel = "test"
# specify which hp you want to optimize over
hyperparameter_ranges = {
"alpha0": ContinuousParameter(1, 10),
"num_topics": IntegerParameter(1, 2),
}
objective_metric_name = "test:pwll"
tuner = HyperparameterTuner(
estimator=lda,
objective_metric_name=objective_metric_name,
hyperparameter_ranges=hyperparameter_ranges,
objective_type="Maximize",
max_jobs=2,
max_parallel_jobs=2,
early_stopping_type="Auto",
)
tuning_job_name = unique_name_from_base("test-lda", max_length=32)
print("Started hyperparameter tuning job with name:" + tuning_job_name)
tuner.fit([record_set, test_record_set],
mini_batch_size=1,
job_name=tuning_job_name)
attached_tuner = HyperparameterTuner.attach(
tuning_job_name, sagemaker_session=sagemaker_session)
assert attached_tuner.early_stopping_type == "Auto"
assert attached_tuner.estimator.alpha0 == 1.0
assert attached_tuner.estimator.num_topics == 1
best_training_job = attached_tuner.best_training_job()
with timeout_and_delete_endpoint_by_name(best_training_job,
sagemaker_session):
predictor = tuner.deploy(1, cpu_instance_type)
predict_input = np.random.rand(1, feature_num)
result = predictor.predict(predict_input)
assert len(result) == 1
for record in result:
assert record.label["topic_mixture"] is not None
