def load_module(workspace, namespace, name, yaml_file_path):
try:
module_func = Module.load(workspace=workspace, namespace=namespace, name=name)
print('found the module of {}'.format(name))
return module_func
except:
print('not found the module of {}, register it now...'.format(name))
module_func = Module.register(workspace=workspace, yaml_file=yaml_file_path)
return module_func
except:
print("Creating new compute target: {}".format(aml_compute_target))
provisioning_config = AmlCompute.provisioning_configuration(
vm_size="STANDARD_D2_V2", min_nodes=1, max_nodes=4)
aml_compute = ComputeTarget.create(workspace, aml_compute_target,
provisioning_config)
aml_compute.wait_for_completion(show_output=True,
min_node_count=None,
timeout_in_minutes=20)
# In[ ]:
try:
mpi_train_module_func = Module.load(
workspace,
namespace="microsoft.com/azureml/samples",
name="Hello World MPI Job")
except:
mpi_train_module_func = Module.register(
workspace, os.path.join('modules', 'mpi_module', 'module_spec.yaml'))
from azureml.pipeline.wrapper._dataset import get_global_dataset_by_path
blob_input_data = get_global_dataset_by_path(
workspace, 'Automobile_price_data',
'GenericCSV/Automobile_price_data_(Raw)')
mpi_train = mpi_train_module_func(input_path=blob_input_data,
string_parameter="test1")
mpi_train.runsettings.configure(node_count=2, process_count_per_node=2)
print(mpi_train.runsettings.node_count)
except:
print("Creating new compute target: {}".format(aml_compute_target))
provisioning_config = AmlCompute.provisioning_configuration(vm_size = "STANDARD_D2_V2",
min_nodes = 1,
max_nodes = 4)
aml_compute = ComputeTarget.create(ws, aml_compute_target, provisioning_config)
aml_compute.wait_for_completion(show_output=True, min_node_count=None, timeout_in_minutes=20)
# In[ ]:
# modules
try:
ejoin_module_func = Module.load(ws, namespace='microsoft.com/bing', name='ejoin')
eselect_module_func = Module.load(ws, namespace='microsoft.com/bing', name='eselect')
except:
ejoin_module_func = Module.register(ws, os.path.join('modules', 'ejoin', 'amlmodule.yaml'))
eselect_module_func = Module.register(ws, os.path.join('modules', 'eselect', 'amlmodule.yaml'))
training_data_name = "Titanic.tsv"
if training_data_name not in ws.datasets:
print('Registering a training dataset for sample pipeline ...')
train_data = Dataset.File.from_files(path=['https://desginerdemo.blob.core.windows.net/demo/titanic.tsv'])
train_data.register(workspace = ws,
name = training_data_name,
description = 'Training data (just for illustrative purpose)')
print('Registerd')
else:
train_data = ws.datasets[training_data_name]
# ---------------------------------------------------------
# Copyright (c) Microsoft Corporation. All rights reserved.
# ---------------------------------------------------------
from azureml.core import Workspace
from azureml.pipeline.wrapper import Module, dsl
ws = Workspace.from_config()
execute_python_script_module = Module.load(ws, namespace='azureml', name='Execute Python Script')
@dsl.pipeline(name='external sub0 graph', description='sub0')
def external_sub_pipeline0(input):
module1 = execute_python_script_module(
# should be pipeline input
dataset1=input,
)
module2 = execute_python_script_module(
dataset1=module1.outputs.result_dataset,
)
return module2.outputs
# register anonymous modules
import os
from azureml.pipeline.wrapper._module_registration import _load_anonymous_module
local_module = _load_anonymous_module(ws,
yaml_file=os.path.join(
'modules', 'hello_world',
'module_spec.yaml'))
github_yaml = "https://github.com/sherry1989/sample_modules/blob/master/3_basic_module/basic_module.yaml"
github_module = _load_anonymous_module(ws, yaml_file=github_yaml)
hello_world_module_id = local_module.module_version_id
basic_module_id = github_module.module_version_id
# In[ ]:
# get modules
hello_world_anonymous = Module.load(ws, id=hello_world_module_id)
basic_module_anonymous = Module.load(ws, id=basic_module_id)
# In[ ]:
# get dataset
from azureml.pipeline.wrapper._dataset import get_global_dataset_by_path
automobile_price_data_raw = get_global_dataset_by_path(
ws, 'automobile_price_data_raw', 'GenericCSV/Automobile_price_data_(Raw)')
# In[ ]:
# define pipeline
@dsl.pipeline(name='module_SDK_test Run 8575',
description='test local module',
print("Found existing compute target: {}".format(aml_compute_target))
except:
print("Creating new compute target: {}".format(aml_compute_target))
provisioning_config = AmlCompute.provisioning_configuration(vm_size="STANDARD_D2_V2",
min_nodes=1,
max_nodes=4)
aml_compute = ComputeTarget.create(ws, aml_compute_target, provisioning_config)
aml_compute.wait_for_completion(show_output=True, min_node_count=None, timeout_in_minutes=20)
# In[ ]:
# Module
select_columns_in_dataset = Module.load(ws, namespace='azureml', name='Select Columns in Dataset')
clean_missing_data = Module.load(ws, namespace='azureml', name='Clean Missing Data')
split_data = Module.load(ws, namespace='azureml', name='Split Data')
join_data = Module.load(ws, namespace='azureml', name='Join Data')
# Dataset
try:
dset = Dataset.get_by_name(ws, 'Automobile_price_data_(Raw)')
except Exception:
global_datastore = Datastore(ws, name="azureml_globaldatasets")
dset = Dataset.File.from_files(global_datastore.path('GenericCSV/Automobile_price_data_(Raw)'))
dset.register(workspace=ws,
name='Automobile_price_data_(Raw)',
create_new_version=True)
blob_input_data = dset
aml_compute = AmlCompute(workspace, aml_compute_target)
print("Found existing compute target: {}".format(aml_compute_target))
except:
print("Creating new compute target: {}".format(aml_compute_target))
provisioning_config = AmlCompute.provisioning_configuration(
vm_size="STANDARD_D2_V2", min_nodes=1, max_nodes=4)
aml_compute = ComputeTarget.create(workspace, aml_compute_target,
provisioning_config)
aml_compute.wait_for_completion(show_output=True,
min_node_count=None,
timeout_in_minutes=20)
try:
mpi_train_module_func = Module.load(
workspace,
namespace="microsoft.com/azureml/samples",
name="Hello World MPI Job")
except:
mpi_train_module_func = Module.register(
workspace, os.path.join('modules', 'mpi_module', 'module_spec.yaml'))
from azureml.pipeline.wrapper._dataset import get_global_dataset_by_path
blob_input_data = get_global_dataset_by_path(
workspace, 'Automobile_price_data',
'GenericCSV/Automobile_price_data_(Raw)')
mpi_train = mpi_train_module_func(input_path=blob_input_data,
string_parameter="test1")
mpi_train.runsettings.configure(node_count=2, process_count_per_node=2)
print(mpi_train.runsettings.node_count)
provisioning_config = AmlCompute.provisioning_configuration(
vm_size="STANDARD_D2_V2", min_nodes=1, max_nodes=4)
aml_compute = ComputeTarget.create(ws, aml_compute_target,
provisioning_config)
aml_compute.wait_for_completion(show_output=True,
min_node_count=None,
timeout_in_minutes=20)
# In[ ]:
# modules
try:
ejoin_module_func = Module.load(ws,
namespace='microsoft.com/bing',
name='ejoin')
eselect_module_func = Module.load(ws,
namespace='microsoft.com/bing',
name='eselect')
except:
ejoin_module_func = Module.register(
ws, os.path.join('modules', 'ejoin', 'amlmodule.yaml'))
eselect_module_func = Module.register(
ws, os.path.join('modules', 'eselect', 'amlmodule.yaml'))
join_data_module_func = Module.load(ws, namespace='azureml', name='Join Data')
train_svd_recommender_module_func = Module.load(ws,
namespace='azureml',
name='Train SVD Recommender')
aml_compute = AmlCompute(ws, aml_compute_target)
print("Found existing compute target: {}".format(aml_compute_target))
except:
print("Creating new compute target: {}".format(aml_compute_target))
provisioning_config = AmlCompute.provisioning_configuration(
vm_size="STANDARD_D2_V2", min_nodes=1, max_nodes=4)
aml_compute = ComputeTarget.create(ws, aml_compute_target,
provisioning_config)
aml_compute.wait_for_completion(show_output=True,
min_node_count=None,
timeout_in_minutes=20)
# In[ ]:
join_data_module_func = Module.load(ws, namespace='azureml', name='Join Data')
execute_python_script_module_func = Module.load(ws,
namespace='azureml',
name='Execute Python Script')
remove_duplicate_rows_module_func = Module.load(ws,
namespace='azureml',
name='Remove Duplicate Rows')
split_data_module_func = Module.load(ws,
namespace='azureml',
name='Split Data')
train_svd_recommender_module_func = Module.load(ws,
namespace='azureml',
name='Train SVD Recommender')
select_columns_module_func = Module.load(ws,
namespace='azureml',
name='Select Columns in Dataset')
except:
print("Creating new compute target: {}".format(aml_compute_target))
provisioning_config = AmlCompute.provisioning_configuration(
vm_size="STANDARD_D2_V2", min_nodes=1, max_nodes=4)
aml_compute = ComputeTarget.create(ws, aml_compute_target,
provisioning_config)
aml_compute.wait_for_completion(show_output=True,
min_node_count=None,
timeout_in_minutes=20)
# In[ ]:
try:
train_module_func = Module.load(ws,
namespace='microsoft.com/aml/samples',
name='Train')
except:
train_module_func = Module.register(
ws, os.path.join('modules', 'train-score-eval', 'train.yaml'))
try:
score_module_func = Module.load(ws,
namespace='microsoft.com/aml/samples',
name='Score')
except:
score_module_func = Module.register(
ws, os.path.join('modules', 'train-score-eval', 'score.yaml'))
try:
eval_module_func = Module.load(ws,