def train_densenet(client, train_dataset_path, val_dataset_path, gpus, hours):
'''
Conducts training of model `PyDenseNetBc` on the CIFAR-10 dataset for IMAGE_CLASSIFICATION.
Demonstrates hyperparameter tuning with distributed parameter sharing on SINGA-Auto.
'''
task = 'IMAGE_CLASSIFICATION'
app_id = gen_id()
app = 'cifar10_densenet_{}'.format(app_id)
model_name = 'PyDenseNetBc_{}'.format(app_id)
print('Preprocessing datasets...')
load_cifar10(train_dataset_path, val_dataset_path)
print('Creating & uploading datasets onto SINGA-Auto...')
train_dataset = client.create_dataset('{}_train'.format(app), task,
train_dataset_path)
pprint(train_dataset)
val_dataset = client.create_dataset('{}_val'.format(app), task,
val_dataset_path)
pprint(val_dataset)
print('Creating model...')
model = client.create_model(
name=model_name,
task='IMAGE_CLASSIFICATION',
model_file_path='examples/models/image_classification/PyDenseNetBc.py',
model_class='PyDenseNetBc',
dependencies={
ModelDependency.TORCH: '1.0.1',
ModelDependency.TORCHVISION: '0.2.2'
})
pprint(model)
print('Creating train job...')
budget = {BudgetOption.TIME_HOURS: hours, BudgetOption.GPU_COUNT: gpus}
train_job = client.create_train_job(app,
task,
train_dataset['id'],
val_dataset['id'],
budget,
models=[model['id']])
pprint(train_job)
print('Monitor the train job on SINGA-Auto Web Admin')
def run_enas(client, train_dataset_path, val_dataset_path, gpus, hours):
'''
Conducts training of model `TfEnas` on the CIFAR-10 dataset for IMAGE_CLASSIFICATION.
Demonstrates architecture tuning with ENAS on SINGA-Auto.
'''
task = 'IMAGE_CLASSIFICATION'
app_id = gen_id()
app = 'cifar10_enas_{}'.format(app_id)
model_name = 'TfEnas_{}'.format(app_id)
print('Preprocessing datasets...')
load_cifar10(train_dataset_path, val_dataset_path)
print('Creating & uploading datasets onto SINGA-Auto...')
train_dataset = client.create_dataset('{}_train'.format(app), task,
train_dataset_path)
pprint(train_dataset)
val_dataset = client.create_dataset('{}_val'.format(app), task,
val_dataset_path)
pprint(val_dataset)
print('Creating model...')
model = client.create_model(
name=model_name,
task='IMAGE_CLASSIFICATION',
model_file_path='examples/models/image_classification/TfEnas.py',
model_class='TfEnas',
dependencies={ModelDependency.TENSORFLOW: '1.12.0'})
pprint(model)
print('Creating train job...')
budget = {BudgetOption.TIME_HOURS: hours, BudgetOption.GPU_COUNT: gpus}
train_job = client.create_train_job(app,
task,
train_dataset['id'],
val_dataset['id'],
budget,
models=[model['id']])
pprint(train_job)
print('Monitor the train job on SINGA-Auto Web Admin')
def quickstart(client, train_dataset_path, val_dataset_path, gpus, hours,
query_paths):
'''
Conducts a full train-inference flow on the Fashion MNIST dataset with
models `SkDt` and `TfFeedForward` for the task `IMAGE_CLASSIFICATION`.
'''
task = 'IMAGE_CLASSIFICATION'
# Randomly generate app & model names to avoid naming conflicts
app_id = gen_id()
app = 'image_classification_app_{}'.format(app_id)
tf_model_name = 'TfFeedForward_{}'.format(app_id)
sk_model_name = 'SkDt_{}'.format(app_id)
print('Preprocessing datasets...')
load_fashion_mnist(train_dataset_path, val_dataset_path)
print('Creating & uploading datasets onto SINGA-Auto...')
train_dataset = client.create_dataset('{}_train'.format(app), task,
train_dataset_path)
pprint(train_dataset)
val_dataset = client.create_dataset('{}_val'.format(app), task,
val_dataset_path)
pprint(val_dataset)
print('Adding models "{}" and "{}" to SINGA-Auto...'.format(
tf_model_name, sk_model_name))
tf_model = client.create_model(
tf_model_name,
task,
'examples/models/image_classification/TfFeedForward.py',
'TfFeedForward',
dependencies={ModelDependency.TENSORFLOW: '1.12.0'})
pprint(tf_model)
sk_model = client.create_model(
sk_model_name,
task,
'examples/models/image_classification/SkDt.py',
'SkDt',
dependencies={ModelDependency.SCIKIT_LEARN: '0.20.0'})
pprint(sk_model)
model_ids = [tf_model['id'], sk_model['id']]
print('Creating train job for app "{}" on SINGA-Auto...'.format(app))
budget = {BudgetOption.TIME_HOURS: hours, BudgetOption.GPU_COUNT: gpus}
train_job = client.create_train_job(app,
task,
train_dataset['id'],
val_dataset['id'],
budget,
models=model_ids)
pprint(train_job)
print('Waiting for train job to complete...')
print('This might take a few minutes')
wait_until_train_job_has_stopped(client, app)
print('Train job has been stopped')
print('Listing best trials of latest train job for app "{}"...'.format(app))
pprint(client.get_best_trials_of_train_job(app))
print('Creating inference job for app "{}" on SINGA-Auto...'.format(app))
pprint(client.create_inference_job(app))
predictor_host = get_predictor_host(client, app)
if not predictor_host:
raise Exception('Inference job has errored')
print('Inference job is running!')
print('Making predictions for query images:')
print(query_paths)
queries = utils.dataset.load_images(query_paths).tolist()
predictions = make_predictions(client, predictor_host, queries)
print('Predictions are:')
print(predictions)
print('Stopping inference job...')
pprint(client.stop_inference_job(app))