def test_serialize_sklearn_model(self):
class MyModeler(MLModeler):
def load_data(self) -> dict:
data = load_iris()
return {'X': data.data, 'y': data.target}
def train(self, prepared_data):
model = LogisticRegression(random_state=0)
X, y = prepared_data['X'], prepared_data['y']
model.fit(X, y)
return model
class MyModel(MLModel):
pass
my_modeler = MyModeler()
my_modeler.model_class = MyModel
model = my_modeler.build()
with tempfile.TemporaryDirectory() as path:
mm = ModelRepository(storage=LocalStorage(storage_path=path))
version = mm.persist(model=model)
model_2 = MyModel()
mm.load(model=model_2, version=version)
assert 'sklearn' in str(type(model_2.base_model))
def load(self, version: str = None) -> "Model":
"""
Load parameters from the specified `version` from the ModelRepository.
Leave version blank to load latest version.
"""
repo = ModelRepository.get_model_repo(self)
repo.load(model=self, version=version)
return self
def persist(self, version=None):
"""
Persist this model's properties to the ModelRepository. Currently, only `stats`, `metrics`, `model` properties are supported.
`model` property could be single model, list or dict of models
Currently, only sklearn and tensorflow-keras are supported.
:param version: model version, leave blank for autogeneration
:returns: model version
"""
repo = ModelRepository.get_model_repo(self)
return repo.persist(model=self, version=version)
def test_serialize_sklearn_model(self):
class MyModel(MLModel):
def __init__(self):
super().__init__()
self.base_model = LogisticRegression(random_state=0)
def train(self, prepared_data):
X, y = prepared_data['X'], prepared_data['y']
self.base_model.fit(X, y)
X, y = load_iris(return_X_y=True)
prepared_data = {'X': X, 'y': y}
model = MyModel()
model.train(prepared_data)
with tempfile.TemporaryDirectory() as path:
mm = ModelRepository(storage=LocalStorage(storage_path=path))
version = mm.persist(model=model)
model_2 = MyModel()
mm.load(model=model_2, version=version)
assert 'sklearn' in str(type(model_2.base_model))
def init_model_repo(cls, repo_path):
from h1st.model.repository import ModelRepository
if not hasattr(ModelRepository, 'MODEL_REPO'):
setattr(ModelRepository, 'MODEL_REPO',
ModelRepository(storage=repo_path))