def load(self, path):
if os.path.isfile(self._keras_module_name_path(path)):
with open(self._keras_module_name_path(path), "rb") as text_file:
keras_module_name = text_file.read().decode("utf-8")
try:
keras_module = importlib.import_module(keras_module_name)
except ImportError:
raise BentoMLArtifactLoadingException(
"Failed to import '{}' module when loading saved "
"KerasModelArtifact".format(keras_module_name))
self.creat_session()
if self.custom_objects is None and os.path.isfile(
self._custom_objects_path(path)):
self.custom_objects = cloudpickle.load(
open(self._custom_objects_path(path), 'rb'))
with self.graph.as_default():
with self.sess.as_default():
# load keras model via json and weights if requested
if self._store_as_json_and_weights:
with open(self._model_json_path(path), 'r') as json_file:
model_json = json_file.read()
model = keras_module.models.model_from_json(
model_json, custom_objects=self.custom_objects)
model.load_weights(self._model_weights_path(path))
# otherwise, load keras model via standard load_model
else:
model = keras_module.models.load_model(
self._model_file_path(path),
custom_objects=self.custom_objects)
return self.pack(model)
def load(self, path):
try:
import torch
except ImportError:
raise ImportError(
"torch package is required to use PytorchModelArtifact")
model = cloudpickle.load(open(self._file_path(path), 'rb'))
if not isinstance(model, torch.nn.Module):
raise TypeError(
"Expecting PytorchModelArtifact loaded object type to be "
"'torch.nn.Module' but actually it is {}".format(type(model)))
return self.pack(model)
def load(self, path):
if tf is None:
raise ImportError(
"Tensorflow package is required to use KerasModelArtifact.")
self.creat_session()
if self.custom_objects is None and os.path.isfile(
self._custom_objects_path(path)):
self.custom_objects = cloudpickle.load(
open(self._custom_objects_path(path), 'rb'))
with self.graph.as_default():
with self.sess.as_default():
model = keras.models.load_model(
self._model_file_path(path),
custom_objects=self.custom_objects)
return self.pack(model)
def load(self, path):
try:
import torch
except ImportError:
raise MissingDependencyException(
"torch package is required to use PytorchModelArtifact")
# TorchScript Models are saved as zip files
if zipfile.is_zipfile(self._file_path(path)):
model = torch.jit.load(self._file_path(path))
else:
model = cloudpickle.load(open(self._file_path(path), 'rb'))
if not isinstance(model, torch.nn.Module):
raise InvalidArgument(
"Expecting PytorchModelArtifact loaded object type to be "
"'torch.nn.Module' or 'torch.jit.ScriptModule' \
but actually it is {}".format(type(model)))
return self.pack(model)
