def predict_proba(self, connector, batch_size=128):
nclasses = self.service_parameters_mllib["nclasses"]
self.predict_parameters_input = connector.predict_parameters_input
self.predict_parameters_mllib = {
"gpu": self.service_parameters_mllib["gpu"],
"net": {"test_batch_size": batch_size}}
if self.gpu:
self.predict_parameters_mllib.update({"gpuid": self.gpuid})
self.predict_parameters_output = {"best": nclasses}
if connector.name == "svm":
data = [connector.path]
if connector.name == "lmdb":
data = [connector.path]
elif connector.name == "array":
if type(connector.X) == np.ndarray:
data = ndarray_to_sparse_strings(connector.X)
elif sparse.issparse(connector.X):
data = sparse_to_sparse_strings(connector.X)
y_score = self._predict_proba(data,
connector.predict_parameters_input,
self.predict_parameters_mllib,
self.predict_parameters_output)
return y_score
def predict_proba(self, X):
data = [X]
if type(X) == np.ndarray:
data = ndarray_to_sparse_strings(X)
elif sparse.issparse(X):
data = sparse_to_sparse_strings(X)
nclasses = self.service_parameters_mllib['nclasses']
self.predict_parameters_input = {}
self.predict_parameters_mllib = {
"gpu": self.service_parameters_mllib['gpu'],
"gpuid ": self.service_parameters_mllib['gpuid']
}
self.predict_parameters_output = {'best': nclasses}
json_dump = self.post_predict(self.sname, data,
self.predict_parameters_input,
self.predict_parameters_mllib,
self.predict_parameters_output)
self.answers.append(json_dump)
with open("{}/model.json".format(self.model['repository'])) as f:
self.calls = [json.loads(line, encoding='utf-8') for line in f]
y_score = to_array(json_dump, nclasses)
return y_score