import bentoml
svc = bentoml.Service("test.bentob")
name="multi_ndarray_pred_runner",
batch=True,
)
echo_multi_ndarray_pred_runner = bentoml.picklable_model.load_runner(
"sk_model",
method_name="echo_multi_ndarray",
name="echo_multi_ndarray_pred_runner",
batch=True,
)
svc = bentoml.Service(
name="general_sync",
runners=[
json_echo_runner,
ndarray_pred_runner,
dataframe_pred_runner,
file_pred_runner,
multi_ndarray_pred_runner,
echo_multi_ndarray_pred_runner,
],
)
@svc.api(input=JSON(), output=JSON())
def echo_json(json_obj: JSONSerializable) -> JSONSerializable:
return json_echo_runner.run(json_obj)
@svc.api(
input=JSON(pydantic_model=_Schema),
output=JSON(),
import bentoml
# import bentoml.sklearn
# from bentoml.io import NumpyNdarray
# iris_model_runner = bentoml.sklearn.load_runner('iris_classifier:latest')
svc = bentoml.Service("test.simplebento",
# runners=[iris_model_runner]
)
# @svc.api(input=NumpyNdarray(), output=NumpyNdarray())
# def predict(request_data: np.ndarray):
# return iris_model_runner.predict(request_data)
# For simple use cases, only models list is required:
# svc.bento_options.models = []
# svc.bento_files.include = ["*"]
# svc.bento_env.pip_install = "./requirements.txt"