BentoML

<@U01NVN84Y57> can you use the `PickleArtifact`  with pickle_module pointed to joblib?

<https://docs.bentoml.org/en/latest/api/artifacts.html#common-artifacts>

Great suggestion, I think that will work 

Thanks <@U01NVU1BL65> <@UKB4CLKP1> I can confirm that PickleArtifact worked as expected

bentoservice.py
`from bentoml.service.artifacts.common import PickleArtifact`

`@env(infer_pip_packages=True)`
`@artifacts([PickleArtifact('model', pickle_module='joblib')])`
`class MyBentoService(BentoService):`

    `@api(input=DataframeInput(), batch=True)`
    `def predict_proba(self, df: pd.DataFrame):`
        `return self.artifacts.model.predict_proba(df)`

main.py
`from bentoservice import MyBentoService`

`model=load('serialized.joblib')`

`# Create  service with the  trained model`
`bento_service = MyBentoService()`
`bento_service.pack("model", model)`