This message was deleted.

Hi Youssef, do you mean that after BentoML's new release (which version?), the probability values range returned by runner changes?

would you elaborate how could i switch from predict to predict proba using the runner. or could you hop on a fast call for help?

Hi Youssef, I think some examples will help you. When saving the model, you need to add the signature of

predict_proba

like this: https://github.com/bentoml/BentoML/blob/main/examples/sklearn/pipeline/train.py#L83 Then when use the runner to do the prediction, use

runner.predict_proba.run

like this: https://github.com/bentoml/BentoML/blob/main/examples/sklearn/pipeline/service.py#L21

thank you larme, that was very helpful and probelm solved.