Any ways for the QnA model tp highlight the content of a answer?

Hi, I have been working on a Question and Answering System

I have done the optimization of the QnA model using the optimum library.
But During the inference I couldn’t find the context of a answer, just like we used to get in the Haystack library.

Is there any work around for this solution.

@lewtun do you have any solution for this?