where to get the "prompt" in context required model

[Elktrn]

Hello and thank you for sharing such a genius work.

As you have mentioned in the research paper the DIPPER model with context which i think provided in the prompt of the model would work better that DIPPEER-no-context model.

I wonder for a dataset like Xsum or a generated sample by a machine or human where we should get the "context" in real world scenario?

Best reagards