Silvia

Results 1 comments of Silvia

I confirm the same situation. The reason seems to be that the factual_correctness used in answer_correctness has a different prompt compared to the one used in factual_correctness alone. It would...