I referred Generate Answer API. My KB has around 2600 QnA pairs and I observed that when I used "top 10", queried using "question", it provided different top results compared to when I used "top 100". My assumption is when "top 10" is used, this API gets first 10 results for that "Question" and stops after that and provides score for them relatively, where as, when "top 100" is used it gets first 100 results and provides scores for them. Is my analysis right? In order to get correct responses should I have "top: [total pairs in KB]"?
Asked
Active
Viewed 31 times
