Hi community.
I would like to know if there is a way to change the replies given by the Generative Answer or the AI Model when the information is coming from uploaded documents on Azure OpenAI side. I've been noticing that the bot replies and at the end adds a reference where the information was taken but I would like to remove that link from the text shared by the bot. Is there a way to "train" the Generative Answers integration to let it know that it should not include references from X site?
Thank you!
Just to update this thread @MattJimison I see two very promising ideas already shared by @dewainr regarding this topic:
I hope we hear back soon since it is becoming a need for us to have this kind of flexibility on PVA as well.
Absolutely, thank you! It is what I was looking for 🙂
Thank you! @MattJimison for your prompt reply!
Yeah, I thought it may be an option to "parse" the information after it was shared to remove those links that I don't want. Also, we can't rate the answer because the thumbs down up built functionality gets lost when I use a single parsed message. That's why I was exhausting all options from Model perspective rather than manipulation of the message. I hope it get's improved soon! 🙂
Hi, @thelopyader .
Generally, the citations make for a good user experience because they help the user understand where the answer came from. This is a built-in part of the Bring Your Own Data experience that cannot be removed. However, you can opt to not automatically output the message in PVA, and instead modify the saved variable accordingly and then output it in your own message node. You can make these changes by going to the Generative Actions topic -> Create generative answers node -> "..." in top-right corner of the node -> Properties -> Click on "Advanced" at bottom of screen -> uncheck "Send a message"
Cheers,
Matt
Find this post helpful? Please mark it as the solution and/or provide kudos so that it will help others in the future.
Cheers,
Matt
Romain The Low-Code...
66
Pablo Roldan
55
stampcoin
10