Replies: 1 comment
-
I think the current workflow/plan for reward model training would not be able to use the labels, we only want rankings for the synthetic data as it stands. We don't want to ask site users to do extra work if it won't be utilised. If someone from the ML team wants to correct me feel free though |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
There should be a way to report harmful assistant replies. Especially when they're synthetic, I'd like to be able to take more drastic action than ranking them worst.
Beta Was this translation helpful? Give feedback.
All reactions