Comparability of the results for different prompts

#9
by treehugg3 - opened

You cannot use this model to compare the same response against two different user prompts, correct?

NVIDIA org

No, you cannot - similar to other Bradley-Terry Reward Models.

Sign up or log in to comment