Here's an experiment you can run on your eval setup in five minutes. Take one question from your test set and two candidate answers. Ask your LLM judge which is better. Then ask again with the two ans

Markdown supported · ⌘↵ to post
No responses yet.