The Importance of Verified Responses in Language Models

We have this verified mechanism. Can you talk about what that is? So, so there are certain questions for which we know factually what the answer is. One thing, language models are not good at is responding to just facts. They don't have any new knowledge since the training. It doesn't know any of those details. There are times when we decide for this question, there is no point in showing the language models response. So we'll just show you the very fight response.