AWS re:Invent 2024 - Streamline RAG and model evaluation with Amazon Bedrock (AIM359)

[@alishaheen9217]

You do realize that you are actually playing a video rather than live demo? :) So you dont trust the own demo? LLM as a judge being presented here is very very superficial. You are basically using a probabilistic model to judge another probabilistic model which by definition is just wrong. I could go into detail of how this entire approach is just superficial and wrong but may be thats for a full article.