llm-as-a-judge

Testing LLM applications: how to test a system that answers differently every time

An application built on a large language model gives a slightly different answer to the same question every time, so the classic assertion of expected...