The idea

  • Use an LLM to evaluate your LLM's outputs; works with or without ground truth 2:54
  • Example judge prompts: "Is this response polite?" or "Does this address the question well?" 2:54

When it shines

  • Especially useful for nuanced or subjective tests 2:54
  • Having an LLM judge another LLM may seem odd, but it is a very practical approach 2:54
  • Covered in depth in a dedicated follow-up video 3:10