LLM-as-judge
Using another LLM to score an output against criteria (also used in evaluation).
Using another LLM to score an output against criteria (also used in evaluation).
Using another LLM to score or compare outputs against criteria — a fast, scalable way to run evals.
Using another LLM call to grade an output against a rubric. Scales nuanced evaluation; must be validated against humans. (Mod 4)