Concept
LLM-as-judge
LLM-as-judge means using a large language model to score or rank outputs instead of relying on human raters or exact-match metrics. It fills the gap where BLEU, ROUGE, or accuracy fall apart: open-ended tasks like summaries, generated slides, or chatbot replies, where…
The rest of “LLM-as-judge” is a premium feature: every concept in the library gets a precise, practitioner-focused write-up like this one, cross-linked straight from the paper summaries that use it.
Log in to unlock→