AI & LLM Evals / Quiz

RAGAS: evaluate() and the Four Columns

user_input, response, retrieved_contexts, reference: the data schema behind ragas.evaluate(), and how its core metrics are actually computed.

Difficulty
Easy
Format
Quiz
Points
100
Estimate
8 min

// MISSION BRIEF

Your Mission

RAGAS evaluates RAG pipelines from a dataset of samples: the question, the answer, the fetched chunks, and (for some metrics) the ground truth. Get the columns right and evaluate() hands you per-metric, per-row scores; get them wrong and it hands you exceptions.

Six questions on the evaluate() workflow and what each core metric computes. Speed bonus active.

// FIRST CONTACT

Battle teaser

A RAGAS evaluation sample carries which fields?

  1. AOnly the model's answer; RAGAS infers the rest
  2. BA screenshot of the chat window
  3. Cuser_input (question), response (answer), retrieved_contexts (chunks), and, for ground-truth metrics, reference
  4. DThe model weights and tokenizer
Answers, scoring, hints, and the full battle stay sealed.

// SKILL TAGS

ragasragevaluatemetrics