Good questions for testing RAG

Test questions map real needs. A good set also includes questions the documents cannot answer.

You can follow this course without an account or programming. Examples use synthetic data and AI outcomes need checking.

How it works, step by step

Define success before optimizing

Prepare representative requests and expected answers or properties. Include ordinary cases, ambiguity, missing data and tool errors. For RAG, measure evidence retrieval separately from answer correctness. Recall@k measures the share of relevant evidence found in the first k results; precision@k measures the share of returned results that are relevant. For agents, also track actions, permissions and stopping. An LLM-as-judge can speed evaluation but needs calibration against humans. Do not change prompt, model and index simultaneously if you want to understand what improved results.

RAG accesses evidence, it does not retrain

RAG means retrieval-augmented generation: retrieval supplies relevant passages before the model answers. Ingestion prepares documents, while query execution searches the index and builds context. An embedding model represents text numerically; the response model writes the answer. Rerankers, vision, graph and router models have different roles when needed. RAG helps with private or changing information but does not guarantee truth. Use an authorized API for balances, exact inventory or permissions; an old document is not a reliable source for current state.

Fluency and truth are checked separately

A hallucination is unsupported or incorrect content presented as an answer. It can arise from missing information, ambiguity or incorrectly combined patterns. Ask for evidence and check the original document, date, units and conditions. A real citation may still fail to support the claim. Use current sources for current facts, a calculator for arithmetic, and “cannot determine” when evidence is missing. Do not treat confidence expressed in prose as a calibrated probability.

The visual map

Follow the solid arrows for the main flow. Dashed blue arrows supply data or context; dashed pink arrows show feedback or returning results. Colors and shapes distinguish models, stores, decisions and outputs. On smaller screens, scroll horizontally to follow the entire diagram.

A complete example

Create four questions each about deadlines, exact codes, relationships, ambiguity and absent data. Record evidence and expected behavior for each.

Try it yourself

1. Complete 20 rows with question, source, expected answer and category.
2. Record the input, source and expected outcome before running the experiment. Use only the fictional data in the example.
3. Follow the diagram stages. At every step record what information is received and produced; do not confuse intermediate output with the final outcome.
4. Repeat after removing necessary information or making the input ambiguous. Check whether the system clarifies, stops or invents an answer.
5. Compare with the explained solution. Keep the configuration, date, result and an explanation for differences. Change one thing and retest.

An explained solution

You have 20 cases including “information unavailable”; invented content does not count as completeness. A successful exercise lets you show the connection between input, stages and outcome. When information is missing, a cautious answer is more useful than invented details. Compare more than style: check conditions, sources and operations too.

When it helps and what can go wrong

Verbatim document questions are too easy as the only test. Choose this approach when it improves a measured need. Keep a simple baseline and compare outcomes using identical inputs. One successful example does not establish reliability in every situation.

Check your understanding

Does a good style score prove agent success?

No. The agent may write beautifully after using the wrong tool or data.

Does RAG change LLM parameters?

No. It supplies request-time context; parameter changes belong to training.

What outcome should this exercise produce?

You have 20 cases including “information unavailable”; invented content does not count as completeness.

Words to remember

• Regresie / Regression: A change that breaks a previously correct case.
• Retrieval: Finding relevant information in accessible sources.
• Grounding: Grounding an answer in verifiable evidence.

Sources and your next step

• LangGraph — Workflows and agents
• Lewis et al. — Retrieval-Augmented Generation
• Hugging Face — LLM Course

To prepare: D03 — Chunking and metadata