Evals And Observability easy
You are choosing between two models for a legal-clause extraction feature. Model X scores higher on a popular general reasoning benchmark. How should you decide?
- A Pick Model X, because higher general benchmark accuracy reliably predicts better quality on downstream product tasks.
- B Pick whichever model has the lower p50 latency, since faster responses correlate with better perceived quality.
- C Ask each model to grade its own extractions on a 1 to 10 scale and choose the model reporting higher confidence.
- D Run both models on a task-specific eval set of real clauses with labeled extractions and compare the results.
Answer: D. Run both models on a task-specific eval set of real clauses with labeled extractions and compare the results.
A task-specific eval costs labeling time but measures the behavior your product actually needs. General benchmark accuracy is tempting because it is free and public, yet it often fails to predict performance on a narrow domain task. A benchmark is a reasonable first filter only when you have no labeled data yet and need to shortlist candidates.
One system design question a day, with spaced repetition so the ones you miss come back.
Get the appRelated questions
- In one release, a RAG assistant got a new embedding model and a new chunking strategy, and the end-to-end judge score dropped 6 points. The team only tracks the end-to-end score and has spent a week tweaking the generation prompt with no recovery. What is the best next step?
- Over two months, your team iterated through 40 prompt versions, each tuned against the same 150-case golden set. The offline score rose from 72 to 91. Online task completion and retry rate have not changed. What is the most likely problem, and what is the fix?
- A prompt change produced an impressive answer in a live demo for leadership. The PM wants to ship it today. What should you do?
- You plan to replace human graders with an LLM judge for your answer-quality eval. Before trusting its scores in a CI gate, what should you do first?
- After you swap to a smaller model, p95 latency falls from 900 ms to 350 ms and the ops dashboard is all green. Leadership asks whether answer quality is still fine. What is the right response?
- Your CI gate fails any prompt change that drops the score 3 or more points on a 100-point rubric over 60 cases. Production generates at temperature 0.7. Running the eval twice on the same commit gives scores that differ by up to 4 points, and engineers have started rerunning failed gates until they pass. How do you fix the gate?