Evals And Observability easy
Your weekly offline eval scores have held steady, but online thumbs-down rates have crept up over two months. Logs show users increasingly ask about a newly launched product. What is the likely cause and fix?
- A The model weights changed silently; pin the model version and the thumbs-down rate will return to its earlier level.
- B Users have grown more critical over time; recalibrate the feedback signal and keep the current eval set as it is.
- C Input drift: the eval set no longer reflects traffic, so add cases about the new product from recent logs.
- D The eval is a one-time baseline that already passed, so look for a UI regression in the frontend instead.
Answer: C. Input drift: the eval set no longer reflects traffic, so add cases about the new product from recent logs.
Drift shows up as a slow change in the input distribution: a static eval set stays green while real traffic moves to topics it never covered. Refreshing the set from recent logs costs ongoing labeling effort but keeps the gate meaningful. Pinning the model version is good practice and the right fix when scores drop on unchanged inputs, but here the inputs changed, not the model.
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?