Choose a training objective from observed failures. Separate preference labels, demonstrated answers and adapter methods from factual grounding.
Updated Oct 2026 · Original practice questions with worked answers and teaching rubrics. Our editorial policy explains the sources, review process and limits.
A support-drafting assistant follows the requested format but gives unhelpful answers. The review buckets and collection quotes are fictional.
| 200 failures | 100 omit required procedure despite having evidence; 60 lack account context; 40 are unnecessarily long after completing the task. |
|---|---|
| Demonstration pilot | 400 attempts at $12 each; 320 accepted. |
| Preference pilot | 500 comparisons at $2 each; 450 accepted. |
| Additional budget | Training, rater calibration and independent evaluation costs are not included in those quotes. |
Your task. Choose the next intervention and feedback objective. Compare accepted-data costs, explain what training will not fix and define protected outcomes.
Choose a training objective from observed failures. Separate preference labels, demonstrated answers and adapter methods from factual grounding.
This reference answer is in Premium. Free signup opens other answers.
9 answers per topic instead of 5, and your progress kept · no cardor unlock all 420 remaining answersNo comments yet. How would you have answered?