C.W.K.
Stream
Lesson 02 of 10 · published

Golden Sets — Building and Maintaining Them

~16 min · evaluation, golden-sets

Level 0Apprentice
0 XP0/100 lessons0/14 achievements
0/120 XP to next level120 XP to go0% complete

Curated, not collected

A golden set is hand-curated examples with known-correct outputs. It's the smallest, most carefully chosen subset of your traffic, used to validate every change. 50 cases is plenty for most prompts; for high-stakes systems, 200–500.

Construction

  • Source — sample real traffic stratified by intent / category.
  • Reference outputs — produced by your best human reviewer or current best model.
  • Annotations — why this output is correct; what an alternative correct output would look like.
  • Versioning — golden sets evolve; tag versions and run new prompts against the version they were built on.

Maintenance

  • Add new cases when production surfaces a new failure mode.
  • Retire cases when the underlying behavior is deprecated.
  • Re-review reference outputs annually — what was correct last year may be wrong now.
  • Audit for representativeness — does the set still match traffic distribution?

Code

Golden set as JSONL·json
{"id": "refund-clear-policy-1", "category": "refund", "input": "...", "reference": "...", "reasoning": "30-day window; eligible.", "created": "2026-04-12", "reviewer": "alice"}
{"id": "refund-edge-shipping-2", "category": "refund", "input": "...", "reference": "...", "reasoning": "Damaged on arrival; full refund regardless of window.", "created": "2026-04-12", "reviewer": "alice"}

External links

Exercise

Build a 30-case golden set for one prompt. Stratify by intent. Annotate each reference output with why it's correct. Run your current prompt against it.

Progress

Progress is local-only — sign in to sync across devices.
Spotted a bug or have feedback on this page?Report an Issue

Comments 0

🔔 Reply notifications (sign in)
Sign inPlease sign in to comment.

No comments yet — be the first.