Evaluations and datasets · Lesson 1 of 18 · 8 min

Inspect a golden dataset

A golden dataset is a versioned set of representative evaluation cases. Keep its intended coverage and split separate from the data used to tune a candidate.

Open the lab Course map

Mastery 0/3Run itExplain itApply it

learner@clairos:~ $

↑ history · Tab complete · Ctrl+C cancel · Esc, Tab to leaveExit 0

Your objectives

Hints are always available. Wrong turns never cost XP.

1
Read the case evidence
Read /opt/ai-lab/cases/aiops-evals-datasets-001.json. Separate the recorded facts from your assumptions.
Hint & worked example
A golden dataset is a versioned set of representative evaluation cases. Keep its intended coverage and split separate from the data used to tune a candidate.
The complete case JSON is read successfully.
```
cat /opt/ai-lab/cases/aiops-evals-datasets-001.json
```
Common mistake to avoid: Use the held-out answers to tune and still call the score independent.
2
Isolate the deciding signal
Select the split field from the case JSON. What does it establish—and what does it not establish?
Hint & worked example
Preserve gold-v4 as held-out evaluation data rather than tuning directly on it.
A successful JSON query isolates split.
```
jq .split /opt/ai-lab/cases/aiops-evals-datasets-001.json
```
Common mistake to avoid: Using a different case or an unrelated signal as the justification.
3
Record your operational decision
Choose the justified code: preserve-held-out, change-without-evidence, or ignore-the-signal. Save that one code followed by a newline in /home/learner/lab/aiops-evals-datasets-001-decision.txt.
Hint & worked example
Preserve gold-v4 as held-out evaluation data rather than tuning directly on it.
The decision artifact contains the correct code, not just a matching command in history.
```
printf '%s\n' 'preserve-held-out' > /home/learner/lab/aiops-evals-datasets-001-decision.txt
```
Common mistake to avoid: Use the held-out answers to tune and still call the score independent.

Understand the pattern

A golden dataset is a versioned set of representative evaluation cases. Keep its intended coverage and split separate from the data used to tune a candidate.

Interpret the evidence

Preserve gold-v4 as held-out evaluation data rather than tuning directly on it.

A tempting mistake

Use the held-out answers to tune and still call the score independent.

All names, measurements, prices, versions, and thresholds in this exercise are fictional training data. The evidence supports this bounded scenario, not a claim about a live provider.

Worked example and expected result

$ cat /opt/ai-lab/cases/aiops-evals-datasets-001.json

{
  "fixture": "Fictional training data; not a live measurement.",
  "dataset": "gold-v4",
  "cases": 100,
  "split": "held-out",
  "used_for_tuning": false
}

Exit status: 0. The command completed successfully in the training fixture.

$ jq .split /opt/ai-lab/cases/aiops-evals-datasets-001.json

"held-out"

Exit status: 0. The command completed successfully in the training fixture.

$ printf '%s\n' 'preserve-held-out' > /home/learner/lab/aiops-evals-datasets-001-decision.txt

(no output; exit 0)

Exit status: 0. The command completed successfully in the training fixture.

Try a variation

Name the assumption that would have to change for “preserve-held-out” to stop being the justified action. Do not silently replace missing evidence with a guess.

Explain it

What does the evidence in “Inspect a golden dataset” justify?Make the change now and gather the missing evidence afterward.Preserve gold-v4 as held-out evaluation data rather than tuning directly on it.Use the held-out answers to tune and still call the score independent.

Apply it elsewhere

A colleague concludes: “Use the held-out answers to tune and still call the score independent.” What should you do?Reject that conclusion; return to the case evidence and its stated limits.Accept it because the training command completed.Treat this fixture as a live measurement.

Keep the momentum.

Complete the lab, then explain and apply the pattern.

Next: Read evaluation cases Try an incident drill

Anonymous lesson and mastery progress lasts for this session. Sign in to keep synced lesson XP between sessions; mastery stays device-local. No speed bonus, lost lives, or hint penalties.

Keep learning

Case methodology: evidence, limits, decision