Retrieval and embeddings · Lesson 13 of 16 · 8 min

Compare retrieval configurations

Compare retrieval configurations on the same query set, corpus and relevance judgments. Otherwise a score change may come from different evidence rather than a better retriever.

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-retrieval-embeddings-013.json. Separate the recorded facts from your assumptions.
Hint & worked example
Compare retrieval configurations on the same query set, corpus and relevance judgments. Otherwise a score change may come from different evidence rather than a better retriever.
The complete case JSON is read successfully.
```
cat /opt/ai-lab/cases/aiops-retrieval-embeddings-013.json
```
Common mistake to avoid: Changing top-k is the only possible cause of a score difference.
2
Isolate the deciding signal
Select the same_corpus field from the case JSON. What does it establish—and what does it not establish?
Hint & worked example
Align the corpus versions before attributing the difference to top-k.
A successful JSON query isolates same_corpus.
```
jq .same_corpus /opt/ai-lab/cases/aiops-retrieval-embeddings-013.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: align-corpus, change-without-evidence, or ignore-the-signal. Save that one code followed by a newline in /home/learner/lab/aiops-retrieval-embeddings-013-decision.txt.
Hint & worked example
Align the corpus versions before attributing the difference to top-k.
The decision artifact contains the correct code, not just a matching command in history.
```
printf '%s\n' 'align-corpus' > /home/learner/lab/aiops-retrieval-embeddings-013-decision.txt
```
Common mistake to avoid: Changing top-k is the only possible cause of a score difference.

Understand the pattern

Compare retrieval configurations on the same query set, corpus and relevance judgments. Otherwise a score change may come from different evidence rather than a better retriever.

Interpret the evidence

Align the corpus versions before attributing the difference to top-k.

A tempting mistake

Changing top-k is the only possible cause of a score difference.

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-retrieval-embeddings-013.json

{
  "fixture": "Fictional training data; not a live measurement.",
  "baseline_top_k": 5,
  "candidate_top_k": 10,
  "same_queries": true,
  "same_corpus": false
}

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

$ jq .same_corpus /opt/ai-lab/cases/aiops-retrieval-embeddings-013.json

false

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

$ printf '%s\n' 'align-corpus' > /home/learner/lab/aiops-retrieval-embeddings-013-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 “align-corpus” to stop being the justified action. Do not silently replace missing evidence with a guess.

Explain it

What does the evidence in “Compare retrieval configurations” justify?Changing top-k is the only possible cause of a score difference.Make the change now and gather the missing evidence afterward.Align the corpus versions before attributing the difference to top-k.

Apply it elsewhere

A colleague concludes: “Changing top-k is the only possible cause of a score difference.” What should you do?Treat this fixture as a live measurement.Reject that conclusion; return to the case evidence and its stated limits.Accept it because the training command completed.

Keep the momentum.

Complete the lab, then explain and apply the pattern.

Next: Prepare an index rebuild plan 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