feat(ms-ai-architect): S1 v2 targeted iteration — GATE PASS (recall 84.2%, precision 84.2%) [skip-docs]

Operatørvalg (c): én målrettet, prinsipiell prompt-iterasjon på den diagnostiserte
grounded-men-feil-feilmoden (eksakt-verdi-entailment). Terskel uendret; v1 frosset.

Full blind v2 fan-out (15 batcher Opus 4.8 xhigh, 255 P-påstander dekket):
- judge v2: recall 84.2% (32/38, PASS >=0.80, Wilson [69.6-92.6%]), presisjon 84.2%
  (PASS >=0.70), F1 0.842, slår staleness 0/38.
- Fiksen løste målet: sku recall 37.5%->75.0%, taxonomy 66.7%->100%.
- +6 ekte fangster (26->32) uten netto nye FP (6->6) => recall OG presisjon opp.

Forbehold (ærlig): andre måling på samme frosne sett etter v1 (erkjent); Wilson nedre
grense 69.6% < 0.80 ved n=38; én iterasjon. Gate-logikk => vei mot S3. Stoppet for
operatør-beslutning (S2/S3), eskalerer ikke selv.

run-judge-bakeoff.mjs: --results/--report-prefix flagg (v2 uten å klobbe v1). Suite 552/552.
This commit is contained in:
Kjell Tore Guttormsen 2026-06-26 21:23:50 +02:00
commit 4cd290c14b
6 changed files with 1436 additions and 3 deletions

View file

@ -0,0 +1,234 @@
{
"_meta": {
"source": "gold-correctness-set.json + judge-bakeoff-results.json",
"thresholds": {
"minRecall": 0.8,
"minPrecision": 0.7
},
"judged": 255
},
"population": {
"total": 255,
"verifiable": 240,
"positives": 38,
"negatives": 202,
"unsourcedInP": 15
},
"arms": {
"staleness": {
"tp": 0,
"fp": 0,
"fn": 38,
"tn": 202,
"positives": 38,
"negatives": 202,
"flagged": 0,
"precision": null,
"recall": 0,
"f1": null,
"recallWilson": {
"p": 0,
"low": 0,
"high": 0.09181293258383999
},
"precisionWilson": null
},
"judge": {
"tp": 32,
"fp": 6,
"fn": 6,
"tn": 196,
"positives": 38,
"negatives": 202,
"flagged": 38,
"precision": 0.8421052631578947,
"recall": 0.8421052631578947,
"f1": 0.8421052631578947,
"recallWilson": {
"p": 0.8421052631578947,
"low": 0.6958287736272311,
"high": 0.9255623777627731
},
"precisionWilson": {
"p": 0.8421052631578947,
"low": 0.6958287736272311,
"high": 0.9255623777627731
}
},
"hybrid": {
"tp": 32,
"fp": 6,
"fn": 6,
"tn": 196,
"positives": 38,
"negatives": 202,
"flagged": 38,
"precision": 0.8421052631578947,
"recall": 0.8421052631578947,
"f1": 0.8421052631578947,
"recallWilson": {
"p": 0.8421052631578947,
"low": 0.6958287736272311,
"high": 0.9255623777627731
},
"precisionWilson": {
"p": 0.8421052631578947,
"low": 0.6958287736272311,
"high": 0.9255623777627731
}
}
},
"sourceSilent": {
"onVerifiableNegative": 3,
"onVerifiableError": 2,
"agreesWithUnsourced": 5,
"disagreesWithUnsourced": 10
},
"byClaimType": {
"version": {
"tp": 6,
"fp": 0,
"fn": 1,
"tn": 21,
"positives": 7,
"negatives": 21,
"flagged": 6,
"precision": 1,
"recall": 0.8571428571428571,
"f1": 0.923076923076923,
"recallWilson": {
"p": 0.8571428571428571,
"low": 0.4868654966809701,
"high": 0.9743210440510252
},
"precisionWilson": {
"p": 1,
"low": 0.6096569663469354,
"high": 0.9999999999999999
}
},
"tpm": {
"tp": 4,
"fp": 0,
"fn": 1,
"tn": 20,
"positives": 5,
"negatives": 20,
"flagged": 4,
"precision": 1,
"recall": 0.8,
"f1": 0.888888888888889,
"recallWilson": {
"p": 0.8,
"low": 0.3755282641185388,
"high": 0.9637768390302125
},
"precisionWilson": {
"p": 1,
"low": 0.5100999795960008,
"high": 1
}
},
"region": {
"tp": 1,
"fp": 0,
"fn": 1,
"tn": 13,
"positives": 2,
"negatives": 13,
"flagged": 1,
"precision": 1,
"recall": 0.5,
"f1": 0.6666666666666666,
"recallWilson": {
"p": 0.5,
"low": 0.09452865480086614,
"high": 0.9054713451991339
},
"precisionWilson": {
"p": 1,
"low": 0.2065432914738929,
"high": 1
}
},
"status": {
"tp": 6,
"fp": 1,
"fn": 1,
"tn": 45,
"positives": 7,
"negatives": 46,
"flagged": 7,
"precision": 0.8571428571428571,
"recall": 0.8571428571428571,
"f1": 0.8571428571428571,
"recallWilson": {
"p": 0.8571428571428571,
"low": 0.4868654966809701,
"high": 0.9743210440510252
},
"precisionWilson": {
"p": 0.8571428571428571,
"low": 0.4868654966809701,
"high": 0.9743210440510252
}
},
"taxonomy": {
"tp": 9,
"fp": 5,
"fn": 0,
"tn": 83,
"positives": 9,
"negatives": 88,
"flagged": 14,
"precision": 0.6428571428571429,
"recall": 1,
"f1": 0.782608695652174,
"recallWilson": {
"p": 1,
"low": 0.7008472464490407,
"high": 1
},
"precisionWilson": {
"p": 0.6428571428571429,
"low": 0.3876400468214041,
"high": 0.8365550926279728
}
},
"sku": {
"tp": 6,
"fp": 0,
"fn": 2,
"tn": 14,
"positives": 8,
"negatives": 14,
"flagged": 6,
"precision": 1,
"recall": 0.75,
"f1": 0.8571428571428571,
"recallWilson": {
"p": 0.75,
"low": 0.40926987910258916,
"high": 0.9285223111419724
},
"precisionWilson": {
"p": 1,
"low": 0.6096569663469354,
"high": 0.9999999999999999
}
}
},
"gate": {
"pass": true,
"recallOk": true,
"precisionOk": true,
"beatsStaleness": true,
"thresholds": {
"minRecall": 0.8,
"minPrecision": 0.7
},
"reasons": [
"all criteria met"
]
}
}

View file

@ -0,0 +1,54 @@
# Judge bake-off-rapport — S1 (Fase 3 de-risk)
_Generert deterministisk av `run-judge-bakeoff.mjs` over `gold-correctness-set.json` + `judge-bakeoff-results.json`. Tall fra testet `lib/judge-bakeoff.mjs`. Ikke rediger for hånd — regenerer._
**Forhåndsregistrert gate (låst FØR fan-out):** recall ≥ 0.8, presisjon ≥ 0.7, OG judge-recall > staleness-recall.
## Evaluerings-populasjon (P)
Volatil stratum + fetchbare claim_types (price ekskludert) — der feilene bor; unngår «invertert leverage».
| metrikk | verdi |
|---|---|
| P totalt | 255 |
| Verifiserbare (correct/outdated/wrong) | 240 |
| Positive (reelle feil å fange) | 38 |
| Negative (correct) | 202 |
| Unsourced i P (kjørt, men utenfor P/R) | 15 |
## Arm-sammenligning (detektering over de 240 verifiserbare)
| arm | TP | FP | FN | TN | presisjon | recall | recall Wilson 95% | F1 |
|---|---|---|---|---|---|---|---|---|
| staleness (billig baseline) | 0 | 0 | 38 | 202 | n/a | 0.0% | [0.0%, 9.2%] | n/a |
| judge (per-påstand groundedness) | 32 | 6 | 6 | 196 | 84.2% | 84.2% | [69.6%, 92.6%] | 0.842 |
| hybrid (union) | 32 | 6 | 6 | 196 | 84.2% | 84.2% | [69.6%, 92.6%] | 0.842 |
## Judge per claim_type (verifiserbar delmengde)
| claim_type | positive | TP | FP | FN | presisjon | recall |
|---|---|---|---|---|---|---|
| taxonomy | 9 | 9 | 5 | 0 | 64.3% | 100.0% |
| sku | 8 | 6 | 0 | 2 | 100.0% | 75.0% |
| version | 7 | 6 | 0 | 1 | 100.0% | 85.7% |
| status | 7 | 6 | 1 | 1 | 85.7% | 85.7% |
| tpm | 5 | 4 | 0 | 1 | 100.0% | 80.0% |
| region | 2 | 1 | 0 | 1 | 100.0% | 50.0% |
## source_silent-diagnostikk
Judgen hentet siden men fant ikke verdien. Diagnostisk, ikke et flagg.
| signal | antall | tolkning |
|---|---|---|
| På verifiserbar feil | 2 | judge-bom: reell feil oversett via «kan ikke verifisere» |
| På verifiserbar correct | 3 | judge reproduserte ikke et korrekt faktum mennesket fant |
| Enig med unsourced | 5 | judge reproduserer den uverifiserbare grensen (godt) |
| Uenig med unsourced | 10 | judge hevdet grunnet/ugrunnet der mennesket ikke fant kilde |
## GATE: ✅ PASS — bygg S3
- recall 0.842 ≥ 0.8? **ja**
- presisjon 0.842 ≥ 0.7? **ja**
- slår staleness (recall 0.000)? **ja**
- begrunnelse: all criteria met

File diff suppressed because it is too large Load diff