feat(ms-ai-architect): Fase 0 steg 2 — verifiserings-fan-out (373 påstander) + gull-sett m/ deterministisk lastmod-join + TDD [skip-docs]
15 batcher (Opus 4.8 xhigh, read-only) verifiserte 55 sample-frame-filer mot live MS Learn (strikt v2 bevis-krav: correct/outdated/wrong krever hentet learn-sitat, ellers unsourced). 373 påstander: 259 correct, 29 outdated, 11 wrong, 74 unsourced. 40 reelle feil. lastmod_changed joinet i kode (registry sitemap_lastmod vs filens dato): 0 filer ville staleness-loopen flagget -> alle 40 feil er judge-unike (rå funn, krever tolkning i base-rate-rapporten). lib/base-rate.mjs (Wilson, lastmod, aggregering) TDD 15 tester. Suite 538/538. Steg 3 (compute-base-rate + gate) gjenstår.
This commit is contained in:
parent
f7fba5e4e7
commit
6985ed1f5d
19 changed files with 5239 additions and 0 deletions
4861
scripts/kb-eval/data/gold-correctness-set.json
Normal file
4861
scripts/kb-eval/data/gold-correctness-set.json
Normal file
File diff suppressed because it is too large
Load diff
Loading…
Add table
Add a link
Reference in a new issue