feat(ms-ai-architect): R7.3 bølge 3 — payload-26..35 dømt+ingestet (193 claims, 10 flagged, 54 flagg; ledger 119→129) [skip-docs]
This commit is contained in:
parent
7011898bd0
commit
6bfe24c7e2
1 changed files with 759 additions and 1 deletions
|
|
@ -4,7 +4,7 @@
|
|||
"derived_from": "scripts/kb-update/data/full-pass-worklist.json (243 due, all never-verified)",
|
||||
"cadence": "R7–R10 (5 økter × ~49, 8–10 samtidige)",
|
||||
"batch": "R7.1",
|
||||
"count": 119,
|
||||
"count": 129,
|
||||
"generated": "2026-07-18"
|
||||
},
|
||||
"files": [
|
||||
|
|
@ -5186,6 +5186,220 @@
|
|||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/multimodal-rag.md",
|
||||
"batch": "R7.3",
|
||||
"judged_at": "2026-07-25",
|
||||
"per_file_verdict": "flagged",
|
||||
"claim_count": 23,
|
||||
"verified": null,
|
||||
"verified_by": null,
|
||||
"flags": [
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/multimodal-rag.md#2",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/search/multimodal-search-overview",
|
||||
"evidence_quote": "In Azure AI Search, retrieving knowledge from images can follow two complementary paths: image verbalization or direct embeddings.",
|
||||
"reason": "Siden beskriver TO komplementære veier (verbalisering vs. direkte multimodale embeddinger), ikke tre; Content Understanding hører til ekstraksjonsaksen som ett av to anbefalte skills, og den navngitte modellen GPT-4v er avløst og står ikke i listen over støttede modeller for bildeverbalisering.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/multimodal-rag.md",
|
||||
"line": 28,
|
||||
"claim": "Azure-stakken tilbyr tre komplementære tilnærminger til multimodal RAG: Image verbalization (GPT-4o/4v konverterer bilder til tekst) | direkte multimodale embeddings (Azure Vision genererer vektorer for bilder og tekst i samme vektorrom) | Azure Content Understanding (konverterer komplekse dokumenter til Markdown).",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/multimodal-rag.md#4",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/cognitive-search-skill-document-extraction",
|
||||
"evidence_quote": "This skill extracts text and images. Text extraction is free. Image extraction is billable by Azure AI Search.",
|
||||
"reason": "To bærende deler er motsagt: skillen ekstraherer nettopp tekst, og formatlisten omfatter CSV, EML, EPUB, HTML, JSON, Markdown, DOCX, XLSX, PPTX, PDF, RTF, XML m.fl. - ikke kun PDF; kun table extraction er reelt Nei (PDFs only gjelder location metadata, ikke filstøtte).",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/multimodal-rag.md",
|
||||
"line": 40,
|
||||
"claim": "Document Extraction-skillen støtter kun PDF, ekstraherer bilder, men ikke tekst og ikke tabeller.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/multimodal-rag.md#12",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/ai-services/content-understanding/concepts/analyzer-reference",
|
||||
"evidence_quote": "Document analyzers ... Supported configuration options: returnDetails, omitContent, enableOcr, enableLayout, enableFormula, enableBarcode, tableFormat, chartFormat, enableFigureDescription, enableFigureAnalysis, enableAnnotations, annotationFormat, enableSegment, segmentPerPage, estimateFieldSourceAndConfidence, contentCategories",
|
||||
"reason": "Den kanoniske config-opplistingen har ingen outputContentFormat (det er Document Layout-skillens outputFormat), og feltet heter enableAnnotations i flertall, ikke enableAnnotation - begge er load-bearing API-strenger en leser ville kopiert inn i kode.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/multimodal-rag.md",
|
||||
"line": 177,
|
||||
"claim": "Konfigurasjonsparametere for Content Understanding i RAG-pipelines: outputContentFormat=markdown | enableFigureAnalysis=true | enableAnnotation=true | chartFormat=markdown.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/multimodal-rag.md#13",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/search-get-started-portal-image-search",
|
||||
"evidence_quote": "Models for image verbalization ... LLMs: - phi-4 - gpt-4o - gpt-4o-mini - gpt-5 - gpt-5-mini - gpt-5-nano",
|
||||
"reason": "Den kanoniske opplistingen av modeller for bildeverbalisering inneholder ingen GPT-4v eller vision-preview; modellsiden bekrefter i tillegg at gpt-4 turbo-2024-04-09 er Replacement for all previous GPT-4 preview models (vision-preview, 1106-Preview, 0125-Preview).",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/multimodal-rag.md",
|
||||
"line": 206,
|
||||
"claim": "GPT-4v har modellvarianten vision-preview og anbefales til bildeberikelse og summary-generering.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/multimodal-rag.md#15",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/search-get-started-portal-image-search",
|
||||
"evidence_quote": "If you're on a free search service, use fewer than 20 files to stay within the free quota for enrichment processing.",
|
||||
"reason": "Free tier støtter multimodal (med kvote) - Document Extraction-siden sier også at kostnaden for 20 transaksjoner per indexer per dag absorberes på en gratis søketjeneste; Basic-kravet i quickstarten gjelder managed identity support, ikke multimodal som sådan, så den første bærende delen er motsagt.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/multimodal-rag.md",
|
||||
"line": 216,
|
||||
"claim": "Free tier i Azure AI Search støtter ikke multimodal; minimum Basic tier kreves.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/multimodal-rag.md#19",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/cognitive-search-skill-vision-vectorize",
|
||||
"evidence_quote": "This skill is in preview under Supplemental Terms of Use. The latest preview version of Skillsets - Create Or Update (REST API) supports this feature.",
|
||||
"reason": "Ingen Learn-side stempler den multimodale pipelinen som GA; oversiktssiden har ingen GA-merking, mens Azure Vision multimodal embeddings-skillen og tilhørende vectorizer er eksplisitt preview, og wizard-genererte indekser følger den nyeste preview-REST-API-en.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/multimodal-rag.md",
|
||||
"line": 331,
|
||||
"claim": "Azure AI Search multimodal pipeline er GA.",
|
||||
"disposition": "outdated"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-caching-optimization.md",
|
||||
"batch": "R7.3",
|
||||
"judged_at": "2026-07-25",
|
||||
"per_file_verdict": "flagged",
|
||||
"claim_count": 21,
|
||||
"verified": null,
|
||||
"verified_by": null,
|
||||
"flags": [
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-caching-optimization.md#1",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/well-architected/ai/application-design#implement-multi-layer-caching-strategies",
|
||||
"evidence_quote": "Result and answer caching: Use this approach to reuse responses for identical or semantically similar queries ... Retrieval and grounding snippet caching: Cache frequently retrieved knowledge fragments and grounding data ... Model output caching: Cache intermediate model outputs that can be reused across requests.",
|
||||
"reason": "Den kanoniske WAF-siden ramser opp nøyaktig TRE cache-lag (result/answer, retrieval/grounding, model output); embedding caching er fraværende som eget nivå, semantisk caching er innbakt i result/answer-laget, og model output caching mangler i påstanden — fire-nivå-taksonomien stemmer ikke med kilden.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-caching-optimization.md",
|
||||
"line": 27,
|
||||
"claim": "Multi-layer caching i RAG dekker fire nivåer: result caching (hele LLM-responser) | retrieval caching (knowledge fragments fra vektorsøk) | embedding caching (forhåndsberegnede vektorrepresentasjoner) | semantic caching (semantisk like prompts).",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-caching-optimization.md#2",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/search-pagination-page-layout",
|
||||
"evidence_quote": "each query is independent and operates on the current view of the data as it exists in the index at query time (in other words, there's no caching or snapshot of results, such as those found in a general purpose database)",
|
||||
"reason": "Redis-, Cosmos DB- og APIM-delene er dekket av Learn, men den bærende delen «Azure AI Search (innebygd caching av søkeresultater)» motsies direkte: Azure AI Search cacher ikke søkeresultater — den eneste dokumenterte cachen er enrichment cache (preview) for skillset-output i Azure Storage.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-caching-optimization.md",
|
||||
"line": 29,
|
||||
"claim": "Microsoft-stakken tilbyr disse cache-tjenestene for AI-workloads: Azure Cache for Redis (tradisjonell og semantisk caching) | Azure Cosmos DB (semantisk cache med vektorsøk) | Azure AI Search (innebygd caching av søkeresultater) | Azure API Management (semantisk caching for LLM-API-er).",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-caching-optimization.md#3",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/well-architected/ai/application-design#implement-multi-layer-caching-strategies",
|
||||
"evidence_quote": "Result and answer caching: ... Retrieval and grounding snippet caching: Cache frequently retrieved knowledge fragments and grounding data to avoid repeated database and search queries or data API operations. Model output caching: Cache intermediate model outputs that can be reused across requests.",
|
||||
"reason": "Kilden lister tre cache-lag, ikke fire; «Embedding caching» finnes ikke i den kanoniske oppramsingen, så påstanden om fire lag er ikke dekket.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-caching-optimization.md",
|
||||
"line": 37,
|
||||
"claim": "Multi-layer caching-tabellen lister fire cache-lag: Result caching (cache hele LLM-responser) | Retrieval caching (cache knowledge fragments fra vector search) | Embedding caching (cache forhåndsberegnede embeddings) | Model output caching (cache intermediate model outputs).",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-caching-optimization.md#5",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/well-architected/ai/application-design#implement-multi-layer-caching-strategies",
|
||||
"evidence_quote": "Invalidation hooks: Implement cache invalidation triggers for data updates, model changes, and prompt modifications.",
|
||||
"reason": "Den kanoniske oppramsingen har nøyaktig tre invalideringstriggere; «manual purge (admin-utløst for compliance eller testing)» er fraværende, og søk fant ingen Learn-side som lister manuell purge som en fjerde trigger for AI-caching.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-caching-optimization.md",
|
||||
"line": 68,
|
||||
"claim": "Cache invalidation utløses av fire triggere: data updates (webhook-triggered ved endring i kildedata) | model changes (ved model deployment/retraining) | prompt modifications (ved endring av prompt-template) | manual purge (admin-utløst for compliance eller testing).",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-caching-optimization.md#7",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/azure-cache-for-redis/cache-overview",
|
||||
"evidence_quote": "Memory: The Basic and Standard tiers offer 250 MB – 53 GB; the Premium tier 6 GB - 1.2 TB; the Enterprise tier 1 GB - 2 TB, and the Enterprise Flash tier 300 GB - 4.5 TB.",
|
||||
"reason": "SLA-nivåene (99,9 % Premium / 99,99 % Enterprise) og Premium-kapasiteten holder, men den bærende delen «Enterprise Flash opptil 13 TB» motsies: siden oppgir 300 GB – 4,5 TB for Enterprise Flash.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-caching-optimization.md",
|
||||
"line": 186,
|
||||
"claim": "Azure Cache for Redis-tiers: Premium tier (99,9 % SLA, opptil 120 GB per shard) | Enterprise tier (99,99 % SLA, active-active geo-replikering, Flash-storage-støtte) | Enterprise Flash tier (opptil 13 TB cache-størrelse, 20 % RAM + 80 % NVMe Flash).",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-caching-optimization.md#11",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R7",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/api-management/llm-semantic-cache-lookup-policy",
|
||||
"evidence_quote": "Use the llm-semantic-cache-lookup policy to perform cache lookup of responses to large language model (LLM) API requests from a configured external cache, based on vector proximity of the prompt to previous requests and a specified score threshold.",
|
||||
"reason": "Attributtene og vary-by-underelementet stemmer, men policy-navnene er superseded: azure-openai-semantic-cache-lookup/-store heter nå llm-semantic-cache-lookup og llm-semantic-cache-store (den gamle URL-en redirigerer til llm-siden, og eksempel-XML-en på how-to-siden bruker llm-navnene). Navnet ER påstanden (R7s load-bearing-carve-out) — en leser kopierer strengen inn i policy-XML.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-caching-optimization.md",
|
||||
"line": 237,
|
||||
"claim": "APIM-policyene for semantisk cache er azure-openai-semantic-cache-lookup (inbound, med attributtene score-threshold, embeddings-backend-id, embeddings-backend-auth, ignore-system-messages, max-message-count og underelementet vary-by) og azure-openai-semantic-cache-store (outbound, med attributtet duration).",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-caching-optimization.md#12",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/api-management/llm-semantic-cache-lookup-policy",
|
||||
"evidence_quote": "Score threshold above 0.2 may lead to cache mismatch. Consider using lower value for sensitive use cases.",
|
||||
"reason": "Retningsdelen er grunngitt («Lower values require higher semantic similarity for a match»), men den bærende tre-bånds-rubrikken er ikke dokumentert og strider mot Learn-veiledningen: doc-en anbefaler å starte på 0,05 og advarer om cache-mismatch over 0,2, mens påstanden kaller 0,3–0,5 «balanced» og 0,6–0,8 «liberal».",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-caching-optimization.md",
|
||||
"line": 253,
|
||||
"claim": "APIM-attributtet score-threshold er en distanse der lavere verdi gir strengere matching og krever høyere semantisk likhet: 0,1–0,2 = strict matching | 0,3–0,5 = balanced | 0,6–0,8 = liberal matching.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-caching-optimization.md#14",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/cosmos-db/how-to-dotnet-vector-index-query",
|
||||
"evidence_quote": "After you decide on the vector embedding paths, you must add vector indexes to the indexing policy. ... The vector path is added to the excludedPaths section of the indexing policy to ensure optimized performance for insertion.",
|
||||
"reason": "Multi-region writes, 99,999 % og TTL er grunngitt, men den bærende delen «automatisk indeksering av vektorer» motsies: vektorindeks må defineres eksplisitt i indekseringspolicyen (kun ved container-opprettelse), og vektorstien legges i excludedPaths — altså holdes vektorer utenfor den automatiske indekseringen.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-caching-optimization.md",
|
||||
"line": 296,
|
||||
"claim": "Azure Cosmos DB som semantisk cache gir: global distribusjon med multi-region writes | automatisk indeksering av vektorer | 99,999 % SLA med multi-region-oppsett | innebygd TTL-støtte.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-caching-optimization.md#18",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/redis/entra-for-authentication",
|
||||
"evidence_quote": "Azure Managed Redis offers a password-free authentication mechanism by integrating with Microsoft Entra ID. Azure Managed Redis caches use Microsoft Entra ID by default. When you create a new cache, managed identity is enabled.",
|
||||
"reason": "Status-påstanden er utdatert: Entra ID-autentisering er ikke merket preview noe sted på den kanoniske siden (som derimot merker underfunksjonen «Configure custom data access permissions (preview)»), den siterte quickstart-en sier «Microsoft Entra ID is enabled by default», og What's New oppgir at Entra ID for autentisering og RBAC ble tilgjengelig i alle regioner i juni 2023.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-caching-optimization.md",
|
||||
"line": 361,
|
||||
"claim": "Microsoft Entra ID-autentisering for Redis er i preview.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-caching-optimization.md#19",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/azure-cache-for-redis/cache-best-practices-performance",
|
||||
"evidence_quote": "| F300 | 384 GB | 8 | 3,200 | 500,000 | 390,000 |",
|
||||
"reason": "E10 = 12 GB stemmer, men den bærende størrelsen for Enterprise Flash er feil: F300 er 384 GB, ikke 345 GB. I tillegg beskriver Learn Standard-tieren som «An OSS Redis cache running on two VMs in a replicated configuration» (primary + én replica), ikke 2 replicas.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-caching-optimization.md",
|
||||
"line": 373,
|
||||
"claim": "Azure Cache for Redis SKU-er med kapasitet: Basic C0 (250 MB, ingen SLA) | Standard C1 (1 GB, 2 replicas, 99,9 % SLA) | Premium P1 (6 GB, clustering, geo-replikering) | Enterprise E10 (12 GB, active-active, 99,99 % SLA) | Enterprise Flash F300 (345 GB, 20 % RAM + 80 % Flash).",
|
||||
"disposition": "outdated"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-context-windows.md",
|
||||
"batch": "R7.1",
|
||||
|
|
@ -5221,6 +5435,136 @@
|
|||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-core-patterns.md",
|
||||
"batch": "R7.3",
|
||||
"judged_at": "2026-07-25",
|
||||
"per_file_verdict": "flagged",
|
||||
"claim_count": 22,
|
||||
"verified": null,
|
||||
"verified_by": null,
|
||||
"flags": [
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-core-patterns.md#12",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/search/search-region-support",
|
||||
"evidence_quote": "You can create an Azure AI Search service in any of the following Azure public regions.",
|
||||
"reason": "Europa-tabellen i den kanoniske regionlisten for Azure AI Search fører opp Norway East, mens Norway West ikke forekommer i noen av regiontabellene — Norway West kan derfor ikke brukes for dataresidens i en RAG-løsning, sa den ene bærende delen av pastanden er feil.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-core-patterns.md",
|
||||
"line": 307,
|
||||
"claim": "Azure har norske regioner Norway East og Norway West som kan brukes for dataresidens i RAG-løsninger.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-core-patterns.md#13",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/search/search-limits-quotas-capacity",
|
||||
"evidence_quote": "",
|
||||
"reason": "Ingen Microsoft Learn-side angir S1 som SKU-en for en RAG-løsning med 10M vektorer; vektorkvoten for S1 (35 GB per partisjon, 12 partisjoner) motsier det heller ikke, sa pastanden kan verken bekreftes eller avkreftes — delen om at Semantic Ranker er et separat tillegg er derimot bekreftet.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-core-patterns.md",
|
||||
"line": 336,
|
||||
"claim": "Azure AI Search S1-tier er SKU-en angitt for en RAG-løsning med 10M vektorer, og Semantic Ranker kommer som et separat tillegg.",
|
||||
"disposition": "unsourced"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-core-patterns.md#18",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/semantic-kernel/memories/",
|
||||
"evidence_quote": "| Text Embeddings (Experimental) | ✅ | ✅ | ✅ | Example: Text-Embeddings-Ada-002 |",
|
||||
"reason": "Den siterte siden merker Text Embeddings som Experimental, ikke GA, og Vector Store-funksjonaliteten er merket preview — den ene bærende delen av pastanden motsies dermed direkte.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-core-patterns.md",
|
||||
"line": 423,
|
||||
"claim": "Semantic Kernel memory og embeddings er GA.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-core-patterns.md#19",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R4",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/ai-studio/how-to/flow-develop",
|
||||
"evidence_quote": "Prompt flow in Microsoft Foundry and Azure Machine Learning will be retired on April 20, 2027. Prompt flow is no longer recommended for new development.",
|
||||
"reason": "Gjeldende status er pensjonering og ikke anbefalt for ny utvikling, og siden gjelder kun Foundry (classic) — ikke den nye Foundry-portalen; en uforbeholden GA-status beskriver en tilstand som er avløst.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-core-patterns.md",
|
||||
"line": 424,
|
||||
"claim": "Prompt flow for RAG i Microsoft Foundry er GA.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-core-patterns.md#20",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/agent-framework/overview/",
|
||||
"evidence_quote": "",
|
||||
"reason": "Ingen side oppgir en samlet release-status for Microsoft Agent Framework: preview er avgrenset til Go (The Agent Framework for Go is in public preview) og til enkelte prerelease-pakker, mens rammeverket ellers anbefales som støttet migreringsmal for prompt flow og Foundry workflows.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-core-patterns.md",
|
||||
"line": 430,
|
||||
"claim": "Microsoft Agent Framework har preview-status.",
|
||||
"disposition": "unsourced"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-cost-optimization.md",
|
||||
"batch": "R7.3",
|
||||
"judged_at": "2026-07-25",
|
||||
"per_file_verdict": "flagged",
|
||||
"claim_count": 18,
|
||||
"verified": null,
|
||||
"verified_by": null,
|
||||
"flags": [
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-cost-optimization.md#8",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/foundry/openai/how-to/batch",
|
||||
"evidence_quote": "The service aims to process batch requests within 24 hours, but it doesn't expire jobs that take longer.",
|
||||
"reason": "Separat quota og ingen forstyrrelse av online workloads stemmer, men 24 timer er et target turnaround som siden eksplisitt ikke garanterer (jobber som tar lengre tid utløper ikke) - claimens SLA-påstand er en bærende del som motsies.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-cost-optimization.md",
|
||||
"line": 68,
|
||||
"claim": "Azure OpenAI Batch API har 24-timers SLA og egen separat quota, uten påvirkning på online workloads.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-cost-optimization.md#9",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/cognitive-search-skill-azure-openai-embedding",
|
||||
"evidence_quote": "text-embedding-ada-002 | 1536 | 1536 | text-embedding-3-large | 1 | 3072 | text-embedding-3-small | 1 | 1536",
|
||||
"reason": "ada-002 1536 stemmer, men Learn oppgir minimum 1 dimensjon for både 3-small (1-1536) og 3-large (1-3072), ikke 512 og 1024 som claimen angir, og ingen Learn-side oppgir Multilingual-e5-large med 1024 dimensjoner.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-cost-optimization.md",
|
||||
"line": 76,
|
||||
"claim": "Embedding-modeller og antall dimensjoner: text-embedding-ada-002 1536 | text-embedding-3-small 512-1536 | text-embedding-3-large 1024-3072 | Multilingual-e5-large 1024.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-cost-optimization.md#10",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/foundry/openai/concepts/model-retirement-schedule",
|
||||
"evidence_quote": "text-embedding-ada-002 | 2 | GA | 2028-02-09 | —",
|
||||
"reason": "Livssyklustabellen bruker egne merkelapper Deprecated, Legacy, Retired og GA; ada-002 står som GA (både versjon 1 og 2) med pensjonsdato 2028-02-09, altså verken legacy eller deprecated.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-cost-optimization.md",
|
||||
"line": 76,
|
||||
"claim": "text-embedding-ada-002 er legacy og deprecated.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-cost-optimization.md#18",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/search-limits-quotas-capacity",
|
||||
"evidence_quote": "Maximum search units (SU) | N/A | 3 SU | 36 SU | 36 SU | 36 SU | 36 SU | 36 SU | 36 SU | N/A",
|
||||
"reason": "S1-L2 med 36 SU stemmer, men Free står med N/A søkeenheter (ikke 1 SU) og har maksimalt 3 indekser (ikke 1) i indekstabellen; i tillegg sier fotnoten at Basic gir tre partisjoner og tre replikaer, totalt ni SU, for tjenester opprettet etter 3. april 2024 - flere bærende deler er feil.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-cost-optimization.md",
|
||||
"line": 365,
|
||||
"claim": "Search Units (SU) per tier i Azure AI Search: Free 1 SU med grense på 1 indeks | Basic 1-3 SU | S1, S2, S3, S3 HD, L1 og L2 1-36 SU.",
|
||||
"disposition": "outdated"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-document-preprocessing.md",
|
||||
"batch": "R7.1",
|
||||
|
|
@ -5292,6 +5636,160 @@
|
|||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-enterprise-scale.md",
|
||||
"batch": "R7.3",
|
||||
"judged_at": "2026-07-25",
|
||||
"per_file_verdict": "flagged",
|
||||
"claim_count": 20,
|
||||
"verified": null,
|
||||
"verified_by": null,
|
||||
"flags": [
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-enterprise-scale.md#5",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/dotnet/api/azure.search.documents.models.indexdocumentsbatch",
|
||||
"evidence_quote": "Package: Azure.Search.Documents v12.0.0 … Package: Azure.Search.Documents v12.1.0-beta.1",
|
||||
"reason": "Den siterte API-siden oppgir Azure.Search.Documents v12.0.0 og v12.1.0-beta.1; claimets v11.7.0 og v11.8.0-beta.1 er avløst av en ny major-versjon.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-enterprise-scale.md",
|
||||
"line": 42,
|
||||
"claim": "Azure.Search.Documents SDK er på versjon v11.7.0, med v11.8.0-beta.1 tilgjengelig som beta.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-enterprise-scale.md#7",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/search-how-to-index-sql-database",
|
||||
"evidence_quote": "For Azure SQL indexers, there are two change detection policies: … \"SqlIntegratedChangeTrackingPolicy\" (applies to tables only) … \"HighWaterMarkChangeDetectionPolicy\" (works for views)",
|
||||
"reason": "Den kanoniske siden som ville listet policyene oppgir nøyaktig to; claimet lister tre der «Integrated Change Tracking» og «SQL Change Tracking» er samme SQL-integrerte policy — den tredje oppføringen finnes ikke.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-enterprise-scale.md",
|
||||
"line": 50,
|
||||
"claim": "Change detection på en Azure AI Search data source støtter High Water Mark | Integrated Change Tracking | SQL Change Tracking.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-enterprise-scale.md#8",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/search-howto-schedule-indexers",
|
||||
"evidence_quote": "(required) The amount of time between the start of two consecutive indexer executions. The smallest interval allowed is 5 minutes, and the longest is 1,440 minutes (24 hours).",
|
||||
"reason": "5 minutter, 15 minutter, time og dag er gyldige, men den bærende delen «hvert 2. minutt» ligger under minimumsintervallet på 5 minutter og kan ikke settes.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-enterprise-scale.md",
|
||||
"line": 51,
|
||||
"claim": "Indexer-schedule i Azure AI Search kan settes til intervaller som hvert 2., 5. og 15. minutt, hver time og hver dag.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-enterprise-scale.md#14",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R4",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/search-limits-quotas-capacity",
|
||||
"evidence_quote": "| Before April 3, 2024 | 2 | 25 | 100 | 200 | 1,024 | 2,048 | N/A | … | After May 17, 2024 ^2^ | 15 | 160 | 512 | 1,024 | 2,048 | 4,096 | N/A |",
|
||||
"reason": "Kun Basic 15 GB / eldre 2 GB treffer gjeldende rad; S1 25, S2 100, S3 200 og L1 1 TB er verdiene fra den tidsstemplede «Before April 3, 2024»-raden — gjeldende verdier er 160, 512, 1 024 og 2 048 GB.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-enterprise-scale.md",
|
||||
"line": 161,
|
||||
"claim": "Storage per partition per tier i Azure AI Search: Basic 15 GB (services opprettet etter april 2024; eldre services 2 GB) | Standard S1 25 GB | Standard S2 100 GB | Standard S3 200 GB | Storage Optimized L1 1 TB.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-enterprise-scale.md#18",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/search/monitor-azure-cognitive-search-data-reference",
|
||||
"evidence_quote": "| **Indexer Processed Files (bytes)** … `DocumentsProcessedBytes` | … | **Document processed count** … `DocumentsProcessedCount` | … | **Storage usage** … `IndexStorageUsage` | … | **Vector Storage usage** … `IndexVectorUsage` | … | **Compute units used** … `PerRequestComputeConsumption` | … | **Search Latency** … `SearchLatency` | … | **Search queries per second** … `SearchQueriesPerSecond` | … | **Skill execution invocation count** … `SkillExecutionCount` | … | **Throttled search queries percentage** … `ThrottledSearchQueriesPercentage` |",
|
||||
"reason": "Den kanoniske enumerasjonen av metrikker for Microsoft.Search/searchServices inneholder ingen «Indexing Failed Documents» (feilede dokumenter er dimensjonen `Failed` på `DocumentsProcessedCount`), metrikken heter «Throttled search queries percentage» og ikke «Throttled Requests», og Search Latency har ingen p95-aggregering (kun Average/Count/Max/Min/Total).",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-enterprise-scale.md",
|
||||
"line": 234,
|
||||
"claim": "Azure Monitor eksponerer metrikkene Queries per Second (QPS) | Indexing Failed Documents | Search Latency (p95) | Throttled Requests for Azure AI Search.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-enterprise-scale.md#19",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/search-region-support",
|
||||
"evidence_quote": "You can create an Azure AI Search service in any of the following Azure public regions. … Europe … | Norway East | ✅ | ✅ | | ✅ | | |",
|
||||
"reason": "Jeg sjekket den kanoniske regionslisten: Europa-tabellen har rad for Norway East, men ingen rad for Norway West — Stavanger-regionen mangler helt, så påstanden om to norske regioner er feil.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-enterprise-scale.md",
|
||||
"line": 266,
|
||||
"claim": "Azure AI Search er tilgjengelig i de norske regionene Norway East (Oslo) | Norway West (Stavanger).",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-enterprise-scale.md#20",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/search-capacity-planning",
|
||||
"evidence_quote": "",
|
||||
"reason": "Ingen Learn-side oppgir QPS-estimat per tier; den siterte kapasitetssiden sier tvert imot at slike retningslinjer ikke finnes («There are no guidelines on how many replicas are needed to accommodate query loads») og henviser til egen benchmarking, så tallene ~15/~15/~60/~120 kan verken bekreftes eller motbevises.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-enterprise-scale.md",
|
||||
"line": 277,
|
||||
"claim": "QPS-estimat per tier i Azure AI Search: Basic ~15 | Standard S1 ~15 | Standard S2 ~60 | Standard S3 ~120.",
|
||||
"disposition": "unsourced"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-evaluation-frameworks.md",
|
||||
"batch": "R7.3",
|
||||
"judged_at": "2026-07-25",
|
||||
"per_file_verdict": "flagged",
|
||||
"claim_count": 17,
|
||||
"verified": null,
|
||||
"verified_by": null,
|
||||
"flags": [
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-evaluation-frameworks.md#1",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/foundry-classic/how-to/develop/evaluate-sdk",
|
||||
"evidence_quote": "Evaluate your generative AI application locally with the Azure AI Evaluation SDK (preview) (classic)",
|
||||
"reason": "Den siterte sidens egen tittel merker Azure AI Evaluation SDK som (preview), ikke GA, og delpåstanden om at de agentiske evaluatorene er i Preview holder heller ikke (flere er ikke preview-merket) - kun Groundedness Pro (preview) stemmer, så minst to bærende deler er motsagt.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-evaluation-frameworks.md",
|
||||
"line": 4,
|
||||
"claim": "Azure AI Evaluation SDK har status GA, mens agentiske evaluatorer og Groundedness Pro er i Preview.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-evaluation-frameworks.md#7",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/foundry/concepts/built-in-evaluators",
|
||||
"evidence_quote": "Tool Call Accuracy | Measures the overall quality of tool calls including selection, parameter correctness, and efficiency.",
|
||||
"reason": "Den kanoniske opplistingen av agent-evaluatorer merker bare noen med (preview) - Task Adherence, Task Completion, Customer Satisfaction, Intent Resolution, Quality Grader - mens Tool Call Accuracy, Tool Selection, Tool Input Accuracy, Tool Output Utilization, Tool Call Success og Task Navigation Efficiency står uten preview-merke; den generelle påstanden om at alle agentiske evaluatorer er i Preview er derfor motsagt.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-evaluation-frameworks.md",
|
||||
"line": 74,
|
||||
"claim": "De agentiske evaluatorene i Azure AI Evaluation er i Preview.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-evaluation-frameworks.md#15",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/foundry/concepts/evaluation-regions-limits-virtual-network",
|
||||
"evidence_quote": "",
|
||||
"reason": "Verken den siterte observability-siden, den kanoniske region-/rate-limit-siden eller Foundry-personvernsiden oppgir noen EU-hostet-mot-US-hostet fordeling av LLM-judge-modeller etter arbeidsområdets region; region-sidene lister kun støttede regioner (både US og EU) og sier ingenting som verken bekrefter eller ordrett motsier påstanden.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-evaluation-frameworks.md",
|
||||
"line": 268,
|
||||
"claim": "LLM-judges i evaluering bruker EU-hostede modeller for EU/EØS-arbeidsområder, og US-hostede modeller for andre regioner.",
|
||||
"disposition": "unsourced"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-evaluation-frameworks.md#17",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/foundry-classic/how-to/develop/evaluate-sdk",
|
||||
"evidence_quote": "",
|
||||
"reason": "Den siterte siden sier kun at prompt-ene til kvalitetsevaluatorene er open-sourcet, ikke at SDK-en er gratis, og ingen learn.microsoft.com-side oppgir prising for SDK-en eller at kostnaden utelukkende kommer fra Azure OpenAI-token-bruk for LLM-judge-kall - prisdetaljene ligger bak Azure-prissiden.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-evaluation-frameworks.md",
|
||||
"line": 290,
|
||||
"claim": "Azure AI Evaluation SDK er gratis (open source); kostnaden kommer fra Azure OpenAI-token-bruk for LLM-judge-kall.",
|
||||
"disposition": "unsourced"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-hallucination-mitigation.md",
|
||||
"batch": "R7.1",
|
||||
|
|
@ -5339,6 +5837,266 @@
|
|||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-iterative-refinement.md",
|
||||
"batch": "R7.3",
|
||||
"judged_at": "2026-07-25",
|
||||
"per_file_verdict": "flagged",
|
||||
"claim_count": 18,
|
||||
"verified": null,
|
||||
"verified_by": null,
|
||||
"flags": [
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-iterative-refinement.md#2",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/agent-framework/tutorials/agents/multi-turn-conversation",
|
||||
"evidence_quote": "var serialized = agent.SerializeSession(session);",
|
||||
"reason": "WhiteboardProvider-delen er korrekt, men Agent Framework serialiserer sesjonen via agent.SerializeSession(session) / AIAgent.SerializeSessionAsync(session) - det finnes ingen AgentSession.Serialize(); én bærende del av påstanden er dermed feil.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-iterative-refinement.md",
|
||||
"line": 41,
|
||||
"claim": "Session state i Microsoft-stakken dekkes av `AgentSession.Serialize()` (Agent Framework) og `WhiteboardProvider` (Semantic Kernel).",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-iterative-refinement.md#11",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/agent-framework/tutorials/agents/multi-turn-conversation",
|
||||
"evidence_quote": "var serialized = agent.SerializeSession(session);",
|
||||
"reason": "CreateSessionAsync(), RunAsync(prompt, session) og DeserializeSessionAsync(serialized) stemmer, men serialiseringen skjer via agent.SerializeSession(session) / SerializeSessionAsync på AIAgent - ikke session.Serialize(); den bærende API-strengen er feil selv om returtypen JsonElement er riktig.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-iterative-refinement.md",
|
||||
"line": 235,
|
||||
"claim": "Agent Framework tilbyr følgende API for multi-turn-samtaler: `agent.CreateSessionAsync()` | `agent.RunAsync(prompt, session)` | `session.Serialize()` som returnerer `JsonElement` | `agent.DeserializeSessionAsync(serializedSession)`.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-iterative-refinement.md#16",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R4",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/foundry/openai/concepts/retired-models",
|
||||
"evidence_quote": "The following models are retired and no longer available for use or for new deployments. | gpt-35-turbo - 0613 | | February 13, 2025 | gpt-35-turbo (0125) gpt-4o-mini | | gpt-4 gpt-4-32k - 0613 | | June 6, 2025 | gpt-4o version: 2024-11-20 |",
|
||||
"reason": "Både GPT-3.5- og GPT-4-modellene står i pensjonert-listen med gpt-4o-mini og gpt-4o som anbefalt erstatning, og gjeldende retirement schedule inneholder verken gpt-35-turbo eller gpt-4 - anbefalingen beskriver en modellverden som ikke lenger finnes.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-iterative-refinement.md",
|
||||
"line": 380,
|
||||
"claim": "Anbefalt modellvalg for kostnadsoptimalisering: GPT-3.5 for oppsummeringer og GPT-4 for svar-generering.",
|
||||
"disposition": "outdated"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-query-understanding.md",
|
||||
"batch": "R7.3",
|
||||
"judged_at": "2026-07-25",
|
||||
"per_file_verdict": "flagged",
|
||||
"claim_count": 18,
|
||||
"verified": null,
|
||||
"verified_by": null,
|
||||
"flags": [
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-query-understanding.md#1",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/azure-arc/edge-rag/search-types",
|
||||
"evidence_quote": "Agentic Retrieval in Foundry Local is currently in PREVIEW.",
|
||||
"reason": "Filen stempler query understanding-kapabilitetene samlet som GA, men minst én løftebærende del svikter: søketype-siden som filen selv siterer er eksplisitt i PREVIEW, og AI Builder «Create text with GPT» er også merket preview.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-query-understanding.md",
|
||||
"line": 4,
|
||||
"claim": "Query understanding-kapabilitetene beskrevet i filen er merket med status GA (generelt tilgjengelig).",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-query-understanding.md#2",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/search-synonyms",
|
||||
"evidence_quote": "If the synonym map exists on the search service, it's used on the next query, with no reindexing or rebuild required.",
|
||||
"reason": "Synonym maps som query expansion er grunnet, men den løftebærende delen «ved indexing-tid» motsies — siden sier ekspansjonen skjer ved query-tid uten reindeksering («Internally, the synonyms feature rewrites the original query with synonyms by using the OR operator»).",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-query-understanding.md",
|
||||
"line": 55,
|
||||
"claim": "Azure AI Search støtter synonym maps der man f.eks. definerer «AI, kunstig intelligens, maskinlæring» for automatisk query expansion ved indexing-tid.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-query-understanding.md#7",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R4",
|
||||
"evidence_url": "https://learn.microsoft.com/ai-builder/azure-openai-model-pauto",
|
||||
"evidence_quote": "This feature is deprecated and isn't be visible anymore.",
|
||||
"reason": "Cloud-flow-handlingen «Create text with GPT» i Power Automate/AI Builder er utgått og erstattet av «Create text with GPT using a prompt» (siden mai 2025 kalt «Run a prompt»); kun en gated preview-handling i Power Automate for desktop bærer fortsatt det gamle navnet, så claimet treffer en avviklet rad.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-query-understanding.md",
|
||||
"line": 337,
|
||||
"claim": "Power Automate med AI Builder har handlingen «Create text with GPT» som kan brukes til å klassifisere intent før conditional branching.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-query-understanding.md#8",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/ai-foundry/openai/concepts/models",
|
||||
"evidence_quote": "",
|
||||
"reason": "Den kanoniske modellsiden omtaler text-embedding-3-large sin flerspråklighet kun via MIRACL-benchmarken (54,9) og oppgir ingen språktelling; ingen learn.microsoft.com-side sier «over 100 språk» for modellen.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-query-understanding.md",
|
||||
"line": 351,
|
||||
"claim": "Embedding-modellen text-embedding-3-large støtter over 100 språk, slik at spørring og dokumenter på ulike språk matcher i samme vector space.",
|
||||
"disposition": "unsourced"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-query-understanding.md#10",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/search-limits-quotas-capacity",
|
||||
"evidence_quote": "The following table describes the semantic ranker throttling limits by tier, subject to available capacity in the region.",
|
||||
"reason": "Første del holder (synonym-tabellen gir Basic 3 synonym maps), men den løftebærende delen «semantic ranker krever Standard-tier eller høyere» motsies: semantic ranker-tabellens tier-kolonner starter på Basic (2 samtidige forespørsler per search unit), og oversiktssiden sier den kan brukes gratis innenfor free-tier-grenser.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-query-understanding.md",
|
||||
"line": 399,
|
||||
"claim": "Azure AI Search Basic tier støtter synonym maps, mens semantic ranker krever Standard-tier eller høyere.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-query-understanding.md#11",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/semantic-kernel/overview/",
|
||||
"evidence_quote": "",
|
||||
"reason": "Learn beskriver Semantic Kernel som «a lightweight, open-source development kit», men ingen learn.microsoft.com-side oppgir MIT-lisens eller «ingen lisenskostnad» — lisensteksten finnes kun i GitHub-repoet, som er utenfor Learn.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-query-understanding.md",
|
||||
"line": 400,
|
||||
"claim": "Semantic Kernel er open source under MIT-lisens og har ingen lisenskostnad.",
|
||||
"disposition": "unsourced"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-query-understanding.md#12",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R3",
|
||||
"evidence_url": "https://learn.microsoft.com/microsoft-copilot-studio/billing-licensing",
|
||||
"evidence_quote": "Starting on September 1, 2025, the common currency for agents changed from *messages* to *Copilot Credits*.",
|
||||
"reason": "«Inkludert i M365 Copilot» og «standalone» er grunnet på samme side, men måleenheten i claimet er erstattet: sesjoner ble avløst av messages og deretter av Copilot Credits, som nå er «the common currency across Copilot Studio capabilities».",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-query-understanding.md",
|
||||
"line": 401,
|
||||
"claim": "Copilot Studio lisensieres per bruker/sesjon og er inkludert i M365 Copilot eller tilgjengelig standalone.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-query-understanding.md#14",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/azure-arc/edge-rag/search-types",
|
||||
"evidence_quote": "The following tables list and describe all parameters you can configure for each search type in Agentic Retrieval.",
|
||||
"reason": "Den kanoniske parameter-oppramsingen lister kun Temperature, Top-P, Top-N documents, Text strictness og Image strictness — verken query expansion, sub-query generation eller hypothetical answer generation forekommer; siden er dessuten omrammet fra Arc Edge RAG til Agentic Retrieval i Foundry Local.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-query-understanding.md",
|
||||
"line": 576,
|
||||
"claim": "Azure Arc Edge RAG tilbyr søketype-parametere for query expansion | sub-query generation | hypothetical answer generation.",
|
||||
"disposition": "outdated"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-security-rbac.md",
|
||||
"batch": "R7.3",
|
||||
"judged_at": "2026-07-25",
|
||||
"per_file_verdict": "flagged",
|
||||
"claim_count": 20,
|
||||
"verified": null,
|
||||
"verified_by": null,
|
||||
"flags": [
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-security-rbac.md#12",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/search-document-level-access-overview",
|
||||
"evidence_quote": "SharePoint site groups (preview, starting in the 2026-05-01-preview REST API). Requires extra index configuration.",
|
||||
"reason": "Claimen påstår at SharePoint-grupper ikke er støttet i preview, men siden lister dem nå som støttede prinsipaltyper fra 2026-05-01-preview; de tre andre gruppetypene stemmer, men denne løftebærende delen er motsagt.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-security-rbac.md",
|
||||
"line": 250,
|
||||
"claim": "Støttede gruppetyper for SharePoint ACL (preview): Microsoft Entra security groups (støttet) | Microsoft 365-grupper (støttet) | mail-enabled security groups (støttet) | SharePoint-grupper (ikke støttet i preview).",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-security-rbac.md#13",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/search/search-region-support",
|
||||
"evidence_quote": "You can create an Azure AI Search service in any of the following Azure public regions. ... France Central | Germany West Central | Italy North | Norway East | North Europe | Poland Central | Spain Central | Sweden Central | Switzerland North | Switzerland West | UK South | UK West | West Europe",
|
||||
"reason": "Den kanoniske regionlisten for Azure AI Search fører opp Norway East, men Norway West er fraværende i hele oppregningen - tjenesten tilbys ikke der, så den ene løftebærende halvdelen av regionpåstanden er feil.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-security-rbac.md",
|
||||
"line": 292,
|
||||
"claim": "Data residency for norsk offentlig sektor løses ved å kjøre Azure AI Search i regionene Norway East/Norway West.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-security-rbac.md#14",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/ai-foundry/openai/how-to/deployment-types",
|
||||
"evidence_quote": "",
|
||||
"reason": "Ingen Learn-side anbefaler Sweden Central for norsk offentlig sektor; deployment-types-siden beskriver bare generiske residency-valg (Global, DataZone, Standard/Regional) og opplyser at EU Data Zone også omfatter Norge, så anbefalingen er verken bekreftet eller motsagt.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-security-rbac.md",
|
||||
"line": 331,
|
||||
"claim": "Azure OpenAI anbefales kjørt i Sweden Central med data residency commitment for norsk offentlig sektor.",
|
||||
"disposition": "unsourced"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/rag-security-rbac.md#16",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/search-indexer-sharepoint-access-control-lists",
|
||||
"evidence_quote": "Prerequisites - Azure AI Search on a billable tier (Basic or higher) in any region. - SharePoint in Microsoft 365 sites, libraries, folders, and files with configured permissions. - Complete all configuration steps in the SharePoint indexer documentation ... - Configure Microsoft Entra application permissions and a credential appropriate for your scenario. ... - REST API version 2026-05-01-preview or an equivalent preview SDK package.",
|
||||
"reason": "Den kanoniske forutsetningslisten for SharePoint-ACL-tilnærmingen nevner ingen Microsoft 365 E3/E5-lisens, og SharePoint-tjenestebeskrivelsen viser at SharePoint inngår i langt flere planer (Business Basic/Standard/Premium, Office 365 E1/E3/E5, F1/F3 og standalone SharePoint Plan 1/2) - det påståtte E3/E5-kravet finnes ikke i kilden.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-security-rbac.md",
|
||||
"line": 346,
|
||||
"claim": "SharePoint ACL-tilnærmingen forutsetter SharePoint-lisensiering via Microsoft 365 E3/E5.",
|
||||
"disposition": "outdated"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/self-reflective-rag.md",
|
||||
"batch": "R7.3",
|
||||
"judged_at": "2026-07-25",
|
||||
"per_file_verdict": "flagged",
|
||||
"claim_count": 16,
|
||||
"verified": null,
|
||||
"verified_by": null,
|
||||
"flags": [
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/self-reflective-rag.md#1",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/foundry/concepts/evaluation-evaluators/rag-evaluators",
|
||||
"evidence_quote": "Groundedness Pro (preview) | System evaluation | You want a strict groundedness definition powered by Azure AI Content Safety and use our service model ... Response Completeness (preview) | System evaluation | You want to ensure the RAG response doesn't miss critical information (recall aspect) from your ground truth",
|
||||
"reason": "Siden merker to av seks RAG-evaluatorer eksplisitt som preview (Groundedness Pro og Response Completeness), så den bærende delen «evaluatorene er GA» er motsagt selv om agentic retrieval-delen holder.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/self-reflective-rag.md",
|
||||
"line": 4,
|
||||
"claim": "Microsoft Foundry-evaluatorene er GA, mens agentic retrieval er i Preview.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/self-reflective-rag.md#2",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/foundry/concepts/evaluation-evaluators/rag-evaluators",
|
||||
"evidence_quote": "Groundedness - Is the response grounded in the provided context without fabrication? Groundedness Pro - Does the response strictly adhere to the context (Azure AI Content Safety)? Relevance - Does the response accurately address the user's query? Response Completeness (preview) - Does the response cover all critical information from ground truth?",
|
||||
"reason": "Den kanoniske oppregningen av RAG-evaluatorer inneholder seks evaluatorer og coherence er ikke blant dem (Coherence er kategorisert som general purpose-evaluator), så påstanden om coherence som innebygd RAG-evaluator er feil.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/self-reflective-rag.md",
|
||||
"line": 29,
|
||||
"claim": "Microsoft Foundry tilbyr innebygde evaluatorer for RAG-kvalitetsvurdering: groundedness | relevance | coherence — alle med 1-5-skala.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/self-reflective-rag.md#3",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "R7",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/search/agentic-retrieval-overview",
|
||||
"evidence_quote": "",
|
||||
"reason": "Preview-statusen er dekket, men ingen learn.microsoft.com-side oppgir noen tallfestet relevansgevinst; oversiktssiden og transparensnotatet beskriver LLM-assistert query planning uten «40 %», og tallet er bærende (ikke illustrativt), så det kan verken bekreftes eller motbevises.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/self-reflective-rag.md",
|
||||
"line": 29,
|
||||
"claim": "Azure AI Search agentic retrieval er i preview og forbedrer retrieval-relevans med opptil 40 % gjennom LLM-assistert query planning.",
|
||||
"disposition": "unsourced"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-governance/references/monitoring-observability/alerting-strategies-escalation.md",
|
||||
"batch": "R7.1",
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue