feat(ms-ai-architect): R7.3 bølge 2 — payload-16..25 dømt+ingestet (190 claims, 10 flagged, 57 flagg; ledger 109→119) [skip-docs]
This commit is contained in:
parent
c4554708ca
commit
7011898bd0
1 changed files with 795 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": 109,
|
||||
"count": 119,
|
||||
"generated": "2026-07-18"
|
||||
},
|
||||
"files": [
|
||||
|
|
@ -4322,6 +4322,53 @@
|
|||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/agentic-rag-patterns.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/agentic-rag-patterns.md#1",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/semantic-kernel/frameworks/agent/agent-rag",
|
||||
"evidence_quote": "The Semantic Kernel Agent RAG functionality is experimental, subject to change, and will only be finalized based on feedback and evaluation.",
|
||||
"reason": "Claimen sier GA, men den siterte siden merker Semantic Kernel Agent RAG som eksperimentell og endringsutsatt - motsatt status.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/agentic-rag-patterns.md",
|
||||
"line": 6,
|
||||
"claim": "Semantic Kernel-baserte agentic RAG-funksjoner har status GA (generelt tilgjengelig).",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/agentic-rag-patterns.md#16",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/semantic-kernel/frameworks/agent/agent-orchestration/",
|
||||
"evidence_quote": "| Magentic | Group chat-like orchestration inspired by MagenticOne. | Complex, generalist multi-agent collaboration. |",
|
||||
"reason": "Siden lister fem støttede orkestreringsmønstre - Concurrent, Sequential, Handoff, Group Chat og Magentic - så claimens uttalte antall fire er en bærende del som motsies av kilden.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/agentic-rag-patterns.md",
|
||||
"line": 176,
|
||||
"claim": "Semantic Kernel har fire agent orchestration patterns: Sequential | Concurrent | Handoff | Group Chat.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/agentic-rag-patterns.md#21",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/architecture/ai-ml/guide/ai-agent-design-patterns",
|
||||
"evidence_quote": "| Magentic | Plan-build-execute model. Manager agent builds and adapts a task ledger. | Manager agent assigns and reorders tasks dynamically. | Open-ended problems that don't have a predetermined solution path. | Slow to converge. Stalls on ambiguous goals. |",
|
||||
"reason": "Jeg hentet den kanoniske enumererende siden (Choose a pattern-tabellen): mønstrene er Sequential, Concurrent (også kalt parallel/fan-out), Group chat, Handoff og Magentic - verken supervisor eller autonomous loop finnes der, og claimen mangler group chat, handoff og magentic.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/agentic-rag-patterns.md",
|
||||
"line": 326,
|
||||
"claim": "AI Agent Design Patterns (Azure Architecture Center) omfatter mønstrene: sequential (pipeline) | parallel fanout | supervisor | autonomous loop.",
|
||||
"disposition": "outdated"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/azure-ai-search-setup.md",
|
||||
"batch": "R7.1",
|
||||
|
|
@ -4357,6 +4404,65 @@
|
|||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/chunking-strategies.md",
|
||||
"batch": "R7.3",
|
||||
"judged_at": "2026-07-25",
|
||||
"per_file_verdict": "flagged",
|
||||
"claim_count": 27,
|
||||
"verified": null,
|
||||
"verified_by": null,
|
||||
"flags": [
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/chunking-strategies.md#6",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/cognitive-search-skill-textsplit",
|
||||
"evidence_quote": "The minimum value is 300, the maximum is 50000, and the default value is 5000.",
|
||||
"reason": "Default for maximumPageLength er 5000 tegn (chunk-artikkelen sier også uttrykkelig 5,000 characters (the default)), ikke 2000; ekvivalensen 2000 tegn ca. 512 tokens er riktig som anbefaling, men den løftebærende default-verdien er feil.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/chunking-strategies.md",
|
||||
"line": 49,
|
||||
"claim": "Text Split skill-parameteren `maximumPageLength` har default 2000 tegn, som tilsvarer omtrent 512 tokens.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/chunking-strategies.md#7",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/cognitive-search-skill-textsplit",
|
||||
"evidence_quote": "",
|
||||
"reason": "Parametertabellen oppgir defaults for maximumPageLength, maximumPagesToTake, defaultLanguageCode og unit, men ingen default for pageOverlapLength - kun at verdien 0 gir ingen overlapp; heller ikke SDK-referansene oppgir en default, og ingen Learn-side motsier 0.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/chunking-strategies.md",
|
||||
"line": 50,
|
||||
"claim": "Text Split skill-parameteren `pageOverlapLength` har default 0, altså ingen overlapp mellom chunks.",
|
||||
"disposition": "unsourced"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/chunking-strategies.md#22",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/cognitive-search-skill-document-intelligence-layout",
|
||||
"evidence_quote": "",
|
||||
"reason": "Learn oppgir ingen regionliste for Document Intelligence - både skill-siden og tjenesteoversikten viser til azure.com Product availability by region; verken West Europe-tilgjengeligheten eller fraværet i Norge kan bekreftes eller motbevises på Learn.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/chunking-strategies.md",
|
||||
"line": 253,
|
||||
"claim": "Document Intelligence er tilgjengelig i West Europe, altså i EU, men ikke i Norge spesifikt.",
|
||||
"disposition": "unsourced"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/chunking-strategies.md#24",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/ai-foundry/openai/concepts/models",
|
||||
"evidence_quote": "",
|
||||
"reason": "Learn oppgir bare at tredjegenerasjons embedding-modeller gir bedre flerspråklig gjenfinning målt med MIRACL; ingen Learn-side lister støttede språk for text-embedding-3, så norsk kan verken bekreftes eller motbevises.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/chunking-strategies.md",
|
||||
"line": 271,
|
||||
"claim": "Embedding-modellene i text-embedding-3-serien støtter norsk.",
|
||||
"disposition": "unsourced"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/citation-tracking.md",
|
||||
"batch": "R7.1",
|
||||
|
|
@ -4392,6 +4498,694 @@
|
|||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/contextual-retrieval.md",
|
||||
"batch": "R7.3",
|
||||
"judged_at": "2026-07-25",
|
||||
"per_file_verdict": "flagged",
|
||||
"claim_count": 15,
|
||||
"verified": null,
|
||||
"verified_by": null,
|
||||
"flags": [
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/contextual-retrieval.md#1",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/search/agentic-retrieval-overview",
|
||||
"evidence_quote": "Some agentic retrieval features are generally available in the 2026-04-01 REST API via programmatic access.",
|
||||
"reason": "Påstandens andre bærende del — «Preview for agentic retrieval» — er superseded: knowledge bases og utvalgte knowledge sources er nå generelt tilgjengelige i 2026-04-01 REST API, så agentic retrieval er ikke lenger ren Preview.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/contextual-retrieval.md",
|
||||
"line": 4,
|
||||
"claim": "Contextual Retrieval har status GA for custom skill-mønsteret og Preview for agentic retrieval.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/contextual-retrieval.md#2",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/search/cognitive-search-custom-skill-interface",
|
||||
"evidence_quote": "",
|
||||
"reason": "Hver enkeltkomponent finnes på Learn (Custom Web API skill, Azure OpenAI GPT-4o, Text Merge skill, Azure OpenAI Embedding skill), men ingen learn.microsoft.com-side oppgir denne fire-komponents-dekomponeringen av kontekstuell prefiks-generering — den er forfatterens egen arkitektursammensetning.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/contextual-retrieval.md",
|
||||
"line": 39,
|
||||
"claim": "Kontekstuell prefiks-generering består av fire komponenter: Custom Web API Skill (Azure Function som mottar chunk + metadata og returnerer kontekstuell prefiks) | LLM (GPT-4o) | Text Merge Skill | Embedding Skill.",
|
||||
"disposition": "unsourced"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/contextual-retrieval.md#3",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/search/cognitive-search-create-custom-skill-example",
|
||||
"evidence_quote": "",
|
||||
"reason": "Microsofts custom skill-eksempel er et C#-eksempel som wrapper Bing Entity Search API og inneholder ingen Azure OpenAI-klient; verken den siden eller Azure OpenAI API-livssyklussiden oppgir api_version «2024-10-01-preview» for et custom skill-eksempel, og versjonsstrengen er heller ikke motbevist.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/contextual-retrieval.md",
|
||||
"line": 88,
|
||||
"claim": "Azure OpenAI-klienten i custom skill-eksempelet bruker api_version «2024-10-01-preview».",
|
||||
"disposition": "unsourced"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/contextual-retrieval.md#6",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R4",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/azure-functions/functions-overview",
|
||||
"evidence_quote": "| [Consumption plan](consumption-plan) | Legacy serverless plan (Windows only). Use the Flex Consumption plan for new apps. |",
|
||||
"reason": "Påstanden fester hostingen til Azure Functions consumption plan med Python/C#, men Learn merker nå Consumption som en legacy-plan som kun er Windows — der Python ikke støttes — og henviser nye apper til Flex Consumption; SKU-en påstanden oppgir er dermed en utgått rad.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/contextual-retrieval.md",
|
||||
"line": 201,
|
||||
"claim": "Custom skill hostes på Azure Functions consumption plan (Python/C#).",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/contextual-retrieval.md#7",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/search/search-how-to-semantic-chunking",
|
||||
"evidence_quote": "",
|
||||
"reason": "Alle fem skills finnes og rollene stemmer, men Microsofts dokumenterte pipeline er Document Layout → Text Split → Azure OpenAI Embedding; ingen Learn-side oppgir denne fem-trinns-pipelinen med custom kontekst-genereringsskill og Text Merge, og ingen motbeviser den heller.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/contextual-retrieval.md",
|
||||
"line": 210,
|
||||
"claim": "Den fullstendige skillset-pipelinen består av fem trinn: Document Layout skill (struktur til Markdown) | Text Split skill | Custom Context Generation skill (GPT-4o) | Text Merge skill | Azure OpenAI Embedding skill (text-embedding-3-large).",
|
||||
"disposition": "unsourced"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/contextual-retrieval.md#10",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure-region-availability",
|
||||
"evidence_quote": "| gpt-4o | 2024-11-20 | ✅ | - | ✅ | - | - | ✅ | ✅ | ✅ | - |",
|
||||
"reason": "I Standard/Regional-tabellen for Europa er norwayeast merket ✅ for gpt-4o 2024-11-20, så den bærende delen «Sweden Central som nærmeste region med GPT-4o» er motbevist — Norway East tilbyr selv GPT-4o, selv om EU/EØS-delen holder.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/contextual-retrieval.md",
|
||||
"line": 227,
|
||||
"claim": "Azure OpenAI må kjøres i Sweden Central som nærmeste region med GPT-4o; data forblir i EU/EØS.",
|
||||
"disposition": "outdated"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/embedding-models-selection.md",
|
||||
"batch": "R7.3",
|
||||
"judged_at": "2026-07-25",
|
||||
"per_file_verdict": "flagged",
|
||||
"claim_count": 25,
|
||||
"verified": null,
|
||||
"verified_by": null,
|
||||
"flags": [
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/embedding-models-selection.md#1",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/search/vector-search-how-to-generate-embeddings",
|
||||
"evidence_quote": "One step involves selecting an embedding model to vectorize your plain text content. The following models are supported: 1. text-embedding-3-small 2. text-embedding-3-large 3. text-embedding-ada-002 4. Cohere-embed-v3-english 5. Cohere-embed-v3-multilingual",
|
||||
"reason": "GA-delene for Azure OpenAI og Azure AI Search holder, men den kanoniske listen over embedding-modeller Azure tilbyr inneholder verken Multilingual E5 eller Custom embeddings, så Preview-statusen for disse to bærende delene er ikke dekket.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/embedding-models-selection.md",
|
||||
"line": 4,
|
||||
"claim": "Filen angir status GA for Azure OpenAI og Azure AI Search, og Preview for Multilingual E5 og Custom embeddings.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/embedding-models-selection.md#7",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/postgresql/extensions/azure-local-ai",
|
||||
"evidence_quote": "",
|
||||
"reason": "Ingen learn.microsoft.com-side oppgir dimensjonstall eller token-tak for multilingual-e5-small; den eneste Azure-omtalen (azure_local_ai for PostgreSQL) nevner modellen uten spesifikasjoner og er dessuten pensjonert.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/embedding-models-selection.md",
|
||||
"line": 43,
|
||||
"claim": "multilingual-e5-small har 384 dimensjoner og maks 512 tokens.",
|
||||
"disposition": "unsourced"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/embedding-models-selection.md#8",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/search/vector-search-how-to-generate-embeddings",
|
||||
"evidence_quote": "The following models are supported: 1. text-embedding-3-small 2. text-embedding-3-large 3. text-embedding-ada-002 4. Cohere-embed-v3-english 5. Cohere-embed-v3-multilingual",
|
||||
"reason": "Jeg hentet de kanoniske listene over embedding-modeller Azure tilbyr (Azure AI Search/Foundry-veiviseren og Foundry Models sold by Azure) — multilingual-e5-small og multilingual-e5-large er fraværende i begge, og fraværet er bevis mot at de tilbys via Azure AI med Preview-status.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/embedding-models-selection.md",
|
||||
"line": 43,
|
||||
"claim": "multilingual-e5-small og multilingual-e5-large har status Preview og tilbys via Azure AI.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/embedding-models-selection.md#9",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/postgresql/extensions/azure-local-ai",
|
||||
"evidence_quote": "",
|
||||
"reason": "Ingen learn.microsoft.com-side oppgir 1024 dimensjoner eller 512 tokens for multilingual-e5-large; modellen omtales ikke i Microsofts dokumentasjon i det hele tatt.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/embedding-models-selection.md",
|
||||
"line": 44,
|
||||
"claim": "multilingual-e5-large har 1024 dimensjoner og maks 512 tokens.",
|
||||
"disposition": "unsourced"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/embedding-models-selection.md#10",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure",
|
||||
"evidence_quote": "The following models are supported for fine-tuning: gpt-4o-mini (2024-07-18), gpt-4o (2024-08-06), gpt-4.1 (2025-04-14), gpt-4.1-mini (2025-04-14), gpt-4.1-nano (2025-04-14), o4-mini (2025-04-16), gpt-5 (2025-08-07), Ministral-3B (2411), Qwen-32B, Llama-3.3-70B-Instruct, gpt-oss-20b",
|
||||
"reason": "Jeg sjekket den kanoniske listen over modeller som kan finjusteres — den inneholder ingen embedding-modell, så Custom embeddings for domenespesifikk fine-tuning finnes ikke som et Azure-tilbud med status Announced.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/embedding-models-selection.md",
|
||||
"line": 45,
|
||||
"claim": "Custom embeddings for domene-spesifikk fine-tuning har status Announced, med variabel dimensjonalitet og variabel maks token-kapasitet.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/embedding-models-selection.md#11",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/ai-foundry/openai/concepts/models",
|
||||
"evidence_quote": "",
|
||||
"reason": "Learn oppgir ingen språkantall for OpenAI-embeddings (kun MIRACL- og MTEB-benchmarktall) og sier ingenting om at E5-modellene er optimalisert for flerspråklig kvalitet; verken 100+ språk eller E5-påstanden kan bekreftes eller avkreftes.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/embedding-models-selection.md",
|
||||
"line": 50,
|
||||
"claim": "OpenAI-embeddingmodellene håndterer 100+ språk, men med fallende kvalitet utenfor engelsk, mens E5-modellene er optimalisert for flerspråklig kvalitet.",
|
||||
"disposition": "unsourced"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/embedding-models-selection.md#13",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R4",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/ai-foundry/openai/api-version-lifecycle",
|
||||
"evidence_quote": "Azure OpenAI API version 2024-10-21 is currently the latest GA API release. This API version is the replacement for the previous 2024-06-01 GA API release.",
|
||||
"reason": "2024-02-01 er en tidsstemplet, avløst rad: Learn oppgir at 2024-06-01 erstattet 2024-02-01, at 2024-10-21 nå er siste GA, og at v1-API-et anbefales — eksempelets api_version peker på en versjon som siden er skiftet ut.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/embedding-models-selection.md",
|
||||
"line": 60,
|
||||
"claim": "Azure OpenAI embeddings-kall i eksempelet bruker api_version 2024-02-01.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/embedding-models-selection.md#15",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/microsoft-copilot-studio/knowledge-azure-ai-search",
|
||||
"evidence_quote": "Copilot Studio supports vectorized indexes using integrated vectorization. Prepare your data and choose an embedded model, then use Import and vectorize data in Azure AI Search to create vector indexes.",
|
||||
"reason": "Koblingen til Azure AI Search som knowledge source stemmer, men Learn krever at du selv velger embedding-modell og vektoriserer indeksen i Azure AI Search før tilkobling — embeddings genereres altså ikke automatisk ved opplasting, og modellvalget er ikke låst slik claimet påstår.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/embedding-models-selection.md",
|
||||
"line": 247,
|
||||
"claim": "Copilot Studio kan kobles til Azure AI Search som knowledge source; embeddings genereres automatisk ved opplasting, og standardmodellen (typisk ada-002 eller text-embedding-3-small) kan ikke endres direkte i UI.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/embedding-models-selection.md#17",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/search/vector-search-overview",
|
||||
"evidence_quote": "",
|
||||
"reason": "Learn omtaler verken self-hosting av Custom embeddings i norske Azure-regioner eller den norske klassifiseringen Høy/Kritisk; anbefalingen kan verken bekreftes eller avkreftes mot Microsoft-dokumentasjon.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/embedding-models-selection.md",
|
||||
"line": 280,
|
||||
"claim": "Custom embeddings kan self-hostes i Azure Norway-regioner, noe som anbefales for data klassifisert Høy/Kritisk.",
|
||||
"disposition": "unsourced"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/embedding-models-selection.md#18",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/search/search-howto-reindex",
|
||||
"evidence_quote": "delete | Removes the entire document from the index. If you want to remove an individual field, use merge instead, setting the field in question to null.",
|
||||
"reason": "Første del stemmer, men Learn dokumenterer at et enkeltfelt — inkludert et vektorfelt — fjernes selektivt med merge og null uten reindeksering, så den bærende begrensningen claimet påstår er motsagt.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/embedding-models-selection.md",
|
||||
"line": 289,
|
||||
"claim": "Azure AI Search støtter sletting av dokumenter, men ikke selektiv sletting av embeddings uten reindeksering.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/embedding-models-selection.md#19",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R1",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/foundry/openai/how-to/embeddings",
|
||||
"evidence_quote": "The maximum input length for the current embedding models is 8,192 tokens. ... If you send an array of inputs in a single embedding request, the maximum array size is 2,048. ... Each /embeddings request has a 300,000-token aggregate limit across all inputs.",
|
||||
"reason": "Taket claimet setter (maks 8191 tokens totalt per batch-kall) er avløst: 8 192 gjelder per input, mens Learns gjeldende aggregerte grense per forespørsel er 300 000 tokens, og maks array-størrelse er 2 048 — det oppgitte taket binder ikke lenger.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/embedding-models-selection.md",
|
||||
"line": 360,
|
||||
"claim": "Et batch-kall til embeddings-API-et kan sende ca. 100 dokumenter per API-kall, med maks 8191 tokens totalt.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/embedding-models-selection.md#22",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure",
|
||||
"evidence_quote": "The following models are supported for fine-tuning: gpt-4o-mini (2024-07-18), gpt-4o (2024-08-06), gpt-4.1 (2025-04-14), gpt-4.1-mini (2025-04-14), gpt-4.1-nano (2025-04-14), o4-mini (2025-04-16), gpt-5 (2025-08-07), Ministral-3B (2411), Qwen-32B, Llama-3.3-70B-Instruct, gpt-oss-20b",
|
||||
"reason": "Jeg hentet den kanoniske listen over modeller som støtter fine-tuning i Foundry — den inneholder kun chat- og tekstmodeller, ingen embedding-modell, og ingen Custom Models-preview for embeddings.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/embedding-models-selection.md",
|
||||
"line": 449,
|
||||
"claim": "Microsoft Foundry støtter fine-tuning av embedding-modeller via Custom Models, som er i preview.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/embedding-models-selection.md#23",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure",
|
||||
"evidence_quote": "The following models are supported for fine-tuning: gpt-4o-mini (2024-07-18), gpt-4o (2024-08-06), gpt-4.1 (2025-04-14), gpt-4.1-mini (2025-04-14), gpt-4.1-nano (2025-04-14), o4-mini (2025-04-16), gpt-5 (2025-08-07), Ministral-3B (2411), Qwen-32B, Llama-3.3-70B-Instruct, gpt-oss-20b",
|
||||
"reason": "text-embedding-3-small og text-embedding-3-large står ikke i Learns liste over finjusterbare modeller, og verken JSON Lines-formatet eller query-document-par for embedding-finjustering er dokumentert.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/embedding-models-selection.md",
|
||||
"line": 453,
|
||||
"claim": "Fine-tuning av embeddings i Microsoft Foundry støtter modellene text-embedding-3-small | text-embedding-3-large, med treningsdata i JSON Lines-format bestående av query-document pairs.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/embedding-models-selection.md#24",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure",
|
||||
"evidence_quote": "The following models are supported for fine-tuning: gpt-4o-mini (2024-07-18), gpt-4o (2024-08-06), gpt-4.1 (2025-04-14), gpt-4.1-mini (2025-04-14), gpt-4.1-nano (2025-04-14), o4-mini (2025-04-16), gpt-5 (2025-08-07), Ministral-3B (2411), Qwen-32B, Llama-3.3-70B-Instruct, gpt-oss-20b",
|
||||
"reason": "Embedding-finjustering finnes ikke i den kanoniske listen over finjusterbare modeller, så kravet om minimum 100 positive query-document-par og anbefalingen om 1000+ har ingen dekning i Microsoft-dokumentasjonen.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/embedding-models-selection.md",
|
||||
"line": 455,
|
||||
"claim": "Fine-tuning av embedding-modeller krever minimum 100 positive query-document-par, med 1000+ som anbefalt antall.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/embedding-models-selection.md#25",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/foundry/foundry-models/concepts/models-sold-directly-by-azure",
|
||||
"evidence_quote": "The following models are supported for fine-tuning: gpt-4o-mini (2024-07-18), gpt-4o (2024-08-06), gpt-4.1 (2025-04-14), gpt-4.1-mini (2025-04-14), gpt-4.1-nano (2025-04-14), o4-mini (2025-04-16), gpt-5 (2025-08-07), Ministral-3B (2411), Qwen-32B, Llama-3.3-70B-Instruct, gpt-oss-20b",
|
||||
"reason": "Foundry tilbyr ikke finjustering av embedding-modeller ifølge den kanoniske listen, og metrikkene Recall@k, NDCG og MRR er ikke dokumentert som evalueringsmetrikker for slike jobber noe sted på learn.microsoft.com.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/embedding-models-selection.md",
|
||||
"line": 457,
|
||||
"claim": "Evaluering av fine-tunede embedding-modeller i Microsoft Foundry skjer med metrikkene Recall@k | NDCG | MRR mot valideringssett.",
|
||||
"disposition": "outdated"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/graphrag-knowledge-graphs.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/graphrag-knowledge-graphs.md#2",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R4",
|
||||
"evidence_url": "https://learn.microsoft.com/fabric/fundamentals/whats-new",
|
||||
"evidence_quote": "June 2026 | Fabric Graph (Generally Available) | Graph in Microsoft Fabric helps you model, visualize, and analyze complex relationships within your data.",
|
||||
"reason": "LPG-delen stemmer, men statusen matcher en utdatert rad: Fabric fører nå Fabric Graph som Generally Available (juni 2026), mens Public Preview er den arkiverte oktober 2025-raden, og oversiktssiden har ingen preview-merking på selve graph-produktet.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/graphrag-knowledge-graphs.md",
|
||||
"line": 40,
|
||||
"claim": "Microsoft Fabric Graph er i Public Preview og bruker Labeled Property Graph (LPG)-modellen.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/graphrag-knowledge-graphs.md#8",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/cosmos-db/gen-ai/cosmos-ai-graph",
|
||||
"evidence_quote": "By utilizing Cosmos DB's scalability and performance in both document and vector form, CosmosAIGraph enables the creation of sophisticated data models that can answer various data questions and uncover hidden relationships and concepts in semi-structured data.",
|
||||
"reason": "Den kanoniske siden beskriver CosmosAIGraph som en løsning som bruker Cosmos DB i document- og vector-form, ikke som en multi-model database; den sier ingen steder at document, vector og graph lagres i samme container.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/graphrag-knowledge-graphs.md",
|
||||
"line": 140,
|
||||
"claim": "CosmosAIGraph er en multi-model database som lagrer document | vector | graph i samme container.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/graphrag-knowledge-graphs.md#9",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/cosmos-db/gen-ai/cosmos-ai-graph",
|
||||
"evidence_quote": "It combines traditional database, vector database, and graph database capabilities with AI to manage and query complex data relationships efficiently. To get started, see the CosmosAIGraph repo.",
|
||||
"reason": "Den kanoniske CosmosAIGraph-siden nevner verken Gremlin API eller SQL API blant grensesnittene; «SQL API» er dessuten det gamle navnet som er erstattet av Azure Cosmos DB for NoSQL.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/graphrag-knowledge-graphs.md",
|
||||
"line": 143,
|
||||
"claim": "CosmosAIGraph eksponerer Gremlin API (graph traversal) | SQL API (document queries).",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/graphrag-knowledge-graphs.md#10",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R4",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/search/search-api-preview",
|
||||
"evidence_quote": "Knowledge bases and select knowledge sources are generally available. However, certain agentic retrieval capabilities remain in preview, including document-level permissions, answer synthesis, configurable retrieval reasoning effort, and multi-turn conversational messages in the retrieve request.",
|
||||
"reason": "Preview-oversikten sier at knowledge bases nå er generelt tilgjengelige og at kun utvalgte agentic retrieval-funksjoner er igjen i preview, så preview-statusen i påstanden er utdatert.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/graphrag-knowledge-graphs.md",
|
||||
"line": 150,
|
||||
"claim": "Knowledge base API i Azure AI Search er en preview-feature for agentic retrieval.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/graphrag-knowledge-graphs.md#12",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R4",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/ai-foundry/openai/concepts/legacy-models",
|
||||
"evidence_quote": "gpt-4 gpt-4-32k - 0314 | | June 6, 2025 | gpt-4o version: 2024-11-20",
|
||||
"reason": "GPT-4 står i tabellen over pensjonerte modeller (utgått 6. juni 2025, erstattet av gpt-4o), altså en legacy-rad; verken den siterte RAG-guiden eller noen annen Learn-side anbefaler GPT-4 eller «Opus» for kompleks graph-resonnering.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/graphrag-knowledge-graphs.md",
|
||||
"line": 156,
|
||||
"claim": "GPT-4 og Opus anbefales for kompleks graph-resonnering (path explanations, multi-hop inferenser).",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/graphrag-knowledge-graphs.md#14",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/privacy/eudb/eu-data-boundary-learn",
|
||||
"evidence_quote": "",
|
||||
"reason": "Learn dokumenterer at Norge er et EFTA-land innenfor EU Data Boundary, men ingen learn.microsoft.com-side konkluderer med at GraphRAG-data lagret i en norsk Azure-region oppfyller EU-kravene til dataresidens «under Schrems II» — en slik juridisk vurdering publiseres ikke der.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/graphrag-knowledge-graphs.md",
|
||||
"line": 190,
|
||||
"claim": "GraphRAG-data lagret i Azure Norway (oslo-region) oppfyller EU data residency-kravene under Schrems II.",
|
||||
"disposition": "unsourced"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/graphrag-knowledge-graphs.md#15",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/search/semantic-search-overview",
|
||||
"evidence_quote": "",
|
||||
"reason": "Semantic ranker har en gratisplan med «a monthly free request allowance», men verken oversikten eller faktureringssiden tallfester den til 1000 spørringer per måned — tallet ligger kun på den JS-rendrede Azure-prissiden.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/graphrag-knowledge-graphs.md",
|
||||
"line": 227,
|
||||
"claim": "Semantic ranking i Azure AI Search har en tier på 1000 queries per måned.",
|
||||
"disposition": "unsourced"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/hierarchical-rag-patterns.md",
|
||||
"batch": "R7.3",
|
||||
"judged_at": "2026-07-25",
|
||||
"per_file_verdict": "flagged",
|
||||
"claim_count": 15,
|
||||
"verified": null,
|
||||
"verified_by": null,
|
||||
"flags": [
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/hierarchical-rag-patterns.md#14",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/search-how-to-define-index-projections",
|
||||
"evidence_quote": "",
|
||||
"reason": "Ingen hentet learn.microsoft.com-side oppgir kostnad for index projections: projeksjonssiden nevner bare «Azure AI Search, any tier or region» som forutsetning, og kostnadssidens premium-funksjonstabell verken nevner index projections eller erklærer dem gratis — påstanden kan verken bekreftes eller motbevises.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/hierarchical-rag-patterns.md",
|
||||
"line": 253,
|
||||
"claim": "Index projections er inkludert i Azure AI Search uten ekstra kostnad.",
|
||||
"disposition": "unsourced"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/hybrid-search-configuration.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/hybrid-search-configuration.md#6",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/semantic-how-to-enable-disable",
|
||||
"evidence_quote": "",
|
||||
"reason": "Billing-siden og semantic-oversikten omtaler kun en månedlig gratis forespørselskvote uten tall, og tjenestegrensene oppgir ingen fri kvote for semantic ranker; tallet 1000 per måned står ikke på noen Learn-side (det ligger på den JS-rendrede prissiden), selv om delen om betaling etter forbrukt kvote er dekket.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/hybrid-search-configuration.md",
|
||||
"line": 97,
|
||||
"claim": "Semantic ranking gir 1000 gratis spørringer per måned; utover dette påløper ekstra kostnad per query.",
|
||||
"disposition": "unsourced"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/hybrid-search-configuration.md#8",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/search-region-support",
|
||||
"evidence_quote": "Europe: 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 finnes ikke i oppregningen; fraværet i den autoritative listen motbeviser at tjenesten er tilgjengelig i begge norske regioner.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/hybrid-search-configuration.md",
|
||||
"line": 167,
|
||||
"claim": "Azure AI Search er tilgjengelig i regionene Norway East og Norway West.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/hybrid-search-configuration.md#9",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R3",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/search-security-overview",
|
||||
"evidence_quote": "Azure AI Search stores and processes your data within that geography, but Microsoft might replicate data to other regions within the same geography for high availability. The exception is Brazil South, where data stays within the region.",
|
||||
"reason": "Garantien er på geografi-nivå, ikke valgt region: Microsoft kan replikere data til andre regioner i samme geografi (og objektnavn kan behandles utenfor valgt region), så region-rammen påstanden bygger på er erstattet; ingen hentet Learn-side sier at EU Data Boundary gjelder for norske Azure AI Search-deployments.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/hybrid-search-configuration.md",
|
||||
"line": 170,
|
||||
"claim": "All indeksdata i Azure AI Search forblir i valgt region, og Microsofts EU Data Boundary gjelder for norske deployments.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/hybrid-search-configuration.md#11",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/semantic-how-to-enable-disable",
|
||||
"evidence_quote": "Free (default) | Provides a monthly free request allowance. After the free allowance is consumed, semantic ranker requests return a billing error. | Available on all pricing tiers. || Standard | Pay-as-you-go pricing after the monthly free allowance is consumed. | Requires the Basic tier or higher.",
|
||||
"reason": "To bærende deler er feil: gratisplanen for semantic ranker er tilgjengelig på ALLE tier (betalt plan krever Basic eller høyere, ikke S1), og vector-search-overview slår fast at vektorsøk er tilgjengelig på alle tier uten ekstra kostnad, så hybrid search krever ikke Basic.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/hybrid-search-configuration.md",
|
||||
"line": 190,
|
||||
"claim": "Minimumstier i Azure AI Search: hybrid search (BM25 + vektor) krever Basic | scoring profiles er tilgjengelig på alle tier | semantic ranking krever S1 eller høyere (1000 gratis per måned) | integrert vektorisering krever Basic eller høyere.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/hybrid-search-configuration.md#12",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/search/vector-search-how-to-quantization",
|
||||
"evidence_quote": "Scalar quantization compresses float values into narrower data types. AI Search currently supports int8, which is 8 bits, reducing vector index size fourfold. Binary quantization converts floats into binary bits, which takes up 1 bit. This results in up to 28 times reduced vector index size.",
|
||||
"reason": "Begge bærende deler svikter: kvantiseringssiden har ingen preview-merking (vektoroptimaliseringsteknikkene omtales som generelt tilgjengelige), og reduksjonen er firedobbelt for scalar og opptil 28 ganger / 96 % for binary, altså langt over taket opptil 50 %.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/hybrid-search-configuration.md",
|
||||
"line": 196,
|
||||
"claim": "Scalar/binary quantization i Azure AI Search er i preview og reduserer vektorlagring med opptil 50 %.",
|
||||
"disposition": "outdated"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/late-chunking-patterns.md",
|
||||
"batch": "R7.3",
|
||||
"judged_at": "2026-07-25",
|
||||
"per_file_verdict": "flagged",
|
||||
"claim_count": 13,
|
||||
"verified": null,
|
||||
"verified_by": null,
|
||||
"flags": [
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/late-chunking-patterns.md#1",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/foundry/openai/concepts/model-retirement-schedule",
|
||||
"evidence_quote": "This section lists the retirement lifecycle for Foundry Models sold by partners via Azure Marketplace.",
|
||||
"reason": "Ingen learn.microsoft.com-side omtaler «late chunking», og den kanoniske opplistingen av modeller solgt via Azure Marketplace (Anthropic, Cohere, Fireworks, Meta, Microsoft, Mistral AI, Nixtla, NTT Data, StabilityAI) har ingen Jina-oppføring — verken GA- eller Preview-status for late chunking på Azure er dokumentert.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/late-chunking-patterns.md",
|
||||
"line": 4,
|
||||
"claim": "Late chunking-støtte er GA i Jina Embeddings og i Preview via Azure Marketplace.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/late-chunking-patterns.md#2",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/foundry/foundry-models/concepts/models-from-partners",
|
||||
"evidence_quote": "Foundry Models from partners and community available for deployment (for example, Cohere models) require Azure Marketplace.",
|
||||
"reason": "Den kanoniske opplistingen av partner-modeller på Azure Marketplace enumererer embedding-modeller (Cohere-embed-v3-english/multilingual) under leverandørene Anthropic, Cohere, Meta, Microsoft, Mistral AI og NTT Data, men Jina Embeddings v3/v4 er fraværende både her og i retirement-planen.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/late-chunking-patterns.md",
|
||||
"line": 31,
|
||||
"claim": "Jina Embeddings v3 og v4 er tilgjengelig på Azure via Azure Marketplace.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/late-chunking-patterns.md#9",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/foundry/foundry-models/concepts/models-from-partners",
|
||||
"evidence_quote": "| **Model** | **Version** | **francecentral** | **germanywestcentral** | **italynorth** | **norwayeast** | **polandcentral** | **spaincentral** | **swedencentral** | **switzerlandnorth** | **switzerlandwest** | **uksouth** | **ukwest** | **westeurope** |",
|
||||
"reason": "Den kanoniske region-tabellen for partner-modeller har norwayeast som egen kolonne, men ingen Jina-rad i det hele tatt — ingen Jina Embeddings-deployering i Norway East er dokumentert.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/late-chunking-patterns.md",
|
||||
"line": 192,
|
||||
"claim": "Jina Embeddings fra Azure Marketplace kan deployes i Norway East, slik at data forblir i Norge.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/late-chunking-patterns.md#12",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/foundry/foundry-models/concepts/models-from-partners",
|
||||
"evidence_quote": "",
|
||||
"reason": "Ingen hentet learn.microsoft.com-side beskriver Jina Embeddings på Azure, og verken Azure Container Instance-pakking eller consumption-basert prismodell for et slikt tilbud er dokumentert — påstanden kan verken bekreftes eller avkreftes mot Learn.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/late-chunking-patterns.md",
|
||||
"line": 220,
|
||||
"claim": "Jina Embeddings på Azure deployes som Azure Container Instance med consumption-basert prismodell.",
|
||||
"disposition": "unsourced"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/late-chunking-patterns.md#13",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/foundry/foundry-models/concepts/models-from-partners",
|
||||
"evidence_quote": "",
|
||||
"reason": "Ingen hentet learn.microsoft.com-side omtaler kvotemodellen for Jina Embeddings på Azure; at Azure OpenAI-kvote ikke kreves står ikke noe sted, og ingen side oppgir en avvikende verdi.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/late-chunking-patterns.md",
|
||||
"line": 222,
|
||||
"claim": "Bruk av Jina Embeddings på Azure krever ingen Azure OpenAI-kvote.",
|
||||
"disposition": "unsourced"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/metadata-management-filtering.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/metadata-management-filtering.md#3",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/search-query-odata-filter",
|
||||
"evidence_quote": "Find all hotels within a given viewport described as a polygon (where Location is a field of type Edm.GeographyPoint).",
|
||||
"reason": "Attributt-tabellen for de øvrige typene stemmer, men den uttalte, handlingsbærende innsnevringen Edm.GeographyPoint er filterable kun via geo.distance er motbevist: geo.intersects filtrerer også på Edm.GeographyPoint-felt.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/metadata-management-filtering.md",
|
||||
"line": 37,
|
||||
"claim": "Metadata-felttyper i Azure AI Search med egenskaper: Edm.String (filterable, facetable, sortable) | Collection(Edm.String) (filterable, facetable) | Edm.Int32/Edm.Int64 (filterable, facetable, sortable) | Edm.DateTimeOffset (filterable, facetable, sortable) | Edm.Boolean (filterable, facetable) | Edm.GeographyPoint (filterable kun via geo.distance, ikke facetable).",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/metadata-management-filtering.md#8",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/search/search-faceted-navigation-examples",
|
||||
"evidence_quote": "Using the latest preview REST API or the Azure portal, you can configure a facet hierarchy using the `>` and `;` operators.",
|
||||
"reason": "Preview-statusen for hierarkiske facetter er korrekt, men den uttalte, handlingsbærende delen med delimiter og levels er feil: hierarkiet konfigureres med operatorene > og ;, og verken delimiter eller levels finnes blant de gyldige facet-parameterne (count, sort, values, interval, timeoffset).",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/metadata-management-filtering.md",
|
||||
"line": 158,
|
||||
"claim": "Hierarkisk faceted navigation (hierarchical facets med delimiter og levels) er en preview-funksjon i Azure AI Search, ikke GA.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/metadata-management-filtering.md#11",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/search-query-odata-filter",
|
||||
"evidence_quote": "",
|
||||
"reason": "Den kanoniske seksjonen Filter size limitations oppgir bevisst ingen tallgrense, bare at hundrevis av klausuler gir risiko for å overskride grensen; ingen Learn-side oppgir en filter-klausulgrense på om lag 1000 (1024/3000 gjelder search-klausuler, ikke filter).",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/metadata-management-filtering.md",
|
||||
"line": 260,
|
||||
"claim": "Azure AI Search har en grense for filter-kompleksitet på om lag 1000 klausuler.",
|
||||
"disposition": "unsourced"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/metadata-management-filtering.md#12",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/search/search-how-to-index-sharepoint-online",
|
||||
"evidence_quote": "",
|
||||
"reason": "SharePoint-indexer-siden viser kun et eksempel (Title/Price/InStock/Category) med kompatible måltyper og gir ingen mapping for Managed Metadata, Date and Time eller Person or Group; ingen Learn-side oppgir den påståtte kolonnetype-til-Edm-tabellen eller at alle er filterable og facetable.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/metadata-management-filtering.md",
|
||||
"line": 270,
|
||||
"claim": "SharePoint-kolonnetyper mappes til Azure AI Search-typer slik: Single line of text → Edm.String | Choice → Edm.String | Managed Metadata → Collection(Edm.String) | Date and Time → Edm.DateTimeOffset | Person or Group → Collection(Edm.String), alle filterable og facetable.",
|
||||
"disposition": "unsourced"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/metadata-management-filtering.md#13",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/search/search-indexer-overview",
|
||||
"evidence_quote": "",
|
||||
"reason": "Dataverse står ikke i listen over støttede indexer-datakilder, og datatype-kartet for indexere dekker kun SQL Server og JSON; ingen Learn-side oppgir en Dataverse-til-Edm-mapping som kan bekrefte eller motbevise claimen.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/metadata-management-filtering.md",
|
||||
"line": 317,
|
||||
"claim": "Dataverse-typer mappes til Azure AI Search-typer slik: Choice → Edm.String | Choices (multi-select) → Collection(Edm.String) | Lookup → Edm.String (ID eller Name) | DateTime → Edm.DateTimeOffset.",
|
||||
"disposition": "unsourced"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/metadata-management-filtering.md#14",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R2",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/search/search-how-to-index-azure-blob-storage",
|
||||
"evidence_quote": "Currently, this indexer doesn't support indexing blob index tags.",
|
||||
"reason": "Den kanoniske enumereringen av standard blob-metadata (metadata_storage_name, _path, _content_type, _last_modified, _size, _content_md5, _sas_token) inneholder ikke metadata_storage_blob_index_tags, og siden slår eksplisitt fast at blob-indexeren ikke støtter indeksering av blob index tags.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/metadata-management-filtering.md",
|
||||
"line": 343,
|
||||
"claim": "Azure AI Search-indexer for Azure Blob Storage eksponerer blob index tags i feltet metadata_storage_blob_index_tags (Edm.String, filterable).",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/metadata-management-filtering.md#16",
|
||||
"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 ... After May 17, 2024 | 15 | 160 | 512 | 1,024",
|
||||
"reason": "Tallene 2/25/100/200 GB treffer kun den tidsstemplede historiske raden Before April 3, 2024; gjeldende rad for nye tjenester er Basic 15, S1 160, S2 512, S3/HD 1 024 GB.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/metadata-management-filtering.md",
|
||||
"line": 435,
|
||||
"claim": "Maks indeksstørrelse per tier i Azure AI Search: Basic 2 GB | S1 25 GB per partisjon | S2 100 GB per partisjon | S3/S3HD 200 GB per partisjon.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/metadata-management-filtering.md#17",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/search-limits-quotas-capacity",
|
||||
"evidence_quote": "Maximum simple fields per index | 1000 | 100 | 1000 | 1000 | 1000 | 1000 | 1000 | 1000 | 1000 ... Basic tier is the only tier with a lower limit of 100 fields per index.",
|
||||
"reason": "Begge de tallfestede delene er feil: Basic har 100 felt per indeks (ikke 1000), og S3/S3 HD har 1000 (ikke 3000) — 3000 gjelder maks indekser per S3 HD-tjeneste og elementer i komplekse samlinger, ikke felt.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/metadata-management-filtering.md",
|
||||
"line": 438,
|
||||
"claim": "Maks antall felt per indeks er 1000 for Basic, S1 og S2, og 3000 for S3/S3HD.",
|
||||
"disposition": "outdated"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/multi-index-federation.md",
|
||||
"batch": "R7.3",
|
||||
"judged_at": "2026-07-25",
|
||||
"per_file_verdict": "flagged",
|
||||
"claim_count": 13,
|
||||
"verified": null,
|
||||
"verified_by": null,
|
||||
"flags": [
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/multi-index-federation.md#10",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/en-us/azure/search/vector-search-multi-vector-fields",
|
||||
"evidence_quote": "This feature is currently in preview. This preview is provided without a service-level agreement and isn't recommended for production workloads.",
|
||||
"reason": "Claimen sier GA (2025), men den kanoniske siden merker funksjonen som preview i dag, og «Preview features in Azure AI Search» lister «Multivector support» under data plane preview features (Create or Update Index (preview)) — status er motsagt.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/multi-index-federation.md",
|
||||
"line": 63,
|
||||
"claim": "Multi-vector field support i Azure AI Search er GA (2025).",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/multi-index-federation.md#11",
|
||||
"judge_verdict": "not_grounded",
|
||||
"rule": "R8",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/search/agentic-retrieval-overview",
|
||||
"evidence_quote": "The knowledge base sends the subqueries to your knowledge sources. All subqueries run simultaneously and can be keyword, vector, or hybrid search.",
|
||||
"reason": "Query planning-delen holder («the knowledge base sends your query and conversation history to an LLM, which generates focused subqueries»), men den bærende delen «fremdeles single-index» er motsagt: en knowledge base spør én eller flere knowledge sources (indekserte og eksterne) parallelt — og «Preview» er også superseded, siden «Some agentic retrieval features are generally available in the 2026-04-01 REST API» og preview-oversikten sier «Knowledge bases and select knowledge sources are generally available».",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/multi-index-federation.md",
|
||||
"line": 64,
|
||||
"claim": "Agentic Retrieval i Azure AI Search er i Preview og gir LLM-assistert query planning, men er fremdeles single-index.",
|
||||
"disposition": "outdated"
|
||||
},
|
||||
{
|
||||
"id": "ms-ai-engineering/rag-architecture/multi-index-federation.md#13",
|
||||
"judge_verdict": "source_silent",
|
||||
"rule": "",
|
||||
"evidence_url": "https://learn.microsoft.com/azure/search/search-multi-region",
|
||||
"evidence_quote": "",
|
||||
"reason": "Claimen beskriver referansefilens eget kodeeksempel (modellvalg gpt-4o-mini via chat.completions), noe ingen learn.microsoft.com-side kan bekrefte eller avkrefte; den siterte siden omtaler ikke query routing eller modellvalg, og søk fant ingen Learn-side som uttaler seg om eksempelet — modellen gpt-4o-mini finnes fortsatt (oppført som Deprecated med retirement 2027-04-14 i retirement-schedule), så ingenting motsier claimen heller.",
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/multi-index-federation.md",
|
||||
"line": 187,
|
||||
"claim": "Eksempelet for LLM-basert query routing bruker modellen gpt-4o-mini via chat.completions-API-et.",
|
||||
"disposition": "unsourced"
|
||||
}
|
||||
]
|
||||
},
|
||||
{
|
||||
"file": "skills/ms-ai-engineering/references/rag-architecture/rag-context-windows.md",
|
||||
"batch": "R7.1",
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue