[{"data":1,"prerenderedAt":1833},["ShallowReactive",2],{"blog-tag-hu-rag":3},[4,88,149,208,272,340,410,465,521,564,614,684,757,825,893,957,1025,1091,1146,1185,1236,1286,1325,1400,1444,1498,1553,1609,1659,1734],{"slug":5,"published":6,"minutes":7,"category":8,"tags":9,"tagPages":15,"keywords":18,"about":31,"sources":41,"cover":78,"og":79,"expertise":80,"locales":81,"lang":84,"title":85,"description":86,"coverAlt":87},"embedding-model-selection-2026","2026-10-17",9,"rag",[10,11,12,13,14],"Embeddings","RAG","MTEB","Vector search","Model selection",[16,8,17],"embeddings","vector-search",[19,20,21,22,23,24,25,26,27,28,29,30],"legjobb embedding modell 2026","hogyan válasszunk embedding modellt","embedding modell összehasonlítás","mteb ranglista magyarázat","rteb benchmark","nyílt forráskódú embedding modell","többnyelvű embedding modell","voyage-4 vagy openai embedding","qwen3-embedding","embeddinggemma","embedding modell rag-hoz","embedding modell csere újraindexelés",[32,35,38],{"name":33,"url":34},"Word embedding","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FWord_embedding",{"name":36,"url":37},"Information retrieval","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FInformation_retrieval",{"name":39,"url":40},"Discounted cumulative gain","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FDiscounted_cumulative_gain",[42,45,48,51,54,57,60,63,66,69,72,75],{"title":43,"url":44},"Enevoldsen et al.: MMTEB, Massive Multilingual Text Embedding Benchmark (2025)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2502.13595",{"title":46,"url":47},"Hugging Face: Introducing RTEB, a new standard for retrieval evaluation (Oct 2025)","https:\u002F\u002Fhuggingface.co\u002Fblog\u002Frteb",{"title":49,"url":50},"OpenAI: Vector embeddings guide","https:\u002F\u002Fdevelopers.openai.com\u002Fapi\u002Fdocs\u002Fguides\u002Fembeddings",{"title":52,"url":53},"OpenAI: API pricing","https:\u002F\u002Fdevelopers.openai.com\u002Fapi\u002Fdocs\u002Fpricing",{"title":55,"url":56},"Voyage AI: The Voyage 4 model family, shared embedding space with MoE architecture (Jan 2026)","https:\u002F\u002Fblog.voyageai.com\u002F2026\u002F01\u002F15\u002Fvoyage-4\u002F",{"title":58,"url":59},"Voyage AI: Text embeddings documentation","https:\u002F\u002Fdocs.voyageai.com\u002Fdocs\u002Fembeddings",{"title":61,"url":62},"Voyage AI: Pricing","https:\u002F\u002Fdocs.voyageai.com\u002Fdocs\u002Fpricing",{"title":64,"url":65},"Google: Gemini API embeddings documentation","https:\u002F\u002Fai.google.dev\u002Fgemini-api\u002Fdocs\u002Fembeddings",{"title":67,"url":68},"Google: Gemini API pricing","https:\u002F\u002Fai.google.dev\u002Fgemini-api\u002Fdocs\u002Fpricing",{"title":70,"url":71},"Cohere: Embed models documentation","https:\u002F\u002Fdocs.cohere.com\u002Fdocs\u002Fcohere-embed",{"title":73,"url":74},"Qwen: Qwen3-Embedding-8B model card","https:\u002F\u002Fhuggingface.co\u002FQwen\u002FQwen3-Embedding-8B",{"title":76,"url":77},"Google Developers Blog: Introducing EmbeddingGemma (Sep 2025)","https:\u002F\u002Fdevelopers.googleblog.com\u002Fen\u002Fintroducing-embeddinggemma\u002F","\u002Fimages\u002Fblog\u002Fembedding-model-selection-2026\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fembedding-model-selection-2026\u002Fog.jpg","ai-engineer",[82,83,84],"en","de","hu","Embedding-modell választása 2026-ban: ranglisták, költségek és összehasonlító teszt a saját adatokon","Hogyan válassz embedding-modellt RAG-hoz 2026-ban: mit mondanak az MTEB és az RTEB, API vagy nyílt súlyok, árak, dimenziók és egy egyhetes teszt.","Embedding-modell választása: három sütemény hűl a formájából, egy villa épp megkóstol egyet.",{"slug":89,"published":6,"minutes":90,"category":8,"tags":91,"tagPages":95,"keywords":96,"about":108,"sources":118,"cover":143,"og":144,"expertise":80,"locales":145,"lang":84,"title":146,"description":147,"coverAlt":148},"contextual-retrieval-late-chunking",10,[11,92,10,93,94],"Chunking","Contextual retrieval","Late chunking",[8,16],[97,98,99,100,101,102,103,104,105,106,107],"kontextuális visszakeresés","anthropic kontextuális visszakeresés","késleltetett chunkelés","kontextuális chunk beágyazások","kontextuális beágyazások rag","contextual bm25","chunk kontextus rag","voyage-context-3","voyage-context-4","late chunking vs kontextuális visszakeresés","rag chunkelés kontextusvesztés",[109,112,115],{"name":110,"url":111},"Retrieval-augmented generation","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FRetrieval-augmented_generation",{"name":113,"url":114},"Okapi BM25","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FOkapi_BM25",{"name":116,"url":117},"Sentence embedding","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FSentence_embedding",[119,122,125,128,131,134,137,140],{"title":120,"url":121},"Anthropic: Introducing Contextual Retrieval (Sep 2024)","https:\u002F\u002Fwww.anthropic.com\u002Fnews\u002Fcontextual-retrieval",{"title":123,"url":124},"Günther et al.: Late Chunking, Contextual Chunk Embeddings Using Long-Context Embedding Models (2024, rev. 2025)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2409.04701",{"title":126,"url":127},"Jina AI: Late Chunking in Long-Context Embedding Models (Aug 2024)","https:\u002F\u002Fjina.ai\u002Fnews\u002Flate-chunking-in-long-context-embedding-models\u002F",{"title":129,"url":130},"Voyage AI: Introducing voyage-context-3 (Jul 2025)","https:\u002F\u002Fblog.voyageai.com\u002F2025\u002F07\u002F23\u002Fvoyage-context-3\u002F",{"title":132,"url":133},"Voyage AI: voyage-context-4 (Jun 2026)","https:\u002F\u002Fblog.voyageai.com\u002F2026\u002F06\u002F29\u002Fvoyage-context-4\u002F",{"title":135,"url":136},"Voyage AI: Contextualized chunk embeddings documentation","https:\u002F\u002Fdocs.voyageai.com\u002Fdocs\u002Fcontextualized-chunk-embeddings",{"title":138,"url":139},"Anthropic: Claude API pricing","https:\u002F\u002Fplatform.claude.com\u002Fdocs\u002Fen\u002Fabout-claude\u002Fpricing",{"title":141,"url":142},"Qu, Tu and Bao: Is Semantic Chunking Worth the Computational Cost? (2024)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2410.13070","\u002Fimages\u002Fblog\u002Fcontextual-retrieval-late-chunking\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fcontextual-retrieval-late-chunking\u002Fog.jpg",[82,83,84],"Kontextuális visszakeresés és késleltetett chunkolás: a RAG-chunkok visszakapják a kontextusukat","A chunkok elveszítik a dokumentumukat. Contextual retrieval, late chunking és kontextusos embedding: mennyibe kerülnek, és mikor hagyhatók ki.","Kontextuális visszakeresés: egy kéz címkés kirakósdarabot tart a helye fölé az asztalon fekvő kirakóban.",{"slug":150,"published":6,"minutes":7,"category":8,"tags":151,"tagPages":155,"keywords":156,"about":169,"sources":177,"cover":202,"og":203,"expertise":80,"locales":204,"lang":84,"title":205,"description":206,"coverAlt":207},"rag-access-control-permissions",[11,152,153,154,13],"Access control","Security","Multi-tenancy",[8,17],[157,158,159,160,161,162,163,164,165,166,167,168],"rag hozzáférés-kezelés","jogosultságtudatos rag","dokumentumszintű biztonság rag","rag multi tenancy","vektoradatbázis hozzáférés-kezelés","pgvector row level security","security trimming azure ai search","openfga rag jogosultságkezelés","qdrant multitenancy","pinecone namespace multitenancy","owasp llm08 vektor és beágyazási gyengeségek","rag adatszivárgás",[170,173,176],{"name":171,"url":172},"Access-control list","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FAccess-control_list",{"name":174,"url":175},"Multitenancy","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FMultitenancy",{"name":110,"url":111},[178,181,184,187,190,193,196,199],{"title":179,"url":180},"OWASP GenAI Security Project: LLM08:2025 Vector and Embedding Weaknesses","https:\u002F\u002Fgenai.owasp.org\u002Fllmrisk\u002Fllm082025-vector-and-embedding-weaknesses\u002F",{"title":182,"url":183},"PostgreSQL documentation: Row Security Policies","https:\u002F\u002Fwww.postgresql.org\u002Fdocs\u002Fcurrent\u002Fddl-rowsecurity.html",{"title":185,"url":186},"pgvector README: filtering and iterative index scans","https:\u002F\u002Fgithub.com\u002Fpgvector\u002Fpgvector",{"title":188,"url":189},"Qdrant: Multitenancy and partitions guide","https:\u002F\u002Fqdrant.tech\u002Fdocumentation\u002Fguides\u002Fmultiple-partitions\u002F",{"title":191,"url":192},"Pinecone: Implement multitenancy using namespaces","https:\u002F\u002Fdocs.pinecone.io\u002Fguides\u002Findex-data\u002Fimplement-multitenancy",{"title":194,"url":195},"Microsoft Learn: Document-level access control in Azure AI Search","https:\u002F\u002Flearn.microsoft.com\u002Fen-us\u002Fazure\u002Fsearch\u002Fsearch-document-level-access-overview",{"title":197,"url":198},"OpenFGA: RAG authorization use case","https:\u002F\u002Fopenfga.dev\u002Fdocs\u002Fuse-cases\u002Frag-authorization",{"title":200,"url":201},"Morris et al.: Text Embeddings Reveal (Almost) As Much As Text (EMNLP 2023)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2310.06816","\u002Fimages\u002Fblog\u002Frag-access-control-permissions\u002Fcover.webp","\u002Fimages\u002Fblog\u002Frag-access-control-permissions\u002Fog.jpg",[82,83,84],"Hozzáférés-kezelés RAG-hoz: jogosultságtudatos visszakeresés dokumentumszivárgás nélkül","Hogyan szivárogtatnak dokumentumot a RAG-asszisztensek, és hogyan állítsd meg: ACL minden chunkon, előszűrés, sorszintű biztonság pgvectorban, jogszinkron.","Hozzáférés-kezelés RAG-hoz: belépőkártyás ember könyvet vesz le, mások lezárt üvegajtó mögött maradnak.",{"slug":209,"published":6,"minutes":90,"category":8,"tags":210,"tagPages":215,"keywords":218,"about":230,"sources":238,"cover":265,"og":266,"expertise":267,"locales":268,"lang":84,"title":269,"description":270,"coverAlt":271},"rag-product-catalog-pim-data",[11,211,212,213,214],"PIM","Product data","B2B e-commerce","Metadata filtering",[8,216,217],"pim","b2b-e-commerce",[219,220,221,222,223,224,225,226,227,228,229],"rag termékkatalógus","rag pim adatok","llm termékadatok","termékkatalógus chatbot","akeneo rag","pimcore ai keresés","rag strukturált adatok","metaadat szűrés rag","termékattribútumok beágyazások","b2b termékasszisztens","etim osztályozás ai",[231,234,235],{"name":232,"url":233},"Product information management","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FProduct_information_management",{"name":110,"url":111},{"name":236,"url":237},"ETIM (standard)","https:\u002F\u002Fwww.etim-international.com\u002Fclassification\u002F",[239,242,245,248,250,253,256,259,262],{"title":240,"url":241},"Akeneo API: Products, product values, metric values and product models","https:\u002F\u002Fapi.akeneo.com\u002Fconcepts\u002Fproducts.html",{"title":243,"url":244},"Akeneo Help Center: What about products with variants?","https:\u002F\u002Fhelp.akeneo.com\u002Fserenity-take-the-power-over-your-products\u002Fserenity-what-about-products-with-variants",{"title":246,"url":247},"Pimcore documentation: Classification Store","https:\u002F\u002Fdocs.pimcore.com\u002Fplatform\u002FPimcore\u002FObjects\u002FObject_Classes\u002FData_Types\u002FClassification_Store\u002F",{"title":249,"url":237},"ETIM International: Classification",{"title":251,"url":252},"ETIM International: ETIM releases and the IXF format","https:\u002F\u002Fetim-international.com\u002F?p=754",{"title":254,"url":255},"Reddy et al.: Shopping Queries Dataset, a large-scale ESCI benchmark for improving product search (2022)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2206.06588",{"title":257,"url":258},"Amazon Science: esci-data repository","https:\u002F\u002Fgithub.com\u002Famazon-science\u002Fesci-data",{"title":260,"url":261},"Qdrant documentation: Filtering","https:\u002F\u002Fqdrant.tech\u002Fdocumentation\u002Fconcepts\u002Ffiltering\u002F",{"title":263,"url":264},"Sui et al.: Table Meets LLM, can large language models understand structured table data? (WSDM 2024)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2305.13062","\u002Fimages\u002Fblog\u002Frag-product-catalog-pim-data\u002Fcover.webp","\u002Fimages\u002Fblog\u002Frag-product-catalog-pim-data\u002Fog.jpg","b2b-ecommerce-developer",[82,83,84],"RAG termékkatalógusra: PIM-adatokból megbízható válaszok","RAG-hoz: a PDF-szerű chunkolás elbukik, mert a termékadat strukturált. Sorosítsd a PIM-attribútumokat, és szűrőkre irányítsd a specifikációs kérdéseket.","RAG termékkatalógusra: nyitott katalógusból termékkártyás szövegbuborék emelkedik fel.",{"slug":273,"published":6,"minutes":274,"category":8,"tags":275,"tagPages":280,"keywords":281,"about":292,"sources":300,"cover":334,"og":335,"expertise":80,"locales":336,"lang":84,"title":337,"description":338,"coverAlt":339},"conversational-search-query-rewriting",12,[276,277,11,278,279],"Conversational search","Query rewriting","HyDE","E-commerce",[8],[282,283,284,285,286,287,288,289,290,291],"beszélgetésalapú keresés lekérdezés-újraírás","korreferencia feloldás keresési lekérdezés","hyde hipotetikus dokumentum beágyazások","többlekérdezéses bővítés rag","llm szűrő kinyerés","b2b termékkereső asszisztens","beszélgetésalapú termékkeresés","lekérdezés-újraírás késleltetés","recall@k kiértékelés","önálló lekérdezés generálás",[293,296,299],{"name":294,"url":295},"Query expansion","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FQuery_expansion",{"name":297,"url":298},"Coreference","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FCoreference",{"name":113,"url":114},[301,304,307,310,313,316,319,322,325,328,331],{"title":302,"url":303},"Gao et al.: Precise Zero-Shot Dense Retrieval without Relevance Labels, ACL 2023","https:\u002F\u002Faclanthology.org\u002F2023.acl-long.99\u002F",{"title":305,"url":306},"UTokyo-Hitotsubashi at TREC 2025 RAG Track: HyDE-Enhanced Sparse-Dense Retrieval Fusion with LLM Reranking","https:\u002F\u002Ftrec.nist.gov\u002Fpubs\u002Ftrec34\u002Fpapers\u002FUTokyo.rag.pdf",{"title":308,"url":309},"Hypothetical Documents or Knowledge Leakage? Rethinking LLM-based Query Expansion","https:\u002F\u002Faclanthology.org\u002F2025.findings-acl.980.pdf",{"title":311,"url":312},"ConvSearch-R1: Enhancing Query Reformulation for Conversational Search with Reasoning via Reinforcement Learning","https:\u002F\u002Faclanthology.org\u002F2025.emnlp-main.1349\u002F",{"title":314,"url":315},"CFDA and CLIP at TREC iKAT 2025: Query Reformulation and Rank Fusion","https:\u002F\u002Farxiv.org\u002Fabs\u002F2509.15588",{"title":317,"url":318},"MiniELM: A Lightweight and Adaptive Query Rewriting Framework for E-Commerce Search Optimization","https:\u002F\u002Faclanthology.org\u002F2025.findings-acl.363\u002F",{"title":320,"url":321},"Microsoft Learn: Rewrite queries with semantic ranker in Azure AI Search (preview)","https:\u002F\u002Flearn.microsoft.com\u002Fen-us\u002Fazure\u002Fsearch\u002Fsemantic-how-to-query-rewrite",{"title":323,"url":324},"Microsoft Learn: Agentic retrieval overview in Azure AI Search","https:\u002F\u002Flearn.microsoft.com\u002Fen-us\u002Fazure\u002Fsearch\u002Fagentic-retrieval-overview",{"title":326,"url":327},"Microsoft Tech Community: Generative query rewriting and a new semantic ranker","https:\u002F\u002Ftechcommunity.microsoft.com\u002Fblog\u002Fazure-ai-services-blog\u002Fraising-the-bar-for-rag-excellence-query-rewriting-and-new-semantic-ranker\u002F4302729\u002F",{"title":329,"url":330},"Amazon Bedrock User Guide: Configure and customize queries and response generation","https:\u002F\u002Fdocs.aws.amazon.com\u002Fbedrock\u002Flatest\u002Fuserguide\u002Fkb-test-config.html",{"title":332,"url":333},"Ragas documentation: List of available metrics","https:\u002F\u002Fdocs.ragas.io\u002Fen\u002Fstable\u002Fconcepts\u002Fmetrics\u002Favailable_metrics\u002F","\u002Fimages\u002Fblog\u002Fconversational-search-query-rewriting\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fconversational-search-query-rewriting\u002Fog.jpg",[82,83,84],"Lekérdezés-újraírás beszélgetésalapú kereséshez: kontextus, HyDE és szűrők","Hogyan írjuk át a követő kérdéseket egy B2B webshop-asszisztensben: korreferencia, HyDE, több lekérdezés, szűrők, latencia és kiértékelés.","Lekérdezés-újraírás kereséshez: szövegbuborék egy varrógépen át rendezett cetliként jön ki, rajta nagyító.",{"slug":341,"published":6,"minutes":90,"category":8,"tags":342,"tagPages":347,"keywords":349,"about":362,"sources":368,"cover":404,"og":405,"expertise":80,"locales":406,"lang":84,"title":407,"description":408,"coverAlt":409},"synthetic-test-data-rag-evals",[11,343,344,345,346],"Evaluation","Synthetic data","Retrieval","Testing",[8,348],"evaluation",[350,351,352,353,354,355,356,357,358,359,360,361],"szintetikus tesztadat rag","rag kiértékelő adathalmaz","kérdésgenerálás chunkekből","hard negatívok visszakeresés kiértékelés","válaszolhatósági szűrő","rag tesztkészlet generálás","szintetikus kérdés-válasz párok","rag eval adat kontamináció","többnyelvű rag kiértékelés német magyar","visszakeresési mutatók recall ndcg","llm bíró emberi ellenőrzés","megválaszolhatatlan kérdések rag",[363,365,367],{"name":110,"url":364},"https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FRetrieval_augmented_generation",{"name":344,"url":366},"https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FSynthetic_data",{"name":36,"url":37},[369,372,375,378,381,384,387,390,393,396,399,401],{"title":370,"url":371},"Red Hat Developers: Synthetic data for RAG evaluation (Feb 23, 2026)","https:\u002F\u002Fdevelopers.redhat.com\u002Farticles\u002F2026\u002F02\u002F23\u002Fsynthetic-data-rag-evaluation-why-your-rag-system-needs-better-testing",{"title":373,"url":374},"Ragas docs: Testset Generation for RAG","https:\u002F\u002Fdocs.ragas.io\u002Fen\u002Fstable\u002Fgetstarted\u002Frag_testset_generation\u002F",{"title":376,"url":377},"Ragas docs: Testset generation concepts (query types, scenarios)","https:\u002F\u002Fdocs.ragas.io\u002Fen\u002Fstable\u002Fconcepts\u002Ftest_data_generation\u002Frag\u002F",{"title":379,"url":380},"NVIDIA Technical Blog: Evaluating and Enhancing RAG Pipeline Performance Using Synthetic Data","https:\u002F\u002Fdeveloper.nvidia.com\u002Fblog\u002Fevaluating-and-enhancing-rag-pipeline-performance-using-synthetic-data\u002F",{"title":382,"url":383},"DeepEval docs: Datasets and the Synthesizer","https:\u002F\u002Fdeepeval.com\u002Fdocs\u002Fevaluation-datasets",{"title":385,"url":386},"Rajpurkar, Jia, Liang: Know What You Don't Know: Unanswerable Questions for SQuAD (ACL 2018)","https:\u002F\u002Farxiv.org\u002Fabs\u002F1806.03822",{"title":388,"url":389},"Rahmani et al.: Towards Understanding Bias in Synthetic Data for Evaluation (CIKM 2025)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2506.10301",{"title":391,"url":392},"Cheng et al.: A Survey on Data Contamination for Large Language Models","https:\u002F\u002Farxiv.org\u002Fabs\u002F2502.14425",{"title":394,"url":395},"Xu et al.: Infini-gram mini: Exact n-gram Search at the Internet Scale (EMNLP 2025)","https:\u002F\u002Faclanthology.org\u002F2025.emnlp-main.1268\u002F",{"title":397,"url":398},"Panickssery, Bowman, Feng: LLM Evaluators Recognize and Favor Their Own Generations","https:\u002F\u002Farxiv.org\u002Fabs\u002F2404.13076",{"title":400,"url":44},"Enevoldsen et al.: MMTEB, Massive Multilingual Text Embedding Benchmark (ICLR 2025)",{"title":402,"url":403},"Evidently AI: How to align an LLM judge with human labels (Nov 2025)","https:\u002F\u002Fwww.evidentlyai.com\u002Fblog\u002Fhow-to-align-llm-judge-with-human-labels","\u002Fimages\u002Fblog\u002Fsynthetic-test-data-rag-evals\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fsynthetic-test-data-rag-evals\u002Fog.jpg",[82,83,84],"Szintetikus tesztadatok generálása a RAG-értékeléshez","Kérdésgenerálás blokkokból, hard negatívok, válaszolhatósági címkék, német és magyar lefedettség, kontamináció és emberi ellenőrzések.","Szintetikus tesztadatok RAG-értékeléshez: robot olyan gépet kezel, amely egy könyvből kérdéskártyákat készít.",{"slug":411,"published":412,"minutes":7,"category":8,"tags":413,"tagPages":418,"keywords":420,"about":431,"sources":439,"cover":459,"og":460,"expertise":80,"locales":461,"lang":84,"title":462,"description":463,"coverAlt":464},"pdf-document-parsing-for-rag","2026-10-16",[11,414,415,416,417],"PDF parsing","OCR","Docling","Document AI",[8,419],"document-ai",[421,422,423,424,425,426,427,428,429,430],"pdf feldolgozás rag-hez","legjobb pdf parser rag-hez","docling vs marker","marker vs mineru","pdf markdown llm-hez","táblázatok kinyerése pdf-ből rag-hez","ocr rag pipeline","olmocr bench","szkennelt pdf rag","dokumentum feldolgozás llm",[432,435,438],{"name":433,"url":434},"PDF","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FPDF",{"name":436,"url":437},"Optical character recognition","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FOptical_character_recognition",{"name":110,"url":111},[440,443,446,449,452,455,458],{"title":441,"url":442},"Docling: README, features and license (GitHub)","https:\u002F\u002Fgithub.com\u002Fdocling-project\u002Fdocling",{"title":444,"url":445},"Auer et al.: Docling: An Efficient Open-Source Toolkit for AI-driven Document Conversion (2025)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2501.17887",{"title":447,"url":448},"Datalab: Marker README, olmOCR-bench results and license (GitHub)","https:\u002F\u002Fgithub.com\u002Fdatalab-to\u002Fmarker",{"title":450,"url":451},"OpenDataLab: MinerU README and license (GitHub)","https:\u002F\u002Fgithub.com\u002Fopendatalab\u002FMinerU",{"title":453,"url":454},"Poznanski et al.: olmOCR: Unlocking Trillions of Tokens in PDFs with Vision Language Models (2025)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2502.18443",{"title":456,"url":457},"Claude Platform Docs: PDF support","https:\u002F\u002Fplatform.claude.com\u002Fdocs\u002Fen\u002Fbuild-with-claude\u002Fpdf-support",{"title":120,"url":121},"\u002Fimages\u002Fblog\u002Fpdf-document-parsing-for-rag\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fpdf-document-parsing-for-rag\u002Fog.jpg",[82,83,84],"PDF-feldolgozás RAG-hoz: layout-pipeline-ok, OCR és vizuális modellek összevetve","Miért a PDF-feldolgozáson múlik a RAG minősége, miben más a Docling, a Marker, a MinerU és a VLM-OCR, és hogyan teszteld a parsert saját dokumentumokon.","PDF-feldolgozás RAG-hoz: dokumentum hengeres gépen halad át, és külön táblázat-, kép- és szövegkártyákként jön ki.",{"slug":466,"published":412,"minutes":467,"category":8,"tags":468,"tagPages":472,"keywords":473,"about":484,"sources":494,"cover":515,"og":516,"expertise":80,"locales":517,"lang":84,"title":518,"description":519,"coverAlt":520},"multilingual-rag-german-hungarian",7,[11,469,10,470,471],"Multilingual","German","Hungarian",[8,16],[474,475,476,477,478,479,480,481,482,483],"többnyelvű rag","német rag beágyazások","magyar nlp keresés","nyelvek közötti visszakeresés","többnyelvű beágyazó modell","bge-m3 német","német összetett szavak keresése","magyar szótövezés postgres","multilingual-e5-large-instruct","qwen3 embedding többnyelvű",[485,488,491],{"name":486,"url":487},"Cross-language information retrieval","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FCross-language_information_retrieval",{"name":489,"url":490},"Agglutinative language","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FAgglutinative_language",{"name":492,"url":493},"Compound (linguistics)","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FCompound_(linguistics)",[495,497,500,503,506,509,512],{"title":496,"url":44},"Enevoldsen et al.: MMTEB: Massive Multilingual Text Embedding Benchmark (2025)",{"title":498,"url":499},"Chen et al.: M3-Embedding (BGE-M3): Multi-Linguality, Multi-Functionality, Multi-Granularity (2024)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2402.03216",{"title":501,"url":502},"Zhang et al.: Qwen3 Embedding: Advancing Text Embedding and Reranking Through Foundation Models (2025)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2506.05176",{"title":504,"url":505},"Hugging Face: intfloat\u002Fmultilingual-e5-large-instruct model card","https:\u002F\u002Fhuggingface.co\u002Fintfloat\u002Fmultilingual-e5-large-instruct",{"title":507,"url":508},"Elasticsearch Reference: Hyphenation decompounder token filter","https:\u002F\u002Fwww.elastic.co\u002Fdocs\u002Freference\u002Ftext-analysis\u002Fanalysis-hyp-decomp-tokenfilter",{"title":510,"url":511},"PostgreSQL 18 Documentation: Text search dictionaries","https:\u002F\u002Fwww.postgresql.org\u002Fdocs\u002Fcurrent\u002Ftextsearch-dictionaries.html",{"title":513,"url":514},"Snowball: stemming algorithms by language","https:\u002F\u002Fsnowballstem.org\u002Falgorithms\u002F","\u002Fimages\u002Fblog\u002Fmultilingual-rag-german-hungarian\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fmultilingual-rag-german-hungarian\u002Fog.jpg",[82,83,84],"Többnyelvű RAG német és magyar tartalomra: embeddingek, összetett szavak és nyelvek közötti keresés","RAG német és magyar tartalom fölött: többnyelvű embeddingek, összetett szavak, szótövezés, ékezetek és nyelvenkénti értékelés.","Többnyelvű RAG német és magyar tartalomra: egy ember könyvet visz át egy kis hídon két könyvespolc között.",{"slug":522,"published":412,"minutes":523,"category":8,"tags":524,"tagPages":527,"keywords":528,"about":539,"sources":549,"cover":558,"og":559,"expertise":80,"locales":560,"lang":84,"title":561,"description":562,"coverAlt":563},"embedding-quantization-binary-matryoshka",5,[10,13,525,526,11],"Quantization","pgvector",[16,17,8],[529,530,531,532,533,534,535,536,537,538],"embedding kvantálás","bináris kvantálás beágyazások","int8 embedding","matryoshka embedding","vektoradatbázis memória csökkentése","pgvector halfvec","pgvector binary_quantize","embedding dimenziók csökkentése","bináris beágyazások újrapontozása","vektorindex memóriaköltség",[540,543,546],{"name":541,"url":542},"Quantization (signal processing)","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FQuantization_(signal_processing)",{"name":544,"url":545},"Nearest neighbor search","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FNearest_neighbor_search",{"name":547,"url":548},"Hamming distance","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FHamming_distance",[550,553,556],{"title":551,"url":552},"Hugging Face: Binary and Scalar Embedding Quantization for Significantly Faster and Cheaper Retrieval (Mar 2024)","https:\u002F\u002Fhuggingface.co\u002Fblog\u002Fembedding-quantization",{"title":554,"url":555},"Kusupati et al.: Matryoshka Representation Learning (2022)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2205.13147",{"title":557,"url":186},"pgvector README: halfvec, bit, binary quantization and subvector indexing","\u002Fimages\u002Fblog\u002Fembedding-quantization-binary-matryoshka\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fembedding-quantization-binary-matryoshka\u002Fog.jpg",[82,83,84],"Embedding-kvantálás: bináris, int8 és Matryoshka vektorok a visszakeresés minőségének elvesztése nélkül","Hogyan zsugorítja 4–32-szeresére a vektorindexet a bináris és int8 kvantálás és a Matryoshka-vágás, mennyibe kerül ez, és hogyan hozza vissza a rescoring.","Embedding-kvantálás: matrjoska babák sorban nagytól kicsiig, a legkisebb egy mérlegen áll.",{"slug":565,"published":566,"minutes":567,"category":8,"tags":568,"tagPages":573,"keywords":575,"about":585,"sources":589,"cover":608,"og":609,"expertise":80,"locales":610,"lang":84,"title":611,"description":612,"coverAlt":613},"zep","2026-10-06",11,[569,570,571,572,11],"Agent memory","Knowledge graph","Temporal graph","Context engineering",[574,8],"context-engineering",[576,577,578,579,580,581,582,583,584],"zep ai","zep teszt","zep ügynök memória","graphiti tudásgráf","zep árak","zep vagy mem0","ügynök memória hosszú távon","időbeli tudásgráf","zep cloud",[586],{"name":587,"url":588},"Zep","https:\u002F\u002Fwww.getzep.com\u002F",[590,593,596,599,602,605],{"title":591,"url":592},"Zep pricing: plans, credits and limits","https:\u002F\u002Fwww.getzep.com\u002Fpricing",{"title":594,"url":595},"Zep documentation","https:\u002F\u002Fhelp.getzep.com\u002F",{"title":597,"url":598},"Graphiti on GitHub","https:\u002F\u002Fgithub.com\u002Fgetzep\u002Fgraphiti",{"title":600,"url":601},"Graphiti product page","https:\u002F\u002Fwww.getzep.com\u002Fplatform\u002Fgraphiti\u002F",{"title":603,"url":604},"Announcing a new direction for Zep's open-source strategy","https:\u002F\u002Fwww.getzep.com\u002Fblog\u002Fannouncing-a-new-direction-for-zeps-open-source-strategy\u002F",{"title":606,"url":607},"Graphiti: temporal knowledge graphs for AI agents (arXiv)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2501.13956","\u002Fimages\u002Fblog\u002Fzep\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fzep\u002Fog.jpg",[82,83,84],"Zep ismertető: ügynök-memory temporális gráfon","A Zep hostolt memória-API temporális tudásgráffal: kreditek íráskor, lekérdezés ingyen, Flex 125 dollártól havonta, a Graphiti pedig önhosztolható.","Zep ismertető: robot egy falon lévő, összekötött kártyákból álló ábrát néz, mindegyiken kis óra.",{"slug":615,"published":616,"minutes":617,"category":8,"tags":618,"tagPages":622,"keywords":623,"about":634,"sources":641,"cover":678,"og":679,"expertise":80,"locales":680,"lang":84,"title":681,"description":682,"coverAlt":683},"graphrag-knowledge-graph-rag","2026-10-02",13,[619,620,621,11],"GraphRAG","Knowledge graphs","LightRAG",[8],[624,625,626,627,628,629,630,631,632,633],"graphrag magyarázat","tudásgráf rag","graphrag vs vektoros rag","microsoft graphrag magyarázat","lightrag vs graphrag","graphrag globális vs lokális keresés","graphrag indexelés ára","mikor használjunk graphrag-et","többlépéses rag","lazygraphrag",[635,637,638],{"name":570,"url":636},"https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FKnowledge_graph",{"name":110,"url":111},{"name":639,"url":640},"Leiden algorithm","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FLeiden_algorithm",[642,645,648,651,654,657,660,663,666,669,672,675],{"title":643,"url":644},"Edge et al.: From Local to Global: A Graph RAG Approach to Query-Focused Summarization (arXiv:2404.16130)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2404.16130",{"title":646,"url":647},"Microsoft GraphRAG documentation: overview","https:\u002F\u002Fmicrosoft.github.io\u002Fgraphrag\u002F",{"title":649,"url":650},"Microsoft GraphRAG documentation: default dataflow","https:\u002F\u002Fmicrosoft.github.io\u002Fgraphrag\u002Findex\u002Fdefault_dataflow\u002F",{"title":652,"url":653},"Microsoft GraphRAG documentation: global search","https:\u002F\u002Fmicrosoft.github.io\u002Fgraphrag\u002Fquery\u002Fglobal_search\u002F",{"title":655,"url":656},"Microsoft GraphRAG documentation: local search","https:\u002F\u002Fmicrosoft.github.io\u002Fgraphrag\u002Fquery\u002Flocal_search\u002F",{"title":658,"url":659},"Microsoft GraphRAG documentation: DRIFT search","https:\u002F\u002Fmicrosoft.github.io\u002Fgraphrag\u002Fquery\u002Fdrift_search\u002F",{"title":661,"url":662},"Microsoft GraphRAG documentation: getting started","https:\u002F\u002Fmicrosoft.github.io\u002Fgraphrag\u002Fget_started\u002F",{"title":664,"url":665},"Microsoft Research: LazyGraphRAG, setting a new standard for quality and cost (25 November 2024)","https:\u002F\u002Fwww.microsoft.com\u002Fen-us\u002Fresearch\u002Fblog\u002Flazygraphrag-setting-a-new-standard-for-quality-and-cost\u002F",{"title":667,"url":668},"Guo et al.: LightRAG: Simple and Fast Retrieval-Augmented Generation (arXiv:2410.05779)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2410.05779",{"title":670,"url":671},"HKUDS\u002FLightRAG on GitHub","https:\u002F\u002Fgithub.com\u002FHKUDS\u002FLightRAG",{"title":673,"url":674},"Gutiérrez et al.: HippoRAG: Neurobiologically Inspired Long-Term Memory for Large Language Models (arXiv:2405.14831)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2405.14831",{"title":676,"url":677},"Xiang et al.: When to use Graphs in RAG: A Comprehensive Analysis for Graph Retrieval-Augmented Generation (arXiv:2506.05690)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2506.05690","\u002Fimages\u002Fblog\u002Fgraphrag-knowledge-graph-rag\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fgraphrag-knowledge-graph-rag\u002Fog.jpg",[82,83,84],"GraphRAG és knowledge-graph RAG: mikor veri a gráf a vektoros keresést","Mit tud a Microsoft GraphRAG és a LightRAG, mibe kerül az indexelés, és mikor veri a tudásgráf a vektoros RAG-et többlépéses és globális kérdéseknél?","GraphRAG és tudásgráf-RAG: nagyító egy kereszteződés fölött egy várostérképen, mellette laza papírkupac.",{"slug":685,"published":686,"updated":6,"minutes":687,"category":8,"tags":688,"tagPages":692,"keywords":694,"about":705,"sources":715,"cover":751,"og":752,"expertise":80,"locales":753,"lang":84,"title":754,"description":755,"coverAlt":756},"code-rag-repository-indexing","2026-10-01",8,[11,689,690,10,691],"Code search","Tree-sitter","Coding agents",[8,16,693],"coding-agents",[695,696,697,698,699,700,701,702,703,704],"rag kódon","ast darabolás","tree-sitter kód darabolás","repótérkép","kód beágyazások","szemantikus kódkeresés","kódbázis indexelés","repo map tokenek","inkrementális kódindex","coir benchmark",[706,707,710,712],{"name":110,"url":111},{"name":708,"url":709},"Abstract syntax tree","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FAbstract_syntax_tree",{"name":690,"url":711},"https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FTree-sitter_%28parser_generator%29",{"name":713,"url":714},"Merkle tree","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FMerkle_tree",[716,719,722,725,728,731,734,737,740,742,745,748],{"title":717,"url":718},"cAST: AST-based chunking for code RAG (arXiv 2506.15655)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2506.15655",{"title":720,"url":721},"Aider docs: repository map","https:\u002F\u002Faider.chat\u002Fdocs\u002Frepomap.html",{"title":723,"url":724},"Aider blog: building a better repository map with tree-sitter (2023)","https:\u002F\u002Faider.chat\u002F2023\u002F10\u002F22\u002Frepomap.html",{"title":726,"url":727},"Anthropic Engineering: effective context engineering for AI agents (Sep 2025)","https:\u002F\u002Fwww.anthropic.com\u002Fengineering\u002Feffective-context-engineering-for-ai-agents",{"title":729,"url":730},"Cursor blog: securely indexing large codebases (Jan 2026)","https:\u002F\u002Fcursor.com\u002Fblog\u002Fsecure-codebase-indexing",{"title":732,"url":733},"turbopuffer: Cursor customer story","https:\u002F\u002Fturbopuffer.com\u002Fcustomers\u002Fcursor",{"title":735,"url":736},"Kilo Code docs: codebase indexing","https:\u002F\u002Fkilo.ai\u002Fdocs\u002Fcustomize\u002Fcontext\u002Fcodebase-indexing",{"title":738,"url":739},"Serena on GitHub","https:\u002F\u002Fgithub.com\u002Foraios\u002Fserena",{"title":741,"url":62},"Voyage AI docs: pricing",{"title":743,"url":744},"Qwen blog: Qwen3-Embedding (Jun 2025)","https:\u002F\u002Fqwenlm.github.io\u002Fblog\u002Fqwen3-embedding\u002F",{"title":746,"url":747},"CoIR: a benchmark for code information retrieval (arXiv 2407.02883)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2407.02883",{"title":749,"url":750},"CocoIndex blog: indexing a codebase","https:\u002F\u002Fcocoindex.io\u002Fblogs\u002Findex-codebase-v1","\u002Fimages\u002Fblog\u002Fcode-rag-repository-indexing\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fcode-rag-repository-indexing\u002Fog.jpg",[82,83,84],"RAG kódon: AST-darabolás, repótérképek és kódbeágyazások ügynökökhöz","Hogyan találják meg a kódot a kódoló ügynökök: AST-darabolás tree-sitterrel, repótérképek, Merkle-fás szinkron, kódbeágyazók 2026-ban és mikor jobb a grep.","RAG kódon (AST-darabolás): elágazó kódfa táplál egy sor kartotékkártyát.",{"slug":758,"published":759,"updated":686,"minutes":687,"category":8,"tags":760,"tagPages":764,"keywords":765,"about":776,"sources":784,"cover":819,"og":820,"expertise":80,"locales":821,"lang":84,"title":822,"description":823,"coverAlt":824},"reranking-models-compared","2026-09-30",[761,11,762,763,343],"Reranking","Search","Cross-encoders",[8,348],[766,767,768,769,770,771,772,773,774,775],"cross-encoder rerankerek összehasonlítása","cohere rerank 4 árak","bge-reranker-v2-m3","voyage rerank-3","jina reranker v3.5","colbert late interaction vs reranker","llm mint reranker","reciprocal rank fusion vs reranker","újrarangsorolás top-k kiértékelés","saját szerveres reranker onnx cpu",[777,780,783],{"name":778,"url":779},"Learning to rank","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FLearning_to_rank",{"name":781,"url":782},"BERT (language model)","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FBERT_%28language_model%29",{"name":36,"url":37},[785,788,791,794,797,800,803,804,807,810,813,816],{"title":786,"url":787},"Sentence-BERT (arXiv 1908.10084)","https:\u002F\u002Farxiv.org\u002Fabs\u002F1908.10084",{"title":789,"url":790},"Sentence Transformers: retrieve and re-rank","https:\u002F\u002Fsbert.net\u002Fexamples\u002Fsentence_transformer\u002Fapplications\u002Fretrieve_rerank\u002FREADME.html",{"title":792,"url":793},"BAAI\u002Fbge-reranker-v2-m3 model card","https:\u002F\u002Fhuggingface.co\u002FBAAI\u002Fbge-reranker-v2-m3",{"title":795,"url":796},"Cohere docs: reranking best practices","https:\u002F\u002Fdocs.cohere.com\u002Fdocs\u002Freranking-best-practices",{"title":798,"url":799},"Cohere pricing","https:\u002F\u002Fcohere.com\u002Fpricing",{"title":801,"url":802},"Voyage AI blog: rerank-3 (30 Sep 2026)","https:\u002F\u002Fblog.voyageai.com\u002F2026\u002F09\u002F30\u002Frerank-3\u002F",{"title":741,"url":62},{"title":805,"url":806},"Jina AI: Reranker v3.5 (3 Aug 2026)","https:\u002F\u002Fjina.ai\u002Fnews\u002Fjina-reranker-v3-5-faster-listwise-reranking-hybrid-attention-self-distillation\u002F",{"title":808,"url":809},"ColBERTv2 (arXiv 2112.01488)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2112.01488",{"title":811,"url":812},"Elasticsearch docs: reciprocal rank fusion","https:\u002F\u002Fwww.elastic.co\u002Fdocs\u002Freference\u002Felasticsearch\u002Frest-apis\u002Freciprocal-rank-fusion",{"title":814,"url":815},"Azure AI Search: semantic ranking overview","https:\u002F\u002Flearn.microsoft.com\u002Fen-us\u002Fazure\u002Fsearch\u002Fsemantic-search-overview",{"title":817,"url":818},"RankGPT (arXiv 2304.09542)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2304.09542","\u002Fimages\u002Fblog\u002Freranking-models-compared\u002Fcover.webp","\u002Fimages\u002Fblog\u002Freranking-models-compared\u002Fog.jpg",[82,83,84],"Cross-encoder rerankerek: Cohere, bge, Voyage, Jina és késői interakció (2026)","Rerankerek 2026 októberében: pointwise, listwise és késői interakció, Cohere, Voyage, bge, Qwen3, Jina és mxbai, árak, saját üzemeltetés és a k választása.","Cross-encoder rerankerek: egy tölcsér sok kártyát ötre szűkít, egy kártya előre emelkedik.",{"slug":826,"published":827,"updated":412,"minutes":687,"category":8,"tags":828,"tagPages":832,"keywords":833,"about":844,"sources":850,"cover":887,"og":888,"expertise":80,"locales":889,"lang":84,"title":890,"description":891,"coverAlt":892},"table-extraction-rag","2026-09-29",[11,829,830,343,831],"Tables","Document parsing","Text-to-SQL",[8,348],[834,835,836,837,838,839,840,841,842,843],"táblázatkinyerés rag-hez","táblázatok rag pipeline-okban","pdfplumber vs camelot","table transformer pubtables-1m","docling table mode tableformer","azure document intelligence táblázatok","teds metrika korlátai","markdown vs html táblázat llm-nek","text-to-sql táblázatkérdésekhez","llamaparse táblázatkinyerés",[845,846,849],{"name":110,"url":111},{"name":847,"url":848},"Table (information)","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FTable_%28information%29",{"name":433,"url":434},[851,854,857,860,863,866,869,872,875,878,881,884],{"title":852,"url":853},"Factorial study of parsers, chunkers and retrievers (arXiv 2609.31660)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.31660",{"title":855,"url":856},"pdfplumber on GitHub","https:\u002F\u002Fgithub.com\u002Fjsvine\u002Fpdfplumber",{"title":858,"url":859},"Camelot documentation","https:\u002F\u002Fcamelot-py.readthedocs.io\u002Fen\u002Flatest\u002F",{"title":861,"url":862},"Microsoft Table Transformer (archived)","https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Ftable-transformer",{"title":864,"url":865},"PubTabNet and TEDS (arXiv 1911.10683)","https:\u002F\u002Farxiv.org\u002Fabs\u002F1911.10683",{"title":867,"url":868},"Semantic evaluation of table extraction (arXiv 2603.18652)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2603.18652",{"title":870,"url":871},"TableFormer (arXiv 2203.01017)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2203.01017",{"title":873,"url":874},"Docling docs: advanced options","https:\u002F\u002Fdocling-project.github.io\u002Fdocling\u002Fusage\u002Fadvanced_options\u002F",{"title":876,"url":877},"Azure Document Intelligence: layout model","https:\u002F\u002Flearn.microsoft.com\u002Fen-us\u002Fazure\u002Fai-services\u002Fdocument-intelligence\u002Fprebuilt\u002Flayout",{"title":879,"url":880},"Table Meets LLM (WSDM 2024)","https:\u002F\u002Fzmy.io\u002Ffiles\u002Fwsdm24-SUC.pdf",{"title":882,"url":883},"TableRAG (arXiv 2410.04739)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2410.04739",{"title":885,"url":886},"Chain-of-Table (arXiv 2401.04398)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2401.04398","\u002Fimages\u002Fblog\u002Ftable-extraction-rag\u002Fcover.webp","\u002Fimages\u002Fblog\u002Ftable-extraction-rag\u002Fog.jpg",[82,83,84],"Táblázatkinyerés RAG-hoz: kinyerési rétegek, szerializálás és megbízható lekérdezések","Táblázatok RAG-ban: miért töri szét őket a darabolás, pdfplumber, Camelot, Docling és felhős kinyerők, mit nem lát a TEDS, szerializálás és SQL-irányítás.","Táblázatkinyerés RAG-hoz: egy oldalon lévő táblázatból különálló sorcsíkok lesznek, mindegyiken fejlécfüllel.",{"slug":894,"published":895,"updated":6,"minutes":687,"category":8,"tags":896,"tagPages":899,"keywords":900,"about":911,"sources":917,"cover":951,"og":952,"expertise":80,"locales":953,"lang":84,"title":954,"description":955,"coverAlt":956},"embedding-fine-tuning-domain","2026-09-28",[10,897,11,343,898],"Fine-tuning","Sentence Transformers",[16,8,348],[901,902,903,904,905,906,907,908,909,910],"beágyazó modellek finomhangolása","domain adaptation beágyazások","hard negative mining visszakeresés","multiplenegativesrankingloss","gistembed vs mnrl","matryoshka beágyazás dimenziók","qwen3-embedding finomhangolás","bge-m3 vs embeddinggemma","beágyazás kiértékelés recall@k","katasztrofális felejtés dense retrieval",[912,913,916],{"name":33,"url":34},{"name":914,"url":915},"Fine-tuning (deep learning)","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FFine-tuning_%28deep_learning%29",{"name":36,"url":37},[918,921,924,927,930,933,936,937,940,943,946,948],{"title":919,"url":920},"Sentence Transformers: training overview","https:\u002F\u002Fsbert.net\u002Fdocs\u002Fsentence_transformer\u002Ftraining_overview.html",{"title":922,"url":923},"Sentence Transformers: loss overview","https:\u002F\u002Fsbert.net\u002Fdocs\u002Fsentence_transformer\u002Floss_overview.html",{"title":925,"url":926},"Sentence Transformers: hard negative mining","https:\u002F\u002Fsbert.net\u002Fdocs\u002Fpackage_reference\u002Futil\u002Fhard_negatives.html",{"title":928,"url":929},"RocketQA (arXiv 2010.08191)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2010.08191",{"title":931,"url":932},"InPars (arXiv 2202.05144)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2202.05144",{"title":934,"url":935},"Promptagator (arXiv 2209.11755)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2209.11755",{"title":743,"url":744},{"title":938,"url":939},"BAAI\u002Fbge-m3 model card","https:\u002F\u002Fhuggingface.co\u002FBAAI\u002Fbge-m3",{"title":941,"url":942},"Google: EmbeddingGemma docs (Oct 2026)","https:\u002F\u002Fai.google.dev\u002Fgemma\u002Fdocs\u002Fembeddinggemma",{"title":944,"url":945},"Sentence Transformers: Matryoshka embeddings","https:\u002F\u002Fsbert.net\u002Fexamples\u002Fsentence_transformer\u002Ftraining\u002Fmatryoshka\u002FREADME.html",{"title":947,"url":44},"MMTEB (arXiv 2502.13595)",{"title":949,"url":950},"ANCE-Tele (arXiv 2210.17167)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2210.17167","\u002Fimages\u002Fblog\u002Fembedding-fine-tuning-domain\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fembedding-fine-tuning-domain\u002Fog.jpg",[82,83,84],"Beágyazó modellek finomhangolása: nehéz negatívok, veszteségek és kiértékelés","Beágyazások finomhangolása: mikor jobb a rerankernél, tanítópárok, nehéz negatívok, veszteségek a sentence-transformers v6-ban, Matryoshka és költség.","Beágyazó modellek finomhangolása: a bal oldali laza pontokból tanítás után szoros csoportok lesznek jobbra.",{"slug":958,"published":959,"updated":960,"minutes":687,"category":8,"tags":961,"tagPages":964,"keywords":965,"about":976,"sources":984,"cover":1019,"og":1020,"expertise":80,"locales":1021,"lang":84,"title":1022,"description":1023,"coverAlt":1024},"knowledge-graph-extraction-llm","2026-09-25","2026-10-12",[620,619,962,11,963],"Entity resolution","Neo4j",[8],[966,967,968,969,970,971,972,973,974,975],"tudásgráf kinyerése llm-ekkel","graphrag indexelés","entitásfeloldás tudásgráf","hármas kinyerő prompt","ontológia tervezés llm kinyeréshez","entitás duplikáció szűrés beágyazásokkal","tudásgráf eredetkövetés","növekményes tudásgráf frissítés","graphrag költség","neo4j simplekgpipeline",[977,978,981],{"name":570,"url":636},{"name":979,"url":980},"Record linkage","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FRecord_linkage",{"name":982,"url":983},"Ontology (information science)","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FOntology_%28information_science%29",[985,988,991,993,996,999,1002,1005,1008,1011,1014,1016],{"title":986,"url":987},"Microsoft GraphRAG: default dataflow","https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Fgraphrag\u002Fblob\u002Fmain\u002Fdocs\u002Findex\u002Fdefault_dataflow.md",{"title":989,"url":990},"Microsoft GraphRAG on GitHub","https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Fgraphrag",{"title":992,"url":665},"Microsoft Research: LazyGraphRAG (Nov 2024)",{"title":994,"url":995},"Microsoft Research: GraphRAG auto-tuning (Sep 2024)","https:\u002F\u002Fwww.microsoft.com\u002Fen-us\u002Fresearch\u002Fblog\u002Fgraphrag-auto-tuning-provides-rapid-adaptation-to-new-domains\u002F",{"title":997,"url":998},"neo4j-graphrag-python on GitHub","https:\u002F\u002Fgithub.com\u002Fneo4j\u002Fneo4j-graphrag-python",{"title":1000,"url":1001},"Neo4j GraphRAG: knowledge graph builder guide","https:\u002F\u002Fneo4j.com\u002Fdocs\u002Fneo4j-graphrag-python\u002Fcurrent\u002Fuser_guide_kg_builder.html",{"title":1003,"url":1004},"Splink on GitHub","https:\u002F\u002Fgithub.com\u002Fmoj-analytical-services\u002Fsplink",{"title":1006,"url":1007},"Google LangExtract on GitHub","https:\u002F\u002Fgithub.com\u002Fgoogle\u002Flangextract",{"title":1009,"url":1010},"KGGen (arXiv 2502.09956)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2502.09956",{"title":1012,"url":1013},"GraphEval (arXiv 2407.10793)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2407.10793",{"title":1015,"url":607},"Zep temporal knowledge graph (arXiv 2501.13956)",{"title":1017,"url":1018},"Kùzu on GitHub","https:\u002F\u002Fgithub.com\u002Fkuzudb\u002Fkuzu","\u002Fimages\u002Fblog\u002Fknowledge-graph-extraction-llm\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fknowledge-graph-extraction-llm\u002Fog.jpg",[82,83,84],"Tudásgráf kinyerése LLM-ekkel: ontológia, entitásfeloldás, eredet","Tudásgráf építése LLM-ekkel: minimális ontológia, kinyerő promptok, entitásfeloldás, eredetkövetés, növekményes frissítés, kiértékelés és tárolás.","Tudásgráf kinyerése LLM-ekkel: csomópontok lépnek ki egy lapkötegből, hálót alkotva, ahol két kör eggyé olvad.",{"slug":1026,"published":959,"updated":6,"minutes":687,"category":8,"tags":1027,"tagPages":1032,"keywords":1033,"about":1044,"sources":1051,"cover":1085,"og":1086,"expertise":80,"locales":1087,"lang":84,"title":1088,"description":1089,"coverAlt":1090},"incremental-indexing-cdc",[11,1028,1029,1030,1031],"Change data capture","PostgreSQL","Vector databases","Data freshness",[8],[1034,1035,1036,1037,1038,1039,1040,1041,1042,1043],"növekményes indexelés rag-hez","change data capture rag","debezium postgres logical decoding","újrabeágyazás költsége","blue\u002Fgreen indexcsere vektoradatbázis","index elavulás monitorozás","törlések vektorindexben","jogosultságváltozások rag indexben","qdrant aliasok","high water mark változásfelismerés",[1045,1047,1048],{"name":1028,"url":1046},"https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FChange_data_capture",{"name":110,"url":111},{"name":1049,"url":1050},"Blue–green deployment","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FBlue%E2%80%93green_deployment",[1052,1055,1058,1061,1064,1067,1070,1073,1076,1078,1081,1084],{"title":1053,"url":1054},"PostgreSQL docs: logical decoding concepts","https:\u002F\u002Fwww.postgresql.org\u002Fdocs\u002Fcurrent\u002Flogicaldecoding-explanation.html",{"title":1056,"url":1057},"Debezium docs: PostgreSQL connector","https:\u002F\u002Fdebezium.io\u002Fdocumentation\u002Freference\u002Fstable\u002Fconnectors\u002Fpostgresql.html",{"title":1059,"url":1060},"PostgreSQL docs: CREATE PUBLICATION","https:\u002F\u002Fwww.postgresql.org\u002Fdocs\u002Fcurrent\u002Fsql-createpublication.html",{"title":1062,"url":1063},"PostgreSQL docs: pg_replication_slots","https:\u002F\u002Fwww.postgresql.org\u002Fdocs\u002Fcurrent\u002Fview-pg-replication-slots.html",{"title":1065,"url":1066},"Azure AI Search: changed and deleted blobs","https:\u002F\u002Flearn.microsoft.com\u002Fen-us\u002Fazure\u002Fsearch\u002Fsearch-howto-index-changed-deleted-blobs",{"title":1068,"url":1069},"Azure AI Search: scheduling indexers","https:\u002F\u002Flearn.microsoft.com\u002Fen-us\u002Fazure\u002Fsearch\u002Fsearch-howto-schedule-indexers",{"title":1071,"url":1072},"Qdrant docs: collections and aliases","https:\u002F\u002Fqdrant.tech\u002Fdocumentation\u002Fconcepts\u002Fcollections\u002F",{"title":1074,"url":1075},"Elastic docs: aliases","https:\u002F\u002Fwww.elastic.co\u002Fdocs\u002Fmanage-data\u002Fdata-store\u002Faliases",{"title":1077,"url":186},"pgvector on GitHub",{"title":1079,"url":1080},"GDPR Article 17","https:\u002F\u002Fgdpr-info.eu\u002Fart-17-gdpr\u002F",{"title":1082,"url":1083},"OpenAI API pricing","https:\u002F\u002Fplatform.openai.com\u002Fdocs\u002Fpricing",{"title":741,"url":62},"\u002Fimages\u002Fblog\u002Fincremental-indexing-cdc\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fincremental-indexing-cdc\u002Fog.jpg",[82,83,84],"A RAG-index frissen tartása: növekményes indexelés, CDC és blue\u002Fgreen újrabeágyazás","A RAG-index frissen tartása: elavulási keret, változáskövetés, CDC Postgresből Debeziummal, törlések, jogosultságok, blue\u002Fgreen csere és költség.","RAG-index frissen tartása: adatbázis változásokat továbbít egy szekrénybe, mellette egy második szekrény és egy kapcsoló.",{"slug":1092,"published":1093,"updated":1094,"minutes":7,"category":8,"tags":1095,"tagPages":1098,"keywords":1100,"about":1108,"sources":1115,"cover":1140,"og":1141,"expertise":80,"locales":1142,"lang":84,"title":1143,"description":1144,"coverAlt":1145},"lancedb","2026-09-21","2026-10-07",[13,1096,1097,11],"Hybrid search","Embedded database",[17,1099,8],"hybrid-search",[1092,1101,1102,1103,1104,1105,1106,1107],"lancedb teszt","lance vektoradatbázis","beágyazott vektoradatbázis","lancedb vagy qdrant","hibrid keresés rrf","lancedb indexelés","lance adatformátum",[1109,1112],{"name":1110,"url":1111},"Vector database","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FVector_database",{"name":1113,"url":1114},"Apache Arrow","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FApache_Arrow",[1116,1119,1122,1125,1128,1131,1134,1137],{"title":1117,"url":1118},"LanceDB quickstart","https:\u002F\u002Fdocs.lancedb.com\u002Fquickstart",{"title":1120,"url":1121},"LanceDB vector indexes","https:\u002F\u002Fdocs.lancedb.com\u002Findexing\u002Fvector-index",{"title":1123,"url":1124},"LanceDB indexing guide","https:\u002F\u002Fdocs.lancedb.com\u002Findexing\u002Findex",{"title":1126,"url":1127},"LanceDB hybrid search","https:\u002F\u002Fdocs.lancedb.com\u002Fsearch\u002Fhybrid-search",{"title":1129,"url":1130},"LanceDB Enterprise","https:\u002F\u002Fdocs.lancedb.com\u002Fenterprise",{"title":1132,"url":1133},"LanceDB frequently asked questions","https:\u002F\u002Fdocs.lancedb.com\u002Ffaq\u002Ffaq-oss",{"title":1135,"url":1136},"LanceDB pricing","https:\u002F\u002Flancedb.com\u002Fpricing",{"title":1138,"url":1139},"LanceDB on PyPI","https:\u002F\u002Fpypi.org\u002Fproject\u002Flancedb\u002F","\u002Fimages\u002Fblog\u002Flancedb\u002Fcover.webp","\u002Fimages\u002Fblog\u002Flancedb\u002Fog.jpg",[82,83,84],"LanceDB: vektoros keresés, amely könyvtárként indul","A LanceDB értékelése: beágyazott, Apache-2.0 vektorkönyvtár, IVF és HNSW indexek, hibrid keresés rangsoregyesítéssel, és mit ad az Enterprise.","LanceDB: egy nyitott könyvből golyókkal teli, szétválogatott polcok nyílnak ki.",{"slug":526,"published":1093,"updated":686,"minutes":90,"category":8,"tags":1147,"tagPages":1151,"keywords":1152,"about":1160,"sources":1164,"cover":1179,"og":1180,"expertise":80,"locales":1181,"lang":84,"title":1182,"description":1183,"coverAlt":1184},[13,1148,1149,11,1150],"Postgres","HNSW","Quantisation",[17,8],[526,1153,1154,1155,1156,1157,1158,1159],"pgvector teszt","pgvector vagy qdrant","postgres vektorkeresés","hnsw index postgres","iteratív indexszkennelés","bináris kvantálás postgres","vektoradatbázis postgres",[1161,1162],{"name":526,"url":186},{"name":1029,"url":1163},"https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FPostgreSQL",[1165,1167,1170,1173,1176],{"title":1166,"url":186},"pgvector README",{"title":1168,"url":1169},"pgvector changelog","https:\u002F\u002Fgithub.com\u002Fpgvector\u002Fpgvector\u002Fblob\u002Fmaster\u002FCHANGELOG.md",{"title":1171,"url":1172},"PostgreSQL news: pgvector 0.8.2 released","https:\u002F\u002Fwww.postgresql.org\u002Fabout\u002Fnews\u002Fpgvector-082-released-3245\u002F",{"title":1174,"url":1175},"AWS: Scale pgvector with binary quantization","https:\u002F\u002Faws.amazon.com\u002Fblogs\u002Fdatabase\u002Fscale-pgvector-with-binary-quantization-on-amazon-aurora-postgresql\u002F",{"title":1177,"url":1178},"pgvector licence","https:\u002F\u002Fgithub.com\u002Fpgvector\u002Fpgvector\u002Fblob\u002Fmaster\u002FLICENSE","\u002Fimages\u002Fblog\u002Fpgvector\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fpgvector\u002Fog.jpg",[82,83,84],"pgvector, értékelés: az a vektordatázis, amelyet nem kell üzemeltetned","pgvector 0.8.7: iteratív szűrt keresés, HNSW és IVFFlat, bináris kvantálás 100 millió vektornál, és a CVE, ami javítandóvá tette az indexépítést.","pgvector értékelés: egy elefánt kis fiókos szekrényt visz a hátán, benne szétválogatott golyók.",{"slug":1186,"published":1187,"updated":686,"minutes":90,"category":8,"tags":1188,"tagPages":1190,"keywords":1191,"about":1198,"sources":1202,"cover":1230,"og":1231,"expertise":80,"locales":1232,"lang":84,"title":1233,"description":1234,"coverAlt":1235},"mem0","2026-09-17",[569,1189,11,13],"Long-term memory",[8,17],[1186,1192,1193,1194,1195,1196,1197],"mem0 teszt","ügynök memória réteg","mem0 saját szerveren","mem0 árak","hosszú távú memória ügynökökhöz","mem0 alternatíva",[1199],{"name":1200,"url":1201},"Mem0","https:\u002F\u002Fmem0.ai",[1203,1206,1209,1212,1215,1218,1221,1224,1227],{"title":1204,"url":1205},"Mem0 documentation","https:\u002F\u002Fdocs.mem0.ai\u002Fintroduction",{"title":1207,"url":1208},"Mem0 quickstart","https:\u002F\u002Fdocs.mem0.ai\u002Fquickstart",{"title":1210,"url":1211},"How Mem0 works","https:\u002F\u002Fdocs.mem0.ai\u002Fcore-concepts\u002Fhow-it-works",{"title":1213,"url":1214},"Mem0 pricing","https:\u002F\u002Fmem0.ai\u002Fpricing",{"title":1216,"url":1217},"Mem0 on GitHub","https:\u002F\u002Fgithub.com\u002Fmem0ai\u002Fmem0",{"title":1219,"url":1220},"mem0ai on PyPI","https:\u002F\u002Fpypi.org\u002Fproject\u002Fmem0ai\u002F",{"title":1222,"url":1223},"Mem0 research and benchmarks","https:\u002F\u002Fmem0.ai\u002Fresearch",{"title":1225,"url":1226},"Mem0 MCP server","https:\u002F\u002Fdocs.mem0.ai\u002Fplatform\u002Fmem0-mcp",{"title":1228,"url":1229},"Mem0 paper on arXiv","https:\u002F\u002Farxiv.org\u002Fabs\u002F2504.19413","\u002Fimages\u002Fblog\u002Fmem0\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fmem0\u002Fog.jpg",[82,83,84],"Mem0: mennyibe kerül egy ügynökmemória körönként","A Mem0 elemzése: körönként kinyert tények, a 2026. áprilisi benchmark-tábla platformos fenntartása, négy felhőcsomag és az önhosztolásból kimaradók.","Mem0 költsége körönként: egy robot jegyzetel egy parkolóóra mellett, mögötte cetlik sora.",{"slug":1237,"published":1238,"updated":1239,"minutes":90,"category":8,"tags":1240,"tagPages":1243,"keywords":1244,"about":1253,"sources":1261,"cover":1280,"og":1281,"expertise":80,"locales":1282,"lang":84,"title":1283,"description":1284,"coverAlt":1285},"milvus-zilliz","2026-09-08","2026-09-20",[13,1096,1241,1242,11],"BM25 full text","Distributed",[17,1099,8],[1245,1246,1247,1248,1249,1250,1251,1252],"milvus","milvus vagy qdrant","zilliz cloud árak","vektoradatbázis összehasonlítás","milvus hibrid keresés","apache milvus saját szerveren","milvus 3.0 teszt","rag vektortár",[1254,1257,1260],{"name":1255,"url":1256},"Milvus","https:\u002F\u002Fmilvus.io",{"name":1258,"url":1259},"Zilliz Cloud","https:\u002F\u002Fzilliz.com",{"name":110,"url":111},[1262,1265,1268,1271,1274,1277],{"title":1263,"url":1264},"Milvus architecture overview","https:\u002F\u002Fmilvus.io\u002Fdocs\u002Farchitecture_overview.md",{"title":1266,"url":1267},"Milvus release notes","https:\u002F\u002Fmilvus.io\u002Fdocs\u002Frelease_notes.md",{"title":1269,"url":1270},"Milvus releases on GitHub","https:\u002F\u002Fgithub.com\u002Fmilvus-io\u002Fmilvus\u002Freleases",{"title":1272,"url":1273},"Milvus README: features and licence","https:\u002F\u002Fgithub.com\u002Fmilvus-io\u002Fmilvus",{"title":1275,"url":1276},"Zilliz Cloud pricing","https:\u002F\u002Fzilliz.com\u002Fpricing",{"title":1278,"url":1279},"Zilliz Cloud list price","https:\u002F\u002Fzilliz.com\u002Fpricing\u002Fpricing-guide","\u002Fimages\u002Fblog\u002Fmilvus-zilliz\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fmilvus-zilliz\u002Fog.jpg",[82,83,84],"Milvus elemzés: a legteljesebb vektordatbázis üzemeltetve","Milvus 3.0.2: a legteljesebb nyílt forráskódú vektoradatbázis és a legnehezebben üzemeltethető. Architektúra, hibrid keresés, költségek és korlátok.","Milvus és Zilliz: magas polcos raktár targoncával, egy dolgozó csavarkulccsal a vezérlőpultnál.",{"slug":1287,"published":1238,"updated":616,"minutes":90,"category":8,"tags":1288,"tagPages":1292,"keywords":1293,"about":1300,"sources":1304,"cover":1319,"og":1320,"expertise":80,"locales":1321,"lang":84,"title":1322,"description":1323,"coverAlt":1324},"graphrag",[570,11,1289,1290,1291],"Global search","Community summaries","Token cost",[8],[1287,1294,1295,1296,1297,1298,625,1299],"microsoft graphrag","graphrag vagy vektoros rag","graphrag indexelési költség","graphrag lekérdezési módok","graphrag vs lightrag összehasonlítás","graphrag dinamikus közösségválasztás",[1301,1303],{"name":619,"url":1302},"https:\u002F\u002Fgithub.com\u002Fmicrosoft\u002Fgraphrag\u002F",{"name":110,"url":111},[1305,1307,1309,1311,1314,1316],{"title":1306,"url":990},"GraphRAG on GitHub",{"title":1308,"url":647},"GraphRAG documentation",{"title":1310,"url":644},"From Local to Global: A Graph RAG Approach to Query-Focused Summarization",{"title":1312,"url":1313},"GraphRAG: Improving global search via dynamic community selection","https:\u002F\u002Fwww.microsoft.com\u002Fen-us\u002Fresearch\u002Fblog\u002Fgraphrag-improving-global-search-via-dynamic-community-selection\u002F",{"title":1315,"url":665},"LazyGraphRAG: Setting a new standard for quality and cost",{"title":1317,"url":1318},"Graph RAG in 2026: What Actually Works in Production","https:\u002F\u002Fwww.paperclipped.de\u002Fen\u002Fblog\u002Fgraph-rag-production\u002F","\u002Fimages\u002Fblog\u002Fgraphrag\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fgraphrag\u002Fog.jpg",[82,83,84],"Microsoft GraphRAG: tudásgráfos RAG, előre kifizetve","A Microsoft GraphRAG 3.2.0 tesztje: MIT, karbantartási mód és egy indexelési számla, amely az első lekérdezés előtt eldől.","Microsoft GraphRAG: parafatábla fonalakkal összekötött kártyákkal, a sarkán árcédula.",{"slug":1326,"published":1327,"updated":1328,"minutes":617,"category":8,"tags":1329,"tagPages":1334,"keywords":1335,"about":1346,"sources":1354,"cover":1394,"og":1395,"expertise":80,"locales":1396,"lang":84,"title":1397,"description":1398,"coverAlt":1399},"llm-hallucination-grounding-citations","2026-08-20","2026-09-01",[1330,11,1331,1332,1333],"Hallucinations","Citations","Grounding","Faithfulness",[8],[1336,1337,1338,1339,1340,1341,1342,1343,1344,1345],"llm hallucinációk csökkentése éles rendszerben","hogyan csökkentsük a hallucinációt rag-ben","llm citations api","anthropic citations api","rag hűségmetrika","llm tartózkodás nem tudom válasz","állításszintű ellenőrzés llm","llm válaszok megalapozása forrásokban","grounding ellenőrző api","források megjelenítése ai chatbot felületen",[1347,1350,1351],{"name":1348,"url":1349},"Hallucination (artificial intelligence)","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FHallucination_(artificial_intelligence)",{"name":110,"url":111},{"name":1352,"url":1353},"Large language model","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FLarge_language_model",[1355,1358,1361,1364,1367,1370,1373,1376,1379,1382,1385,1388,1391],{"title":1356,"url":1357},"Anthropic: Citations (Claude API documentation)","https:\u002F\u002Fplatform.claude.com\u002Fdocs\u002Fen\u002Fbuild-with-claude\u002Fcitations",{"title":1359,"url":1360},"Anthropic: Search results (Claude API documentation)","https:\u002F\u002Fplatform.claude.com\u002Fdocs\u002Fen\u002Fbuild-with-claude\u002Fsearch-results",{"title":1362,"url":1363},"Anthropic: Reduce hallucinations (Claude API documentation)","https:\u002F\u002Fplatform.claude.com\u002Fdocs\u002Fen\u002Ftest-and-evaluate\u002Fstrengthen-guardrails\u002Freduce-hallucinations",{"title":1365,"url":1366},"Anthropic: Introducing Citations on the Anthropic API","https:\u002F\u002Fclaude.com\u002Fblog\u002Fintroducing-citations-api",{"title":1368,"url":1369},"Simon Willison: Anthropic's new Citations API (24 January 2025)","https:\u002F\u002Fsimonwillison.net\u002F2025\u002FJan\u002F24\u002Fanthropics-new-citations-api\u002F",{"title":1371,"url":1372},"OpenAI: Web search guide (url_citation annotations and display requirement)","https:\u002F\u002Fdevelopers.openai.com\u002Fapi\u002Fdocs\u002Fguides\u002Ftools-web-search",{"title":1374,"url":1375},"Cohere: Documents and citations","https:\u002F\u002Fdocs.cohere.com\u002Fdocs\u002Fdocuments-and-citations",{"title":1377,"url":1378},"Google Cloud: Check grounding API","https:\u002F\u002Fdocs.cloud.google.com\u002Fgenerative-ai-app-builder\u002Fdocs\u002Fcheck-grounding",{"title":1380,"url":1381},"AWS: Amazon Bedrock Guardrails contextual grounding check","https:\u002F\u002Fdocs.aws.amazon.com\u002Fbedrock\u002Flatest\u002Fuserguide\u002Fguardrails-contextual-grounding-check.html",{"title":1383,"url":1384},"Ragas: Faithfulness metric","https:\u002F\u002Fdocs.ragas.io\u002Fen\u002Fstable\u002Fconcepts\u002Fmetrics\u002Favailable_metrics\u002Ffaithfulness\u002F",{"title":1386,"url":1387},"Kalai, Nachum, Vempala, Zhang: Why Language Models Hallucinate (arXiv 2509.04664)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2509.04664",{"title":1389,"url":1390},"Magesh et al.: Hallucination-Free? Assessing the Reliability of Leading AI Legal Research Tools (arXiv 2405.20362)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2405.20362",{"title":1392,"url":1393},"Wallat, Heuss, de Rijke, Anand: Correctness is not Faithfulness in RAG Attributions (arXiv 2412.18004)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2412.18004","\u002Fimages\u002Fblog\u002Fllm-hallucination-grounding-citations\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fllm-hallucination-grounding-citations\u002Fog.jpg",[82,83,84],"LLM-hallucinációk csökkentése éles rendszerben: grounding, hivatkozások és a nemet mondás tudománya","Hallucinációk csökkentése éles RAG-ban: hivatkozás-API-k, tartózkodás, állításszintű ellenőrzés, faithfulness-mérés, forrás-UI és ami mégis átcsúszik.","LLM-hallucinációk csökkentése: hőlégballont kötelek rögzítenek egy könyvkupachoz.",{"slug":1401,"published":1402,"updated":1403,"minutes":687,"category":8,"tags":1404,"tagPages":1407,"keywords":1408,"about":1417,"sources":1421,"cover":1438,"og":1439,"expertise":80,"locales":1440,"lang":84,"title":1441,"description":1442,"coverAlt":1443},"rag-2026-hybrid-agentic-long-context","2026-07-30","2026-09-22",[11,1405,1406,1096,93],"Long context","Agentic search",[8,1099],[1409,1410,1411,1412,1413,1414,1415,1416],"rag 2026-ban","hosszú kontextus vs rag","agentikus rag","hibrid keresés","contextual retrieval rag 2026","meghalt a rag","ügynöki keresés vs vektorkeresés","rag vs 1 millió tokenes kontextablak",[1418,1419,1420],{"name":110,"url":111},{"name":1352,"url":1353},{"name":113,"url":114},[1422,1424,1427,1430,1432,1435],{"title":1423,"url":121},"Anthropic: Introducing Contextual Retrieval",{"title":1425,"url":1426},"Chroma: Context Rot","https:\u002F\u002Fwww.trychroma.com\u002Fresearch\u002Fcontext-rot",{"title":1428,"url":1429},"Liu et al., Lost in the Middle (2023)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2307.03172",{"title":1431,"url":139},"Claude docs: Pricing",{"title":1433,"url":1434},"Claude docs: Models overview","https:\u002F\u002Fplatform.claude.com\u002Fdocs\u002Fen\u002Fabout-claude\u002Fmodels\u002Foverview",{"title":1436,"url":1437},"Awesome Agentic Patterns: Agentic search over vector embeddings","https:\u002F\u002Fgithub.com\u002Fnibzard\u002Fawesome-agentic-patterns\u002Fblob\u002Fmain\u002Fpatterns\u002Fagentic-search-over-vector-embeddings.md","\u002Fimages\u002Fblog\u002Frag-2026-hybrid-agentic-long-context\u002Fcover.webp","\u002Fimages\u002Fblog\u002Frag-2026-hybrid-agentic-long-context\u002Fog.jpg",[82,83,84],"RAG 2026-ban: hibrid retrieval, ügynöki keresés vagy egymillió tokenes kontextus?","RAG 2026-ban: mikor nyer a hibrid keresés, mikor illik az ügynöki keresés, és mit ér az egymillió tokenes kontextus kérésenként.","RAG 2026-ban: egy ember három út közül választ: könyvespolc, zseblámpás robot és hosszú tekercs.",{"slug":1445,"published":1446,"updated":1402,"minutes":90,"category":8,"tags":1447,"tagPages":1448,"keywords":1449,"about":1459,"sources":1463,"cover":1492,"og":1493,"expertise":80,"locales":1494,"lang":84,"title":1495,"description":1496,"coverAlt":1497},"rag-pipeline-chunking-hybrid-search-reranking","2026-07-24",[11,1096,92,761,526],[8,1099],[1450,1451,1452,1453,1454,1455,1456,1457,1458],"rag pipeline","visszakereséssel bővített generálás pipeline","chunkelési stratégia rag-hez","hibrid keresés bm25 vektor","reciprocal rank fusion","cross-encoder újrarangsorolás","pgvector hibrid keresés","rag minőségmérés","hogyan építsünk rag pipeline-t",[1460,1461,1462],{"name":110,"url":111},{"name":113,"url":114},{"name":526,"url":186},[1464,1467,1470,1473,1474,1477,1480,1483,1484,1487,1488,1490],{"title":1465,"url":1466},"Lewis et al., Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks (2020)","https:\u002F\u002Farxiv.org\u002Fabs\u002F2005.11401",{"title":1468,"url":1469},"Cormack, Clarke and Büttcher, Reciprocal Rank Fusion (SIGIR 2009)","https:\u002F\u002Fcormack.uwaterloo.ca\u002Fcormacksigir09-rrf.pdf",{"title":1471,"url":1472},"Malkov and Yashunin, HNSW graphs","https:\u002F\u002Farxiv.org\u002Fabs\u002F1603.09320",{"title":1166,"url":186},{"title":1475,"url":1476},"Elasticsearch: similarity settings","https:\u002F\u002Fwww.elastic.co\u002Fdocs\u002Freference\u002Felasticsearch\u002Findex-settings\u002Fsimilarity",{"title":1478,"url":1479},"PostgreSQL: controlling text search","https:\u002F\u002Fwww.postgresql.org\u002Fdocs\u002Fcurrent\u002Ftextsearch-controls.html",{"title":1481,"url":1482},"Chroma: Evaluating Chunking Strategies for Retrieval","https:\u002F\u002Fwww.trychroma.com\u002Fresearch\u002Fevaluating-chunking",{"title":1423,"url":121},{"title":1485,"url":1486},"Sentence Transformers: Cross-Encoders","https:\u002F\u002Fsbert.net\u002Fexamples\u002Fcross_encoder\u002Fapplications\u002FREADME.html",{"title":1428,"url":1429},{"title":1489,"url":1357},"Claude docs: Citations",{"title":1491,"url":1384},"RAGAS: Faithfulness","\u002Fimages\u002Fblog\u002Frag-pipeline-chunking-hybrid-search-reranking\u002Fcover.webp","\u002Fimages\u002Fblog\u002Frag-pipeline-chunking-hybrid-search-reranking\u002Fog.jpg",[82,83,84],"Termelési RAG pipeline lépésről lépésre: chunkelés, hibrid keresés, újrarangsorolás","Építs RAG pipeline-t lépésről lépésre: parse, chunkelés, pgvector, BM25 plusz vektorok RRF-fel, újrarangsorolás, hivatkozások és retrieval evalek.","Termelési RAG pipeline: egy kenyeret felszeletelnek, a szeletek szitán át rendezett sorban kerülnek a tányérra.",{"slug":1499,"published":1500,"minutes":90,"category":8,"tags":1501,"tagPages":1503,"keywords":1504,"about":1512,"sources":1519,"cover":1547,"og":1548,"expertise":80,"locales":1549,"lang":84,"title":1550,"description":1551,"coverAlt":1552},"chroma","2026-07-20",[13,10,11,1096,1502],"Metadata filter",[17,16,8,1099],[1505,1506,1507,1508,1509,1510,1511],"chroma vektoradatbázis","chroma vs qdrant összehasonlítás","chroma vagy pgvector","hibrid keresés reciprocal rank fusion","saját szerveren futó vektoradatbázis","chroma cloud árak","metaadat szűrés vektorkeresés",[1513,1516],{"name":1514,"url":1515},"Chroma","https:\u002F\u002Fgithub.com\u002Fchroma-core\u002Fchroma",{"name":1517,"url":1518},"Apache License 2.0","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FApache_License",[1520,1523,1526,1529,1532,1535,1538,1541,1544],{"title":1521,"url":1522},"Chroma documentation: introduction","https:\u002F\u002Fdocs.trychroma.com\u002Fdocs\u002Foverview\u002Fintroduction",{"title":1524,"url":1525},"Chroma architecture overview","https:\u002F\u002Fdocs.trychroma.com\u002Freference\u002Farchitecture\u002Foverview",{"title":1527,"url":1528},"Chroma: getting started with the SDK","https:\u002F\u002Fdocs.trychroma.com\u002Fdocs\u002Foverview\u002Fgetting-started",{"title":1530,"url":1531},"Chroma: query and get","https:\u002F\u002Fdocs.trychroma.com\u002Fdocs\u002Fquerying-collections\u002Fquery-and-get",{"title":1533,"url":1534},"Chroma: metadata filtering","https:\u002F\u002Fdocs.trychroma.com\u002Fdocs\u002Fquerying-collections\u002Fmetadata-filtering",{"title":1536,"url":1537},"Chroma Cloud: Search API overview","https:\u002F\u002Fdocs.trychroma.com\u002Fcloud\u002Fsearch-api\u002Foverview",{"title":1539,"url":1540},"Chroma Cloud: hybrid search with RRF","https:\u002F\u002Fdocs.trychroma.com\u002Fcloud\u002Fsearch-api\u002Fhybrid-search",{"title":1542,"url":1543},"Chroma Cloud: pricing","https:\u002F\u002Fwww.trychroma.com\u002Fpricing",{"title":1545,"url":1546},"Chroma changelog","https:\u002F\u002Fwww.trychroma.com\u002Fchangelog","\u002Fimages\u002Fblog\u002Fchroma\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fchroma\u002Fog.jpg",[82,83,84],"Chroma: egyszerű vektorkeresés egy bökkenővel","A Chroma Apache-2.0 vektoradatbázis, beágyazva, egy csomóponton vagy Chroma Cloudként. Hol kényelmes, és hol érnek véget a jó keresési funkciók a felhőnél?","Chroma vektorkeresés: fiókos szekrény szétválogatott golyókkal, a tetején egy repedt kavics.",{"slug":1554,"published":1555,"updated":1093,"minutes":90,"category":1556,"tags":1557,"tagPages":1561,"keywords":1563,"about":1571,"sources":1576,"cover":1603,"og":1604,"expertise":80,"locales":1605,"lang":84,"title":1606,"description":1607,"coverAlt":1608},"llamaindex","2026-07-14","agents",[1558,11,1559,830,1560],"Agent framework","Python","Workflows",[8,1562],"python",[1554,1564,1565,1566,1567,1568,1569,1570],"llamaindex vagy langgraph","llamaindex workflow-k","llama index ügynök keretrendszer","llamaparse árak","rag keretrendszer python","tartós workflow-k python","dokumentum ügynök keretrendszer",[1572,1575],{"name":1573,"url":1574},"LlamaIndex","https:\u002F\u002Fgithub.com\u002Frun-llama\u002Fllama_index",{"name":110,"url":111},[1577,1579,1582,1585,1588,1591,1594,1597,1600],{"title":1578,"url":1574},"LlamaIndex repository README and focus note",{"title":1580,"url":1581},"Agent Workflows: introduction","https:\u002F\u002Fdevelopers.llamaindex.ai\u002Fpython\u002Fllamaagents\u002Fworkflows\u002F",{"title":1583,"url":1584},"Agent Workflows: writing durable workflows","https:\u002F\u002Fdevelopers.llamaindex.ai\u002Fpython\u002Fllamaagents\u002Fworkflows\u002Fdurable_workflows\u002F",{"title":1586,"url":1587},"Agent Workflows: observability","https:\u002F\u002Fdevelopers.llamaindex.ai\u002Fpython\u002Fllamaagents\u002Fworkflows\u002Fobservability\u002F",{"title":1589,"url":1590},"LlamaAgents overview","https:\u002F\u002Fdevelopers.llamaindex.ai\u002Fpython\u002Fllamaagents\u002Foverview\u002F",{"title":1592,"url":1593},"llama-index-core on PyPI","https:\u002F\u002Fpypi.org\u002Fproject\u002Fllama-index-core\u002F",{"title":1595,"url":1596},"LlamaIndex and LlamaParse pricing","https:\u002F\u002Fwww.llamaindex.ai\u002Fpricing",{"title":1598,"url":1599},"LiteParse documentation","https:\u002F\u002Fdevelopers.llamaindex.ai\u002Fliteparse\u002F",{"title":1601,"url":1602},"LangGraph on GitHub","https:\u002F\u002Fgithub.com\u002Flangchain-ai\u002Flanggraph","\u002Fimages\u002Fblog\u002Fllamaindex\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fllamaindex\u002Fog.jpg",[82,83,84],"LlamaIndex teszt: a legszélesebb adat-eszközkészlet, aminek a súlypontja már elmozdult","LlamaIndex 2026: MIT-licencű Python-keretrendszer adatokhoz és ügynökökhöz, 300-nál több integrációval, Workflows-szal és a LlamaParse felé fordult céggel.","LlamaIndex teszt: egy láma egy nagy nyitott szerszámláda mellett egy kis felhőre néz.",{"slug":1610,"published":1611,"minutes":567,"category":8,"tags":1612,"tagPages":1614,"keywords":1615,"about":1624,"sources":1628,"cover":1653,"og":1654,"expertise":80,"locales":1655,"lang":84,"title":1656,"description":1657,"coverAlt":1658},"pinecone","2026-07-09",[13,10,11,1613,1096],"Namespaces",[17,16,8,1099],[1610,1616,1617,1618,1619,1620,1621,1622,1623],"pinecone vektoradatbázis","pinecone árak","pinecone read unit","pinecone namespace","pinecone vagy qdrant","pinecone vagy pgvector","pinecone documents api","serverless vektoradatbázis költség",[1625,1626,1627],{"name":1110,"url":1111},{"name":544,"url":545},{"name":113,"url":114},[1629,1632,1635,1638,1641,1644,1647,1650],{"title":1630,"url":1631},"Pinecone docs: Index data overview, indexes, namespaces and metadata","https:\u002F\u002Fdocs.pinecone.io\u002Fguides\u002Findex-data\u002Findexing-overview",{"title":1633,"url":1634},"Pinecone docs: Adopt the Documents API, API version 2026-07","https:\u002F\u002Fdocs.pinecone.io\u002Fguides\u002Findex-data\u002Fadopt-the-documents-api",{"title":1636,"url":1637},"Pinecone docs: Create an index, schemas, regions and metrics","https:\u002F\u002Fdocs.pinecone.io\u002Fguides\u002Findex-data\u002Fcreate-an-index",{"title":1639,"url":1640},"Pinecone docs: Understanding Pinecone cost, read units, write units and egress","https:\u002F\u002Fdocs.pinecone.io\u002Fguides\u002Forganizations\u002Fmanage-cost\u002Funderstanding-cost",{"title":1642,"url":1643},"Pinecone pricing: plans, limits and rates","https:\u002F\u002Fwww.pinecone.io\u002Fpricing\u002F",{"title":1645,"url":1646},"Pinecone docs: Semantic search","https:\u002F\u002Fdocs.pinecone.io\u002Fguides\u002Fsearch\u002Fsemantic-search",{"title":1648,"url":1649},"Pinecone docs: Local development with Pinecone Local","https:\u002F\u002Fdocs.pinecone.io\u002Fguides\u002Foperations\u002Flocal-development",{"title":1651,"url":1652},"Pinecone docs: 2026 release notes","https:\u002F\u002Fdocs.pinecone.io\u002Frelease-notes\u002F2026","\u002Fimages\u002Fblog\u002Fpinecone\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fpinecone\u002Fog.jpg",[82,83,84],"Pinecone: értékelés a menedzselt vektordatabázisról","Mennyibe kerül valójában a Pinecone: a read unit a namespace méretével nő, a sémát utólag nem lehet módosítani, és hol éri meg saját vektoradatbázis.","Pinecone értékelés: fenyőtoboz egy felhőn golyókat tart a pikkelyei között, alatta kulcs.",{"slug":1660,"published":1661,"updated":686,"minutes":617,"category":8,"tags":1662,"tagPages":1664,"keywords":1665,"about":1676,"sources":1680,"cover":1728,"og":1729,"expertise":80,"locales":1730,"lang":84,"title":1731,"description":1732,"coverAlt":1733},"pgvector-vs-vector-databases","2026-06-15",[526,1030,11,1096,1663],"EU hosting",[8,1099],[1666,1667,1668,1669,1670,1671,1672,1673,1674,1675],"pgvector vagy vektoradatbázis","pgvector vs qdrant összehasonlítás","pgvector vagy pinecone","legjobb vektoradatbázis 2026","pgvector hnsw iteratív keresés","pgvector halfvec típus","opensearch vs elasticsearch vektorkeresés","vektoradatbázis eu tárhely","hibrid keresés postgres","weaviate vs milvus",[1677,1678,1679],{"name":1110,"url":1111},{"name":1029,"url":1163},{"name":110,"url":111},[1681,1684,1686,1689,1690,1693,1696,1699,1702,1705,1708,1711,1713,1716,1719,1722,1725],{"title":1682,"url":1683},"pgvector README (index limits, HNSW defaults, iterative scans, filtering, halfvec)","https:\u002F\u002Fgithub.com\u002Fpgvector\u002Fpgvector\u002Fblob\u002Fmaster\u002FREADME.md",{"title":1685,"url":1169},"pgvector CHANGELOG (0.4.0 to 0.8.7)",{"title":1687,"url":1688},"pgvectorscale: StreamingDiskANN, statistical binary quantization, filtered search","https:\u002F\u002Fgithub.com\u002Ftimescale\u002Fpgvectorscale",{"title":260,"url":261},{"title":1691,"url":1692},"Qdrant documentation: Hybrid queries","https:\u002F\u002Fqdrant.tech\u002Fdocumentation\u002Fconcepts\u002Fhybrid-queries\u002F",{"title":1694,"url":1695},"Qdrant documentation: Create a cluster (providers, free tier, Hybrid Cloud)","https:\u002F\u002Fqdrant.tech\u002Fdocumentation\u002Fcloud\u002Fcreate-cluster\u002F",{"title":1697,"url":1698},"Weaviate documentation: Hybrid search","https:\u002F\u002Fdocs.weaviate.io\u002Fweaviate\u002Fconcepts\u002Fsearch\u002Fhybrid-search",{"title":1700,"url":1701},"Weaviate documentation: Vector index types","https:\u002F\u002Fdocs.weaviate.io\u002Fweaviate\u002Fconcepts\u002Fvector-index",{"title":1703,"url":1704},"Weaviate Cloud pricing and deployment options","https:\u002F\u002Fweaviate.io\u002Fpricing",{"title":1706,"url":1707},"Milvus documentation: Overview","https:\u002F\u002Fmilvus.io\u002Fdocs\u002Foverview.md",{"title":1709,"url":1710},"Pinecone documentation: Database architecture","https:\u002F\u002Fdocs.pinecone.io\u002Fguides\u002Fget-started\u002Fdatabase-architecture",{"title":1712,"url":1637},"Pinecone documentation: Create an index (clouds, regions, sparse and hybrid)",{"title":1714,"url":1715},"OpenSearch documentation: Methods and engines","https:\u002F\u002Fdocs.opensearch.org\u002Flatest\u002Fmappings\u002Fsupported-field-types\u002Fknn-methods-engines\u002F",{"title":1717,"url":1718},"OpenSearch documentation: Efficient k-NN filtering","https:\u002F\u002Fdocs.opensearch.org\u002Flatest\u002Fvector-search\u002Ffilter-search-knn\u002Fefficient-knn-filtering\u002F",{"title":1720,"url":1721},"Elasticsearch documentation: Dense vector search","https:\u002F\u002Fwww.elastic.co\u002Fdocs\u002Fsolutions\u002Fsearch\u002Fvector\u002Fdense-vector",{"title":1723,"url":1724},"Elasticsearch documentation: kNN query (filter as pre-filter)","https:\u002F\u002Fwww.elastic.co\u002Fdocs\u002Freference\u002Fquery-languages\u002Fquery-dsl\u002Fquery-dsl-knn-query",{"title":1726,"url":1727},"GitHub releases: Qdrant, Weaviate, Milvus, OpenSearch, pgvectorscale (versions as of 1 October 2026)","https:\u002F\u002Fgithub.com\u002Fqdrant\u002Fqdrant\u002Freleases","\u002Fimages\u002Fblog\u002Fpgvector-vs-vector-databases\u002Fcover.webp","\u002Fimages\u002Fblog\u002Fpgvector-vs-vector-databases\u002Fog.jpg",[82,83,84],"pgvector vagy vektoradatbázis? Így válassz vektortárolót 2026-ban","pgvector, Qdrant, Weaviate, Milvus, Pinecone, OpenSearch vagy Elasticsearch? 2026-os útmutató: szűrés, hibrid keresés, skálázás, költség és EU-s hosztolás.","pgvector vagy vektoradatbázis: egy ember az elágazásnál választ egy fiókos elefánt és egy magas raktár között.",{"slug":1735,"published":1736,"updated":616,"minutes":274,"category":1737,"tags":1738,"tagPages":1741,"keywords":1742,"about":1753,"sources":1760,"cover":1827,"og":1828,"expertise":80,"locales":1829,"lang":84,"title":1830,"description":1831,"coverAlt":1832},"fine-tuning-vs-rag-vs-prompting","2026-05-12","llmops",[897,11,1739,1740],"Prompting","Distillation",[8],[1743,1744,1745,1746,1747,1748,1749,1750,1751,1752],"finomhangolás vs rag","prompting vs rag vs finomhangolás","mikor finomhangoljunk llm-et","rag vagy finomhangolás","sft vs dpo vs rft","megerősítéses finomhangolás","llm desztilláció","openai finomhangolás leállítás","finomhangolás döntési fa","finomhangolás tudásért",[1754,1756,1757],{"name":914,"url":1755},"https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FFine-tuning_(deep_learning)",{"name":110,"url":111},{"name":1758,"url":1759},"Knowledge distillation","https:\u002F\u002Fen.wikipedia.org\u002Fwiki\u002FKnowledge_distillation",[1761,1764,1767,1770,1773,1776,1779,1782,1785,1788,1791,1794,1797,1800,1803,1806,1809,1812,1815,1818,1821,1824],{"title":1762,"url":1763},"OpenAI community: OpenAI self-serve fine-tuning availability (wind-down announcement)","https:\u002F\u002Fcommunity.openai.com\u002Ft\u002Fopenai-s-self-serve-fine-tuning-availability\u002F1380481",{"title":1765,"url":1766},"Tessl: OpenAI is shutting down self-serve fine-tuning","https:\u002F\u002Ftessl.io\u002Fblog\u002Fopenai-shutting-fine-tuning-signals-for-enterprise-ai",{"title":1768,"url":1769},"OpenAI API docs: Model optimization","https:\u002F\u002Fdevelopers.openai.com\u002Fapi\u002Fdocs\u002Fguides\u002Fmodel-optimization",{"title":1771,"url":1772},"OpenAI API docs: Supervised fine-tuning","https:\u002F\u002Fdevelopers.openai.com\u002Fapi\u002Fdocs\u002Fguides\u002Fsupervised-fine-tuning",{"title":1774,"url":1775},"OpenAI API docs: Direct preference optimization","https:\u002F\u002Fdevelopers.openai.com\u002Fapi\u002Fdocs\u002Fguides\u002Fdirect-preference-optimization",{"title":1777,"url":1778},"OpenAI API docs: Reinforcement fine-tuning","https:\u002F\u002Fdevelopers.openai.com\u002Fapi\u002Fdocs\u002Fguides\u002Freinforcement-fine-tuning",{"title":1780,"url":1781},"OpenAI API docs: Fine-tuning best practices","https:\u002F\u002Fdevelopers.openai.com\u002Fapi\u002Fdocs\u002Fguides\u002Ffine-tuning-best-practices",{"title":1783,"url":1784},"OpenAI Cookbook: Choosing between SFT, DPO and RFT","https:\u002F\u002Fdevelopers.openai.com\u002Fcookbook\u002Fexamples\u002Ffine_tuning_direct_preference_optimization_guide",{"title":1786,"url":1787},"Google Cloud: Introduction to tuning (Gemini)","https:\u002F\u002Fdocs.cloud.google.com\u002Fvertex-ai\u002Fgenerative-ai\u002Fdocs\u002Fmodels\u002Ftune-models",{"title":1789,"url":1790},"Google Cloud: About supervised fine-tuning for Gemini models","https:\u002F\u002Fdocs.cloud.google.com\u002Fvertex-ai\u002Fgenerative-ai\u002Fdocs\u002Fmodels\u002Fgemini-supervised-tuning",{"title":1792,"url":1793},"Google Cloud: About preference tuning for Gemini models","https:\u002F\u002Fdocs.cloud.google.com\u002Fvertex-ai\u002Fgenerative-ai\u002Fdocs\u002Fmodels\u002Fgemini-preference-tuning",{"title":1795,"url":1796},"Google Cloud: Reward functions for reinforcement learning fine-tuning","https:\u002F\u002Fdocs.cloud.google.com\u002Fgemini-enterprise-agent-platform\u002Fmodels\u002Ftuning\u002Freinforcement-tuning\u002Freinforcement-tuning-job\u002Freward-functions",{"title":1798,"url":1799},"Google Cloud: Supervised and distillation fine-tuning for open models","https:\u002F\u002Fdocs.cloud.google.com\u002Fvertex-ai\u002Fgenerative-ai\u002Fdocs\u002Fmodels\u002Fopen-model-tuning",{"title":1801,"url":1802},"AWS: Fine-tuning for Claude 3 Haiku in Amazon Bedrock is now generally available","https:\u002F\u002Faws.amazon.com\u002Fblogs\u002Faws\u002Ffine-tuning-for-anthropics-claude-3-haiku-model-in-amazon-bedrock-is-now-generally-available\u002F",{"title":1804,"url":1805},"AWS: Amazon Bedrock now supports reinforcement fine-tuning","https:\u002F\u002Faws.amazon.com\u002Fabout-aws\u002Fwhats-new\u002F2025\u002F12\u002Fbedrock-reinforcement-fine-tuning-66-base-models",{"title":1807,"url":1808},"AWS: Amazon Bedrock Model Distillation (preview announcement)","https:\u002F\u002Faws.amazon.com\u002Fblogs\u002Faws\u002Fbuild-faster-more-cost-efficient-highly-accurate-models-with-amazon-bedrock-model-distillation-preview\u002F",{"title":1810,"url":1811},"AWS: Bedrock reinforcement fine-tuning adds open-weight models","https:\u002F\u002Faws.amazon.com\u002Fabout-aws\u002Fwhats-new\u002F2026\u002F02\u002Famazon-bedrock-reinforcement-fine-tuning-openai",{"title":1813,"url":1814},"Microsoft Foundry blog: What is new in Foundry fine-tuning, April 2026","https:\u002F\u002Fdevblogs.microsoft.com\u002Ffoundry\u002Fwhats-new-in-foundry-finetune-april-2026\u002F",{"title":1816,"url":1817},"Microsoft: Announcing new fine-tuning models and techniques in Azure AI Foundry","https:\u002F\u002Fazure.microsoft.com\u002Fen-us\u002Fblog\u002Fannouncing-new-fine-tuning-models-and-techniques-in-azure-ai-foundry\u002F",{"title":1819,"url":1820},"Ovadia et al.: Fine-Tuning or Retrieval? Comparing Knowledge Injection in LLMs","https:\u002F\u002Farxiv.org\u002Fabs\u002F2312.05934",{"title":1822,"url":1823},"Gekhman et al.: Does Fine-Tuning LLMs on New Knowledge Encourage Hallucinations?","https:\u002F\u002Farxiv.org\u002Fabs\u002F2405.05904",{"title":1825,"url":1826},"Hu et al.: LoRA: Low-Rank Adaptation of Large Language Models","https:\u002F\u002Farxiv.org\u002Fabs\u002F2106.09685","\u002Fimages\u002Fblog\u002Ffine-tuning-vs-rag-vs-prompting\u002Fcover.webp","\u002Fimages\u002Fblog\u002Ffine-tuning-vs-rag-vs-prompting\u002Fog.jpg",[82,83,84],"Prompting vs RAG vs fine-tuning vs distillation: döntési útmutató 2026-ra","Fine-tuning, RAG vagy prompting: mit változtat (tudás, viselkedés, formátum), mennyibe kerül, ki kínál 2026-ban SFT-t, DPO-t és RFT-t, és egy döntési fa.","Prompting vs RAG vs fine-tuning: térképes robot négy ajtó előtt, szövegbuborék, könyvek, csavarkulcs és tölcsér jelekkel.",1793451237700]