[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-github-qdrant-search-strategies":3,"mdc--rqud3c-key":47,"related-repo-github-qdrant-search-strategies":454,"related-org-github-qdrant-search-strategies":561},{"slug":4,"name":4,"fn":5,"description":6,"org":7,"tags":11,"stars":28,"repoUrl":29,"updatedAt":30,"license":31,"forks":32,"topics":33,"repo":42,"sourceUrl":45,"mdContent":46},"qdrant-search-strategies","select optimal Qdrant search strategies","Guides Qdrant search strategy selection. Use when someone asks 'should I use hybrid search?', 'BM25 or sparse vectors?', 'how to rerank?', 'results are not relevant', 'I don't get needed results from my dataset but they're there', 'retrieval quality is not good enough', 'results too similar', 'need diversity', 'MMR', 'relevance feedback', 'recommendation API', 'discovery API', 'ColBERT reranking', or 'missing keyword matches'",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},"github","GitHub","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Fgithub.png",[12,16,19,22,25],{"name":13,"slug":14,"type":15},"Architecture","architecture","tag",{"name":17,"slug":18,"type":15},"Qdrant","qdrant",{"name":20,"slug":21,"type":15},"Search","search",{"name":23,"slug":24,"type":15},"AI Infrastructure","ai-infrastructure",{"name":26,"slug":27,"type":15},"Analytics","analytics",36978,"https:\u002F\u002Fgithub.com\u002Fgithub\u002Fawesome-copilot","2026-04-18T04:46:18.812306",null,4632,[34,35,36,37,38,39,40,41],"agent-skills","agents","ai","awesome","custom-agents","github-copilot","hacktoberfest","prompt-engineering",{"repoUrl":29,"stars":28,"forks":32,"topics":43,"description":44},[34,35,36,37,38,39,40,41],"Community-contributed instructions, agents, skills, and configurations to help you make the most of GitHub Copilot.","https:\u002F\u002Fgithub.com\u002Fgithub\u002Fawesome-copilot\u002Ftree\u002FHEAD\u002Fskills\u002Fqdrant-search-quality\u002Fsearch-strategies","---\nname: qdrant-search-strategies\ndescription: \"Guides Qdrant search strategy selection. Use when someone asks 'should I use hybrid search?', 'BM25 or sparse vectors?', 'how to rerank?', 'results are not relevant', 'I don't get needed results from my dataset but they're there', 'retrieval quality is not good enough', 'results too similar', 'need diversity', 'MMR', 'relevance feedback', 'recommendation API', 'discovery API', 'ColBERT reranking', or 'missing keyword matches'\"\n---\n\n# How to Improve Search Results with Advanced Strategies\n\nThese strategies complement basic vector search. Use them after confirming the embedding model is fitting the task and HNSW config is correct. If exact search returns bad results, verify the selection of the embedding model (retriever) first.\nIf the user wants to use a weaker embedding model because it is small, fast, and cheap, use reranking or relevance feedback to improve search quality.\n\n## Missing Obvious Keyword Matches\n\nUse when: pure vector search misses results that contain obvious keyword matches. Domain terminology not in embedding training data, exact keyword matching critical (brand names, SKUs), acronyms common. Skip when: pure semantic queries, all data in training set, latency budget very tight.\n\n- Dense + sparse with `prefetch` and fusion [Hybrid search](https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fsearch\u002Fhybrid-queries\u002F?s=hybrid-search)\n- Prefer learned sparse ([miniCOIL](https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Ffastembed\u002Ffastembed-minicoil\u002F), SPLADE, GTE) over raw BM25 if applicable (when user needs smart keywords matching and learned sparse models know the vocabulary of the domain)\n- For non-English languages, [configure sparse BM25 parameters accordingly](https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fsearch\u002Ftext-search\u002F?s=language-specific-settings)\n- RRF: good default, supports weighted (v1.17+) [RRF](https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fsearch\u002Fhybrid-queries\u002F?s=reciprocal-rank-fusion-rrf)\n- DBSF with asymmetric limits (sparse_limit=250, dense_limit=100) can outperform RRF for technical docs [DBSF](https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fsearch\u002Fhybrid-queries\u002F?s=distribution-based-score-fusion-dbsf)\n- Fusion can also be done through reranking\n\n## Right Documents Found But Wrong Order\n\nUse when: good recall but poor precision (right docs in top-100, not top-10).\n\n- Cross-encoder rerankers via FastEmbed [Rerankers](https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Ffastembed\u002Ffastembed-rerankers\u002F)\n- See how to use [Multistage queries](https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fsearch\u002Fhybrid-queries\u002F?s=multi-stage-queries) in Qdrant\n- ColBERT and ColPali\u002FColQwen reranking is especially precise due to late interaction mechanisms, but it is heavy. It is important to configure and store multivectors without building HNSW for them to save resources. See [Multivector representation](https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Ftutorials-search-engineering\u002Fusing-multivector-representations\u002F)\n\n## Right Documents Not Found But They Are There\n\nUse when: basic retrieval is in place but the retriever misses relevant items you know exist in the dataset. Works on any embeddable data (text, images, etc.).\n\nRelevance Feedback (RF) Query uses a feedback model's scores on retrieved results to steer the retriever through the full vector space on subsequent iterations, like reranking the entire collection through the retriever. Complementary to reranking: a reranker sees a limited subset, RF leverages feedback signals collection-wide. Even 3–5 feedback scores are enough. Can run multiple iterations.\n\nA feedback model is anything producing a relevance score per document: a bi-encoder, cross-encoder, late-interaction model, LLM-as-judge. Fuzzy relevance scores work, not just binary (good\u002Fbad, relevant\u002Firrelevant), due to the fact that feedback is expressed as a graded relevance score (higher = more relevant).\n\nSkip when: if the retriever already has strong recall, or if retriever and feedback model strongly agree on relevance.\n\n- RF Query is currently based on a [3-parameter naive formula](https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fsearch\u002Fsearch-relevance\u002F?s=naive-strategy) with no universal defaults, so it must be tuned per dataset, retriever, and feedback model\n- Use [qdrant-relevance-feedback](https:\u002F\u002Fpypi.org\u002Fproject\u002Fqdrant-relevance-feedback\u002F) to tune parameters, evaluate impact with Evaluator, and check retriever-feedback agreement. See README for setup instructions. No GPUs are needed, and the framework also provides predefined retriever and feedback model options.\n- Check the configuration of the [Relevance Feedback Query API](https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fsearch\u002Fsearch-relevance\u002F?s=relevance-feedback)\n- Use this as a helper end-to-end text retrieval example with parameter tuning and evals to understand how to use the API and run the `qdrant-relevance-feedback` framework: [RF tutorial](https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Ftutorials-search-engineering\u002Fusing-relevance-feedback\u002F)\n\n## Results Too Similar\n\nUse when: top results are redundant, near-duplicates, or lack diversity. Common in dense content domains (academic papers, product catalogs).\n\n- Use MMR (v1.15+) as a query parameter with `diversity` to balance relevance and diversity [MMR](https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fsearch\u002Fsearch-relevance\u002F?s=maximal-marginal-relevance-mmr)\n- Start with `diversity=0.5`, lower for more precision, higher for more exploration\n- MMR is slower than standard search. Only use when redundancy is an actual problem.\n\n## Know What Good Results Could Look Like But Can't Get Them\n\nUse when: you can provide positive and negative example points to steer search closer to positive and further from negative.\n\n- Recommendation API: positive\u002Fnegative examples to recommend fitting vectors [Recommendation API](https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fsearch\u002Fexplore\u002F?s=recommendation-api)\n  - Best score strategy: better for diverse examples, supports negative-only [Best score](https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fsearch\u002Fexplore\u002F?s=best-score-strategy)\n- Discovery API: context pairs (positive\u002Fnegative) to constrain search regions without a request target [Discovery](https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fsearch\u002Fexplore\u002F?s=discovery-api)\n\n## Have Business Logic Behind Relevance\nUse when: results should be additionally ranked according to some business logic based on data, like recency or distance.\n\nCheck how to set up in [Score Boosting docs](https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fsearch\u002Fsearch-relevance\u002F?s=score-boosting)\n\n## What NOT to Do\n\n- Use hybrid search before verifying pure vector quality (adds complexity, may mask model issues)\n- Use BM25 on non-English text without correctly configuring language-specific stop-word removal (severely degraded results)\n- Skip evaluation when adding relevance feedback (it's good to check on real queries that it actually could help)\n",{"data":48,"body":49},{"name":4,"description":6},{"type":50,"children":51},"root",[52,61,67,74,79,162,168,173,214,220,225,230,235,240,302,308,313,354,360,365,407,413,418,430,436],{"type":53,"tag":54,"props":55,"children":57},"element","h1",{"id":56},"how-to-improve-search-results-with-advanced-strategies",[58],{"type":59,"value":60},"text","How to Improve Search Results with Advanced Strategies",{"type":53,"tag":62,"props":63,"children":64},"p",{},[65],{"type":59,"value":66},"These strategies complement basic vector search. Use them after confirming the embedding model is fitting the task and HNSW config is correct. If exact search returns bad results, verify the selection of the embedding model (retriever) first.\nIf the user wants to use a weaker embedding model because it is small, fast, and cheap, use reranking or relevance feedback to improve search quality.",{"type":53,"tag":68,"props":69,"children":71},"h2",{"id":70},"missing-obvious-keyword-matches",[72],{"type":59,"value":73},"Missing Obvious Keyword Matches",{"type":53,"tag":62,"props":75,"children":76},{},[77],{"type":59,"value":78},"Use when: pure vector search misses results that contain obvious keyword matches. Domain terminology not in embedding training data, exact keyword matching critical (brand names, SKUs), acronyms common. Skip when: pure semantic queries, all data in training set, latency budget very tight.",{"type":53,"tag":80,"props":81,"children":82},"ul",{},[83,107,121,133,145,157],{"type":53,"tag":84,"props":85,"children":86},"li",{},[87,89,96,98],{"type":59,"value":88},"Dense + sparse with ",{"type":53,"tag":90,"props":91,"children":93},"code",{"className":92},[],[94],{"type":59,"value":95},"prefetch",{"type":59,"value":97}," and fusion ",{"type":53,"tag":99,"props":100,"children":104},"a",{"href":101,"rel":102},"https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fsearch\u002Fhybrid-queries\u002F?s=hybrid-search",[103],"nofollow",[105],{"type":59,"value":106},"Hybrid search",{"type":53,"tag":84,"props":108,"children":109},{},[110,112,119],{"type":59,"value":111},"Prefer learned sparse (",{"type":53,"tag":99,"props":113,"children":116},{"href":114,"rel":115},"https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Ffastembed\u002Ffastembed-minicoil\u002F",[103],[117],{"type":59,"value":118},"miniCOIL",{"type":59,"value":120},", SPLADE, GTE) over raw BM25 if applicable (when user needs smart keywords matching and learned sparse models know the vocabulary of the domain)",{"type":53,"tag":84,"props":122,"children":123},{},[124,126],{"type":59,"value":125},"For non-English languages, ",{"type":53,"tag":99,"props":127,"children":130},{"href":128,"rel":129},"https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fsearch\u002Ftext-search\u002F?s=language-specific-settings",[103],[131],{"type":59,"value":132},"configure sparse BM25 parameters accordingly",{"type":53,"tag":84,"props":134,"children":135},{},[136,138],{"type":59,"value":137},"RRF: good default, supports weighted (v1.17+) ",{"type":53,"tag":99,"props":139,"children":142},{"href":140,"rel":141},"https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fsearch\u002Fhybrid-queries\u002F?s=reciprocal-rank-fusion-rrf",[103],[143],{"type":59,"value":144},"RRF",{"type":53,"tag":84,"props":146,"children":147},{},[148,150],{"type":59,"value":149},"DBSF with asymmetric limits (sparse_limit=250, dense_limit=100) can outperform RRF for technical docs ",{"type":53,"tag":99,"props":151,"children":154},{"href":152,"rel":153},"https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fsearch\u002Fhybrid-queries\u002F?s=distribution-based-score-fusion-dbsf",[103],[155],{"type":59,"value":156},"DBSF",{"type":53,"tag":84,"props":158,"children":159},{},[160],{"type":59,"value":161},"Fusion can also be done through reranking",{"type":53,"tag":68,"props":163,"children":165},{"id":164},"right-documents-found-but-wrong-order",[166],{"type":59,"value":167},"Right Documents Found But Wrong Order",{"type":53,"tag":62,"props":169,"children":170},{},[171],{"type":59,"value":172},"Use when: good recall but poor precision (right docs in top-100, not top-10).",{"type":53,"tag":80,"props":174,"children":175},{},[176,188,202],{"type":53,"tag":84,"props":177,"children":178},{},[179,181],{"type":59,"value":180},"Cross-encoder rerankers via FastEmbed ",{"type":53,"tag":99,"props":182,"children":185},{"href":183,"rel":184},"https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Ffastembed\u002Ffastembed-rerankers\u002F",[103],[186],{"type":59,"value":187},"Rerankers",{"type":53,"tag":84,"props":189,"children":190},{},[191,193,200],{"type":59,"value":192},"See how to use ",{"type":53,"tag":99,"props":194,"children":197},{"href":195,"rel":196},"https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fsearch\u002Fhybrid-queries\u002F?s=multi-stage-queries",[103],[198],{"type":59,"value":199},"Multistage queries",{"type":59,"value":201}," in Qdrant",{"type":53,"tag":84,"props":203,"children":204},{},[205,207],{"type":59,"value":206},"ColBERT and ColPali\u002FColQwen reranking is especially precise due to late interaction mechanisms, but it is heavy. It is important to configure and store multivectors without building HNSW for them to save resources. See ",{"type":53,"tag":99,"props":208,"children":211},{"href":209,"rel":210},"https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Ftutorials-search-engineering\u002Fusing-multivector-representations\u002F",[103],[212],{"type":59,"value":213},"Multivector representation",{"type":53,"tag":68,"props":215,"children":217},{"id":216},"right-documents-not-found-but-they-are-there",[218],{"type":59,"value":219},"Right Documents Not Found But They Are There",{"type":53,"tag":62,"props":221,"children":222},{},[223],{"type":59,"value":224},"Use when: basic retrieval is in place but the retriever misses relevant items you know exist in the dataset. Works on any embeddable data (text, images, etc.).",{"type":53,"tag":62,"props":226,"children":227},{},[228],{"type":59,"value":229},"Relevance Feedback (RF) Query uses a feedback model's scores on retrieved results to steer the retriever through the full vector space on subsequent iterations, like reranking the entire collection through the retriever. Complementary to reranking: a reranker sees a limited subset, RF leverages feedback signals collection-wide. Even 3–5 feedback scores are enough. Can run multiple iterations.",{"type":53,"tag":62,"props":231,"children":232},{},[233],{"type":59,"value":234},"A feedback model is anything producing a relevance score per document: a bi-encoder, cross-encoder, late-interaction model, LLM-as-judge. Fuzzy relevance scores work, not just binary (good\u002Fbad, relevant\u002Firrelevant), due to the fact that feedback is expressed as a graded relevance score (higher = more relevant).",{"type":53,"tag":62,"props":236,"children":237},{},[238],{"type":59,"value":239},"Skip when: if the retriever already has strong recall, or if retriever and feedback model strongly agree on relevance.",{"type":53,"tag":80,"props":241,"children":242},{},[243,257,271,283],{"type":53,"tag":84,"props":244,"children":245},{},[246,248,255],{"type":59,"value":247},"RF Query is currently based on a ",{"type":53,"tag":99,"props":249,"children":252},{"href":250,"rel":251},"https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fsearch\u002Fsearch-relevance\u002F?s=naive-strategy",[103],[253],{"type":59,"value":254},"3-parameter naive formula",{"type":59,"value":256}," with no universal defaults, so it must be tuned per dataset, retriever, and feedback model",{"type":53,"tag":84,"props":258,"children":259},{},[260,262,269],{"type":59,"value":261},"Use ",{"type":53,"tag":99,"props":263,"children":266},{"href":264,"rel":265},"https:\u002F\u002Fpypi.org\u002Fproject\u002Fqdrant-relevance-feedback\u002F",[103],[267],{"type":59,"value":268},"qdrant-relevance-feedback",{"type":59,"value":270}," to tune parameters, evaluate impact with Evaluator, and check retriever-feedback agreement. See README for setup instructions. No GPUs are needed, and the framework also provides predefined retriever and feedback model options.",{"type":53,"tag":84,"props":272,"children":273},{},[274,276],{"type":59,"value":275},"Check the configuration of the ",{"type":53,"tag":99,"props":277,"children":280},{"href":278,"rel":279},"https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fsearch\u002Fsearch-relevance\u002F?s=relevance-feedback",[103],[281],{"type":59,"value":282},"Relevance Feedback Query API",{"type":53,"tag":84,"props":284,"children":285},{},[286,288,293,295],{"type":59,"value":287},"Use this as a helper end-to-end text retrieval example with parameter tuning and evals to understand how to use the API and run the ",{"type":53,"tag":90,"props":289,"children":291},{"className":290},[],[292],{"type":59,"value":268},{"type":59,"value":294}," framework: ",{"type":53,"tag":99,"props":296,"children":299},{"href":297,"rel":298},"https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Ftutorials-search-engineering\u002Fusing-relevance-feedback\u002F",[103],[300],{"type":59,"value":301},"RF tutorial",{"type":53,"tag":68,"props":303,"children":305},{"id":304},"results-too-similar",[306],{"type":59,"value":307},"Results Too Similar",{"type":53,"tag":62,"props":309,"children":310},{},[311],{"type":59,"value":312},"Use when: top results are redundant, near-duplicates, or lack diversity. Common in dense content domains (academic papers, product catalogs).",{"type":53,"tag":80,"props":314,"children":315},{},[316,336,349],{"type":53,"tag":84,"props":317,"children":318},{},[319,321,327,329],{"type":59,"value":320},"Use MMR (v1.15+) as a query parameter with ",{"type":53,"tag":90,"props":322,"children":324},{"className":323},[],[325],{"type":59,"value":326},"diversity",{"type":59,"value":328}," to balance relevance and diversity ",{"type":53,"tag":99,"props":330,"children":333},{"href":331,"rel":332},"https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fsearch\u002Fsearch-relevance\u002F?s=maximal-marginal-relevance-mmr",[103],[334],{"type":59,"value":335},"MMR",{"type":53,"tag":84,"props":337,"children":338},{},[339,341,347],{"type":59,"value":340},"Start with ",{"type":53,"tag":90,"props":342,"children":344},{"className":343},[],[345],{"type":59,"value":346},"diversity=0.5",{"type":59,"value":348},", lower for more precision, higher for more exploration",{"type":53,"tag":84,"props":350,"children":351},{},[352],{"type":59,"value":353},"MMR is slower than standard search. Only use when redundancy is an actual problem.",{"type":53,"tag":68,"props":355,"children":357},{"id":356},"know-what-good-results-could-look-like-but-cant-get-them",[358],{"type":59,"value":359},"Know What Good Results Could Look Like But Can't Get Them",{"type":53,"tag":62,"props":361,"children":362},{},[363],{"type":59,"value":364},"Use when: you can provide positive and negative example points to steer search closer to positive and further from negative.",{"type":53,"tag":80,"props":366,"children":367},{},[368,395],{"type":53,"tag":84,"props":369,"children":370},{},[371,373,380],{"type":59,"value":372},"Recommendation API: positive\u002Fnegative examples to recommend fitting vectors ",{"type":53,"tag":99,"props":374,"children":377},{"href":375,"rel":376},"https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fsearch\u002Fexplore\u002F?s=recommendation-api",[103],[378],{"type":59,"value":379},"Recommendation API",{"type":53,"tag":80,"props":381,"children":382},{},[383],{"type":53,"tag":84,"props":384,"children":385},{},[386,388],{"type":59,"value":387},"Best score strategy: better for diverse examples, supports negative-only ",{"type":53,"tag":99,"props":389,"children":392},{"href":390,"rel":391},"https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fsearch\u002Fexplore\u002F?s=best-score-strategy",[103],[393],{"type":59,"value":394},"Best score",{"type":53,"tag":84,"props":396,"children":397},{},[398,400],{"type":59,"value":399},"Discovery API: context pairs (positive\u002Fnegative) to constrain search regions without a request target ",{"type":53,"tag":99,"props":401,"children":404},{"href":402,"rel":403},"https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fsearch\u002Fexplore\u002F?s=discovery-api",[103],[405],{"type":59,"value":406},"Discovery",{"type":53,"tag":68,"props":408,"children":410},{"id":409},"have-business-logic-behind-relevance",[411],{"type":59,"value":412},"Have Business Logic Behind Relevance",{"type":53,"tag":62,"props":414,"children":415},{},[416],{"type":59,"value":417},"Use when: results should be additionally ranked according to some business logic based on data, like recency or distance.",{"type":53,"tag":62,"props":419,"children":420},{},[421,423],{"type":59,"value":422},"Check how to set up in ",{"type":53,"tag":99,"props":424,"children":427},{"href":425,"rel":426},"https:\u002F\u002Fsearch.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fsearch\u002Fsearch-relevance\u002F?s=score-boosting",[103],[428],{"type":59,"value":429},"Score Boosting docs",{"type":53,"tag":68,"props":431,"children":433},{"id":432},"what-not-to-do",[434],{"type":59,"value":435},"What NOT to Do",{"type":53,"tag":80,"props":437,"children":438},{},[439,444,449],{"type":53,"tag":84,"props":440,"children":441},{},[442],{"type":59,"value":443},"Use hybrid search before verifying pure vector quality (adds complexity, may mask model issues)",{"type":53,"tag":84,"props":445,"children":446},{},[447],{"type":59,"value":448},"Use BM25 on non-English text without correctly configuring language-specific stop-word removal (severely degraded results)",{"type":53,"tag":84,"props":450,"children":451},{},[452],{"type":59,"value":453},"Skip evaluation when adding relevance feedback (it's good to check on real queries that it actually could help)",{"items":455,"total":560},[456,472,490,506,518,534,548],{"slug":457,"name":457,"fn":458,"description":459,"org":460,"tags":461,"stars":28,"repoUrl":29,"updatedAt":471},"qdrant-horizontal-scaling","guide Qdrant horizontal scaling decisions","Diagnoses and guides Qdrant horizontal scaling decisions. Use when someone asks 'vertical or horizontal?', 'how many nodes?', 'how many shards?', 'how to add nodes', 'resharding', 'data doesn't fit', or 'need more capacity'. Also use when data growth outpaces current deployment.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[462,463,464,467,470],{"name":23,"slug":24,"type":15},{"name":13,"slug":14,"type":15},{"name":465,"slug":466,"type":15},"Capacity Planning","capacity-planning",{"name":468,"slug":469,"type":15},"Database","database",{"name":17,"slug":18,"type":15},"2026-04-18T04:46:16.349561",{"slug":473,"name":473,"fn":474,"description":475,"org":476,"tags":477,"stars":28,"repoUrl":29,"updatedAt":489},"qdrant-indexing-performance-optimization","optimize Qdrant indexing performance","Diagnoses and fixes slow Qdrant indexing and data ingestion. Use when someone reports 'uploads are slow', 'indexing takes forever', 'optimizer is stuck', 'HNSW build time too long', or 'data uploaded but search is bad'. Also use when optimizer status shows errors, segments won't merge, or indexing threshold questions arise.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[478,481,482,485,488],{"name":479,"slug":480,"type":15},"Data Engineering","data-engineering",{"name":468,"slug":469,"type":15},{"name":483,"slug":484,"type":15},"ETL","etl",{"name":486,"slug":487,"type":15},"Performance","performance",{"name":17,"slug":18,"type":15},"2026-04-18T04:46:02.766357",{"slug":491,"name":491,"fn":492,"description":493,"org":494,"tags":495,"stars":28,"repoUrl":29,"updatedAt":505},"qdrant-memory-usage-optimization","optimize Qdrant memory usage","Diagnoses and reduces Qdrant memory usage. Use when someone reports 'memory too high', 'RAM keeps growing', 'node crashed', 'out of memory', 'memory leak', or asks 'why is memory usage so high?', 'how to reduce RAM?'. Also use when memory doesn't match calculations, quantization didn't help, or nodes crash during recovery.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[496,499,500,503,504],{"name":497,"slug":498,"type":15},"Cost Optimization","cost-optimization",{"name":468,"slug":469,"type":15},{"name":501,"slug":502,"type":15},"Observability","observability",{"name":486,"slug":487,"type":15},{"name":17,"slug":18,"type":15},"2026-04-18T04:46:01.525023",{"slug":507,"name":507,"fn":508,"description":509,"org":510,"tags":511,"stars":28,"repoUrl":29,"updatedAt":517},"qdrant-minimize-latency","minimize Qdrant query latency","Guides Qdrant query latency optimization. Use when someone asks 'search is slow', 'how to reduce latency', 'p99 is too high', 'tail latency', 'single query too slow', 'how to make search faster', or 'latency spikes'.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[512,513,514,515,516],{"name":13,"slug":14,"type":15},{"name":468,"slug":469,"type":15},{"name":486,"slug":487,"type":15},{"name":17,"slug":18,"type":15},{"name":20,"slug":21,"type":15},"2026-04-18T04:46:10.126667",{"slug":519,"name":519,"fn":520,"description":521,"org":522,"tags":523,"stars":28,"repoUrl":29,"updatedAt":533},"qdrant-monitoring-debugging","debug Qdrant production issues with metrics","Diagnoses Qdrant production issues using metrics and observability tools. Use when someone reports 'optimizer stuck', 'indexing too slow', 'memory too high', 'OOM crash', 'queries are slow', 'latency spike', or 'search was fast now it's slow'. Also use when performance degrades without obvious config changes.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[524,527,530,531,532],{"name":525,"slug":526,"type":15},"Debugging","debugging",{"name":528,"slug":529,"type":15},"Monitoring","monitoring",{"name":501,"slug":502,"type":15},{"name":486,"slug":487,"type":15},{"name":17,"slug":18,"type":15},"2026-04-18T04:46:06.450784",{"slug":535,"name":535,"fn":536,"description":537,"org":538,"tags":539,"stars":28,"repoUrl":29,"updatedAt":547},"qdrant-monitoring-setup","set up Qdrant monitoring and alerting","Guides Qdrant monitoring setup including Prometheus scraping, health probes, Hybrid Cloud metrics, alerting, and log centralization. Use when someone asks 'how to set up monitoring', 'Prometheus config', 'Grafana dashboard', 'health check endpoints', 'how to scrape Hybrid Cloud', 'what alerts to set', 'how to centralize logs', or 'audit logging'.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[540,541,542,543,544],{"name":23,"slug":24,"type":15},{"name":528,"slug":529,"type":15},{"name":501,"slug":502,"type":15},{"name":17,"slug":18,"type":15},{"name":545,"slug":546,"type":15},"SRE","sre","2026-04-18T04:46:05.217192",{"slug":549,"name":549,"fn":550,"description":551,"org":552,"tags":553,"stars":28,"repoUrl":29,"updatedAt":559},"qdrant-scaling-data-volume","scale Qdrant data volume","Guides Qdrant data volume scaling decisions. Use when someone asks 'data doesn't fit on one node', 'too much data', 'need more storage', 'vertical or horizontal scaling', 'tenant scaling', 'time window rotation', or 'data growth exceeds capacity'.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[554,555,556,557,558],{"name":23,"slug":24,"type":15},{"name":13,"slug":14,"type":15},{"name":465,"slug":466,"type":15},{"name":468,"slug":469,"type":15},{"name":17,"slug":18,"type":15},"2026-04-18T04:46:07.684464",15,{"items":562,"total":677},[563,571,579,587,595,603,611,619,631,643,657,669],{"slug":457,"name":457,"fn":458,"description":459,"org":564,"tags":565,"stars":28,"repoUrl":29,"updatedAt":471},{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[566,567,568,569,570],{"name":23,"slug":24,"type":15},{"name":13,"slug":14,"type":15},{"name":465,"slug":466,"type":15},{"name":468,"slug":469,"type":15},{"name":17,"slug":18,"type":15},{"slug":473,"name":473,"fn":474,"description":475,"org":572,"tags":573,"stars":28,"repoUrl":29,"updatedAt":489},{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[574,575,576,577,578],{"name":479,"slug":480,"type":15},{"name":468,"slug":469,"type":15},{"name":483,"slug":484,"type":15},{"name":486,"slug":487,"type":15},{"name":17,"slug":18,"type":15},{"slug":491,"name":491,"fn":492,"description":493,"org":580,"tags":581,"stars":28,"repoUrl":29,"updatedAt":505},{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[582,583,584,585,586],{"name":497,"slug":498,"type":15},{"name":468,"slug":469,"type":15},{"name":501,"slug":502,"type":15},{"name":486,"slug":487,"type":15},{"name":17,"slug":18,"type":15},{"slug":507,"name":507,"fn":508,"description":509,"org":588,"tags":589,"stars":28,"repoUrl":29,"updatedAt":517},{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[590,591,592,593,594],{"name":13,"slug":14,"type":15},{"name":468,"slug":469,"type":15},{"name":486,"slug":487,"type":15},{"name":17,"slug":18,"type":15},{"name":20,"slug":21,"type":15},{"slug":519,"name":519,"fn":520,"description":521,"org":596,"tags":597,"stars":28,"repoUrl":29,"updatedAt":533},{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[598,599,600,601,602],{"name":525,"slug":526,"type":15},{"name":528,"slug":529,"type":15},{"name":501,"slug":502,"type":15},{"name":486,"slug":487,"type":15},{"name":17,"slug":18,"type":15},{"slug":535,"name":535,"fn":536,"description":537,"org":604,"tags":605,"stars":28,"repoUrl":29,"updatedAt":547},{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[606,607,608,609,610],{"name":23,"slug":24,"type":15},{"name":528,"slug":529,"type":15},{"name":501,"slug":502,"type":15},{"name":17,"slug":18,"type":15},{"name":545,"slug":546,"type":15},{"slug":549,"name":549,"fn":550,"description":551,"org":612,"tags":613,"stars":28,"repoUrl":29,"updatedAt":559},{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[614,615,616,617,618],{"name":23,"slug":24,"type":15},{"name":13,"slug":14,"type":15},{"name":465,"slug":466,"type":15},{"name":468,"slug":469,"type":15},{"name":17,"slug":18,"type":15},{"slug":620,"name":620,"fn":621,"description":622,"org":623,"tags":624,"stars":28,"repoUrl":29,"updatedAt":630},"qdrant-scaling-qps","scale Qdrant query throughput","Guides Qdrant query throughput (QPS) scaling. Use when someone asks 'how to increase QPS', 'need more throughput', 'queries per second too low', 'batch search', 'read replicas', or 'how to handle more concurrent queries'.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[625,626,627,628,629],{"name":23,"slug":24,"type":15},{"name":13,"slug":14,"type":15},{"name":468,"slug":469,"type":15},{"name":486,"slug":487,"type":15},{"name":17,"slug":18,"type":15},"2026-04-18T04:46:08.905219",{"slug":632,"name":632,"fn":633,"description":634,"org":635,"tags":636,"stars":28,"repoUrl":29,"updatedAt":642},"qdrant-scaling-query-volume","scale Qdrant query volume and pagination","Guides Qdrant query volume scaling. Use when someone asks 'query returns too many results', 'scroll performance', 'large limit values', 'paginating search results', 'fetching many vectors', or 'high cardinality results'.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[637,638,639,640,641],{"name":13,"slug":14,"type":15},{"name":468,"slug":469,"type":15},{"name":486,"slug":487,"type":15},{"name":17,"slug":18,"type":15},{"name":20,"slug":21,"type":15},"2026-04-18T04:46:11.371326",{"slug":644,"name":644,"fn":645,"description":646,"org":647,"tags":648,"stars":28,"repoUrl":29,"updatedAt":656},"qdrant-search-quality-diagnosis","diagnose Qdrant search quality issues","Diagnoses Qdrant search quality issues. Use when someone reports 'results are bad', 'wrong results', 'not relevant results', 'missing matches', 'recall is low', 'approximate search worse than exact', 'which embedding model', or 'quality dropped after quantization'. Also use when search quality degrades without obvious changes.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[649,650,653,654,655],{"name":26,"slug":27,"type":15},{"name":651,"slug":652,"type":15},"Data Quality","data-quality",{"name":525,"slug":526,"type":15},{"name":17,"slug":18,"type":15},{"name":20,"slug":21,"type":15},"2026-04-18T04:46:17.579894",{"slug":658,"name":658,"fn":659,"description":660,"org":661,"tags":662,"stars":28,"repoUrl":29,"updatedAt":668},"qdrant-search-speed-optimization","optimize Qdrant search speed","Diagnoses and fixes slow Qdrant search. Use when someone reports 'search is slow', 'high latency', 'queries take too long', 'low QPS', 'throughput too low', 'filtered search is slow', or 'search was fast but now it's slow'. Also use when search performance degrades after config changes or data growth.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[663,664,665,666,667],{"name":26,"slug":27,"type":15},{"name":468,"slug":469,"type":15},{"name":486,"slug":487,"type":15},{"name":17,"slug":18,"type":15},{"name":20,"slug":21,"type":15},"2026-04-18T04:46:03.987332",{"slug":4,"name":4,"fn":5,"description":6,"org":670,"tags":671,"stars":28,"repoUrl":29,"updatedAt":30},{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[672,673,674,675,676],{"name":23,"slug":24,"type":15},{"name":26,"slug":27,"type":15},{"name":13,"slug":14,"type":15},{"name":17,"slug":18,"type":15},{"name":20,"slug":21,"type":15},45]