id
stringlengths
36
36
url
stringclasses
504 values
title
stringclasses
475 values
section
stringlengths
0
148
content_type
stringclasses
3 values
has_code
bool
2 classes
text
stringlengths
147
22.5k
vector
listlengths
384
384
e8d9172b-b13f-5da1-b516-3cc9c9113647
/documentation/search/text-search/sparse-vector-search/
Sparse Vector Search
BM25 > Configuring BM25 Parameters
documentation
true
Sparse Vector Search > BM25 > Configuring BM25 Parameters The sparse vector BM25 model has three adjustable parameters, based on the BM25 [ranking function](https://en.wikipedia.org/wiki/Okapi_BM25#The_ranking_function), that you can set at ingest time to optimize search results for your specific use case: - `k`. Con...
[ 0.022786682471632957, -0.04762823134660721, -0.03827657178044319, -0.051991283893585205, 0.0643417239189148, -0.056237295269966125, 0.06908965110778809, 0.03191110119223595, 0.0623151920735836, -0.01254532765597105, 0.03110687993466854, 0.01777651160955429, 0.05884314700961113, -0.03285682...
e8e92691-cb74-5f4f-a9fe-0aa15f49613d
/documentation/tutorials-build-essentials/semantic-cache/
Semantic Caching for RAG
Set Up Cache Lookup > Calibrate the Threshold
documentation
true
Semantic Caching for RAG > Set Up Cache Lookup > Calibrate the Threshold To calibrate a threshold, you'd need a golden set for your data: incoming questions labeled with whether the cache should answer them. You could build one based on your query logs: take questions that score close to each other and label whether ...
[ 0.023344531655311584, -0.04019191116094589, -0.07581079006195068, -0.019324952736496925, 0.05582099035382271, -0.010435108095407486, 0.025835717096924782, 0.0174905713647604, 0.05681702122092247, -0.05196624994277954, 0.02489212527871132, 0.0016249879263341427, 0.0728101059794426, -0.03588...
e917c13c-5c49-5233-9bed-42d70877b36e
/documentation/search-precision/reranking-semantic-search/
Reranking for Better Search
Implementing Vector Search with Reranking > Implementation > Steps to Set Up Qdrant Cloud
documentation
true
Reranking for Better Search > Implementing Vector Search with Reranking > Implementation > Steps to Set Up Qdrant Cloud 1. **Sign Up**: Head to Qdrant’s website and sign up for a cloud account using your email, Google, or GitHub credentials. 2. **Create Your First Cluster**: Once you’re in, navigate to the Overview se...
[ 0.04073410481214523, -0.06543227285146713, -0.03632901981472969, -0.013293188996613026, 0.0957220122218132, -0.03134356439113617, 0.05558133125305176, 0.04120175167918205, 0.07367057353258133, -0.053301986306905746, 0.014640246517956257, 0.03419659286737442, 0.047558289021253586, -0.041489...
e93c8acd-a68d-57ed-9792-90c98a9bfb00
/articles/dedicated-vector-search/
Why Vector Search Needs a Dedicated Database
Vector Index > Immutability
article
false
Why Vector Search Needs a Dedicated Database > Vector Index > Immutability With index maintenance divided between segments, Qdrant can ensure high performance even during heavy load, and additional optimizations secure that further. These optimizations come from an idea that working with immutable structures introduce...
[ 0.0321638323366642, -0.04783949255943298, -0.07942978292703629, -0.029503561556339264, 0.03830396384000778, -0.05286335200071335, 0.0687585324048996, 0.014083589427173138, 0.08578388392925262, -0.012339032255113125, 0.006199145223945379, 0.021740896627306938, 0.04681084677577019, -0.031994...
e9443390-d685-5865-856d-0d5f731812a6
/articles/io_uring/
Faster Disk I/O for Vector Search Using io_uring
Input+Output > Combating the Poll-ution
article
false
Faster Disk I/O for Vector Search Using io_uring > Input+Output > Combating the Poll-ution There were multiple experiments to improve matters, some even going so far as moving a HTTP server into the kernel, which of course brought its own share of problems. Others like Intel added their own APIs that ignored the kerne...
[ 0.04001334682106972, -0.04156298562884331, -0.05642528831958771, 0.0027422173880040646, 0.07939878851175308, -0.020945796743035316, 0.05592238903045654, -0.014156517572700977, 0.10622674226760864, 0.010205940343439579, 0.005479741841554642, 0.004799934104084969, 0.08605539798736572, -0.018...
e95f4c27-588e-542e-88f6-cfe4cb3edff5
/documentation/fastembed/fastembed-postprocessing/
Multi-Vector Postprocessing
Using MUVERA in FastEmbed > Upload to Qdrant
documentation
true
Multi-Vector Postprocessing > Using MUVERA in FastEmbed > Upload to Qdrant Install `qdrant-client`. ```python pip install "qdrant-client>=1.14.2" ``` We'll use Qdrant running locally in a Docker container for this example. Alternatively, you can use [a free cluster](/documentation/cloud/create-cluster/#create-a-clu...
[ 0.058392561972141266, -0.04302908852696419, -0.013452553190290928, -0.05475664138793945, 0.06991387903690338, -0.032356422394514084, 0.07681617140769958, 0.009531295858323574, 0.05576780065894127, -0.004162488505244255, 0.009601933881640434, 0.0361432209610939, 0.04413756728172302, -0.0325...
e96b1eea-3b86-5fd5-ab77-59095d7c3536
/articles/what-is-a-vector-database/
What is a Vector Database?
The Architecture of a Vector Database > Points > 2. The Vector(s): One or More Representations of the Data
article
false
What is a Vector Database? > The Architecture of a Vector Database > Points > 2. The Vector(s): One or More Representations of the Data The vector is the numerical representation that powers similarity search (see [What Is a Vector?](#what-is-a-vector)). It is possible to attach more than one vector to a single point...
[ 0.03298228606581688, -0.059028200805187225, -0.0668046772480011, -0.043320149183273315, 0.03481755778193474, -0.017351225018501282, 0.06739094853401184, 0.0346335843205452, 0.0695834681391716, -0.02424442768096924, 0.057811036705970764, 0.02490435168147087, 0.08108557760715485, -0.01884342...
e9839ff1-850f-5ef6-bd52-479ea692dc22
/documentation/tutorials-build-essentials/video-anomaly-edge-part-1/
Video Anomaly Detection Part 1: Architecture, Twelve Labs, and NVIDIA VSS
Why kNN Beats Classifiers for Anomaly Detection
documentation
true
Video Anomaly Detection Part 1: Architecture, Twelve Labs, and NVIDIA VSS > Why kNN Beats Classifiers for Anomaly Detection Before jumping into the code, it's important to understand *why* we use kNN vector search instead of a traditional classifier for anomaly detection. Binary classifiers require labeled examples o...
[ 0.037506770342588425, -0.04764488339424133, -0.08383957296609879, -0.058097243309020996, 0.05937295779585838, -0.035104844719171524, 0.03955204784870148, 0.005702164489775896, 0.06356500089168549, -0.0389559231698513, 0.041063789278268814, 0.02660951390862465, 0.07669515907764435, -0.03124...
e985d951-e82e-5f8d-9ea8-b6ba771bed13
/documentation/search-tuning/when-your-collection-outgrows-ram/
When Your Collection Outgrows RAM
If TurboQuant Is Not the Right Fit > Use turbo4 in Constrained Environments
documentation
false
When Your Collection Outgrows RAM > If TurboQuant Is Not the Right Fit > Use turbo4 in Constrained Environments The tradeoff changes once no full-precision copy has to survive anywhere, which is the case `turbo4` with no further quantization makes. A separate benchmark across five BEIR datasets, comparing datatypes an...
[ 0.047401782125234604, -0.029737846925854683, -0.04073371738195419, -0.0323801152408123, 0.06878075748682022, 0.0020257614087313414, 0.044723112136125565, 0.013395475223660469, 0.10117656737565994, -0.029599107801914215, 0.020853765308856964, 0.016924003139138222, 0.06706569343805313, -0.01...
e9ac408c-d1af-58f9-801d-7e80bd087251
/articles/muvera-embeddings/
MUVERA: Making Multivectors More Performant
How MUVERA Embeddings are created > SimHash projection for clustering
article
false
MUVERA: Making Multivectors More Performant > How MUVERA Embeddings are created > SimHash projection for clustering SimHash is one of the LSH techniques that uses random hyperplanes to hash input vectors into binary codes. In the first step, this method chooses `k_sim` random hyperplanes (normal vectors) from a standa...
[ 0.08584453910589218, -0.025565605610609055, -0.04468924179673195, -0.033083509653806686, 0.10556010156869888, -0.018233636394143105, 0.06363511830568314, -0.002388747874647379, 0.06335955113172531, 0.007840731181204319, 0.04849907383322716, 0.021463798359036446, 0.061099421232938766, -0.02...
e9ad6802-f73d-539a-ba1e-4a7feb6e8c3e
/documentation/frameworks/rocketride/
RocketRide
documentation
false
RocketRide [RocketRide](https://github.com/rocketride-org/rocketride-server) is an open-source AI development environment and runtime for building, running, and integrating AI systems. Pipelines are portable JSON and execute on a multithreaded C++ engine. You can compose them in the visual IDE, run them from the CLI, ...
[ 0.033605221658945084, -0.003091386752203107, -0.03915935382246971, -0.03580633923411369, 0.08125633001327515, -0.06814973801374435, 0.07247095555067062, 0.03234263136982918, 0.016653578728437424, -0.041010767221450806, 0.059923723340034485, -0.010450496338307858, 0.016592271625995636, -0.0...
e9da991f-b2f4-5abc-94f6-4419b170801a
/course/beginners/module-6/ranking-and-reranking/
Ranking: Score Boosting and Reranking
Reranking
course
false
Ranking: Score Boosting and Reranking > Reranking A reranker reorders a shortlist with a slower, more accurate model. It reads the query and one document together, rather than comparing two vectors computed separately, which is what makes it both more accurate and too expensive to run over a whole collection. Reranki...
[ 0.03486800938844681, -0.026198510080575943, -0.05016285553574562, -0.052755389362573624, 0.06396885961294174, -0.013753186911344528, 0.0560004897415638, 0.00837902445346117, 0.06605342030525208, -0.01985055021941662, 0.025775719434022903, 0.015888923779129982, 0.057716004550457, -0.0033570...
e9dabf08-d882-50d1-b471-4e42aeac22e6
/documentation/edge/edge-api/shard-lifecycle/
Shard Lifecycle
Create a New Edge Shard
documentation
true
Shard Lifecycle > Create a New Edge Shard Creates a new Edge Shard at `path` using the supplied configuration. ```python @staticmethod def create(path: str, config: EdgeConfig) -> EdgeShard ``` | Parameter | Description | | --- | --- | | `path` | Path to the shard directory. Must not already contain segment data. | ...
[ 0.0043239900842309, -0.03200915828347206, -0.04351459816098213, -0.030847694724798203, 0.06466878205537796, -0.07253511995077133, 0.034453872591257095, 0.06189039722084999, 0.036882027983665466, -0.03230643272399902, 0.03555827960371971, 0.011535939760506153, 0.023308290168642998, -0.04699...
e9db342a-2fa3-5325-b019-510a0b634a36
/course/beginners/module-0/qdrant-cloud/
Qdrant Setup
Access the Web UI
course
false
Qdrant Setup > Access the Web UI The **Web UI** is a dashboard for looking at your data and running searches without writing code. It's the fastest way to see what's happening inside your cluster while you learn. 1. Select **Cluster UI** in the top corner of the cluster page to open the dashboard. ![Screenshot of th...
[ 0.03402495011687279, -0.049037594348192215, -0.055389221757650375, -0.0371074341237545, 0.039558954536914825, 0.003920892719179392, 0.04583851248025894, 0.032371290028095245, 0.016562148928642273, -0.028195811435580254, 0.01720830425620079, 0.05450647696852684, 0.10135091841220856, -0.0068...
e9dd4f80-d9cb-5a77-bf96-b537adbad8e0
/documentation/ops-configuration/configuration/
Configuration
Environment Variables
documentation
true
Configuration > Environment Variables You can also configure Qdrant using environment variables, which always take the highest priority and override any file-based settings. Environment variables follow this format: they should be prefixed with `QDRANT__`, and nested properties should be separated by double underscor...
[ 0.062150802463293076, -0.02956625632941723, -0.02560097724199295, -0.03918595612049103, 0.03867360204458237, -0.052421651780605316, 0.01713920757174492, 0.058991629630327225, 0.05371500551700592, -0.0043716407380998135, -0.005055427551269531, -0.019021622836589813, 0.08233974128961563, -0....
e9deab56-69ff-5384-9d01-7de8e81c808b
/documentation/tutorials-search-engineering/pdf-retrieval-at-scale/
Multivector Document Retrieval
Upload to Qdrant
documentation
false
Multivector Document Retrieval > Upload to Qdrant The upload process is trivial; the only thing to pay attention to is the compute cost for ColPali and ColQwen2 models. In low-resource environments, it's recommended to use a smaller batch size for embedding and mean pooling. Full version of the upload code is availa...
[ -0.0187235027551651, -0.042500995099544525, -0.05322230979800224, -0.07557713240385056, 0.07289500534534454, -0.023102033883333206, 0.0728907361626625, 0.07333850860595703, 0.05528406426310539, -0.020559431985020638, 0.017226992174983025, 0.0018245602259412408, 0.031612247228622437, -0.050...
e9eab568-c269-5307-88e8-3d702f4865ca
/documentation/production-operations/vector-search-in-production/
Vector Search in Production: Scaling, HA & Tuning Guide
4. Ensuring Disaster Recovery With Database Backups and Snapshots
documentation
false
Vector Search in Production: Scaling, HA & Tuning Guide > 4. Ensuring Disaster Recovery With Database Backups and Snapshots || |:-:| |**"I tried to restore a snapshot, and now everything's broken."**| **Use Case:** A digital publisher endured a catastrophic outage and attempted to restore from backups. They discovere...
[ 0.036059193313121796, -0.047431956976652145, -0.05247560143470764, -0.020840968936681747, 0.05620523914694786, -0.013999479822814465, 0.04179994761943817, -0.015439586713910103, 0.07818572968244553, -0.02326151914894581, 0.009248471818864346, 0.018221242353320122, 0.03292938321828842, -0.0...
e9ec239a-37b5-5d40-a436-0da97748b770
/documentation/manage-data/quantization/
Quantization
Binary Quantization > Rescoring and Oversampling Results
documentation
false
Quantization > Binary Quantization > Rescoring and Oversampling Results **Limits of this run.** Treat the ranges as indicative, not as confidence intervals: 100 queries is a small sample. The run used 1-bit binary quantization only; the [1.5-bit and 2-bit encodings](#15-bit-and-2-bit-quantization) and [asymmetric quan...
[ 0.057005785405635834, -0.06545870751142502, -0.02073388546705246, -0.053711265325546265, 0.06344035267829895, 0.013563492335379124, 0.03344907984137535, -0.007424734998494387, 0.09642311930656433, -0.02855485863983631, 0.02143084816634655, 0.01713331788778305, 0.06707203388214111, -0.02217...
e9f2d55a-1739-5fed-930e-ef50cd3eef63
/articles/gridstore-key-value-storage/
Introducing Gridstore: Qdrant's Custom Key-Value Store
Gridstore Architecture: Three Main Components > 1. The Data Layer for Fast Retrieval
article
false
Introducing Gridstore: Qdrant's Custom Key-Value Store > Gridstore Architecture: Three Main Components > 1. The Data Layer for Fast Retrieval At the core of Gridstore is **The Data Layer**, which is designed to store and retrieve values quickly based on their keys. This layer allows us to do efficient reads and lets u...
[ 0.011017447337508202, -0.00552374916151166, -0.03852386772632599, -0.023266732692718506, 0.06411145627498627, 0.011960359290242195, 0.02335713617503643, 0.021295860409736633, 0.06980728358030319, -0.02889193966984749, 0.0035840701311826706, -0.012713422998785973, 0.05489100143313408, 0.015...
ea1c0686-1257-50a2-ac47-2f7604208e66
/articles/filterable-hnsw/
Filterable HNSW Without Recall Loss
article
false
Filterable HNSW Without Recall Loss > **Note:** This article describes the original design for filterable HNSW, proposed in 2019. It is not a complete description of how Qdrant performs filtered vector search today. For learning more, see our Qdrant Beginners course: [Fast Approximate Search: HNSW](/course/beginners/m...
[ 0.07454654574394226, -0.056998513638973236, -0.06388512998819351, -0.034895796328783035, 0.08233574032783508, -0.03891121968626976, 0.03852017968893051, 0.020067673176527023, 0.050306838005781174, -0.022916819900274277, 0.04780520871281624, 0.04513419046998024, 0.08551821857690811, -0.0415...
ea20ebbd-a3a8-50db-90ef-ec716cc02106
/documentation/private-cloud/api-reference/
API Reference
qdrant.io/v1 > Resource Types > QdrantCluster
documentation
false
API Reference > qdrant.io/v1 > Resource Types > QdrantCluster QdrantCluster is the Schema for the qdrantclusters API _Appears in:_ - [QdrantClusterList](#qdrantclusterlist) | Field | Description | Default | Validation | | --- | --- | --- | --- | | `apiVersion` _string_ | `qdrant.io/v1` | | | | `kind` _string_ | `Qdr...
[ 0.03560888394713402, -0.034890953451395035, -0.046203650534152985, -0.06410188972949982, 0.07257761806249619, -0.042024001479148865, 0.07162706553936005, 0.07455487549304962, 0.04340845346450806, -0.058640070259571075, 0.05778934434056282, 0.0032510284800082445, 0.0548301599919796, -0.0063...
ea319566-d5cd-5976-b364-f680f2a42c3c
/documentation/upgrades/
Upgrades
Kubernetes
documentation
true
Upgrades > Kubernetes If you are using Helm, you can upgrade your Qdrant cluster by upgrading the Helm release. For example: ```bash helm upgrade qdrant qdrant/qdrant --version <target-version> -n <namespace> ``` Kubernetes will automatically perform a rolling update of the Qdrant StatefulSet.
[ 0.04998791962862015, -0.06069750338792801, -0.013980618678033352, -0.036267466843128204, 0.06555313616991043, -0.0701666995882988, 0.04926421120762825, 0.01634873077273369, 0.025330692529678345, -0.023668894544243813, 0.04102402552962303, 0.03566369786858559, 0.05271514877676964, -0.035295...
ea4ec6c3-952a-5581-8538-2be11bdbdb35
/documentation/tutorials-search-engineering/sparse-embeddings-ecommerce-part-2/
Fine-Tuning Sparse Embeddings for E-Commerce Search | Part 2: Training SPLADE on Modal
Setting Up the Modal App
documentation
true
Fine-Tuning Sparse Embeddings for E-Commerce Search | Part 2: Training SPLADE on Modal > Setting Up the Modal App Modal gives us serverless GPUs. No provisioning, no idle hardware, pay-per-second billing. Here's the app configuration: ```python import modal app = modal.App("esci-sparse-encoder") # Persistent storag...
[ 0.002610887633636594, -0.0450664758682251, -0.01965472660958767, -0.0030821028631180525, 0.022784896194934845, -0.013717507012188435, 0.05076141282916069, 0.03011169657111168, 0.06044452637434006, -0.013938786461949348, -0.0022353518288582563, 0.009836905635893345, 0.029255563393235207, -0...
ea5e05e0-8096-5d15-96d7-b562301e6418
/documentation/hybrid-cloud/hybrid-cloud-cluster-creation/
Create a Cluster
Scheduling Configuration
documentation
false
Create a Cluster > Scheduling Configuration When creating or editing a cluster, you can configure Pod to node scheduling for your Kubernetes cluster. This can be useful to ensure that the Qdrant databases will run on dedicated nodes. You can configure the necessary nodeSelectors and tolerations and topologySpreadConst...
[ 0.028859062120318413, -0.04287954419851303, -0.04657416418194771, -0.03147109970450401, 0.03665865212678909, -0.06350576132535934, 0.04973573237657547, 0.068787582218647, 0.040685128420591354, -0.007406159769743681, 0.06116821616888046, -0.025794707238674164, 0.04679398611187935, -0.028146...
ea7b0b06-0f7a-5935-be02-56b98c9c1f2a
/documentation/frameworks/rig-rs/
Rig-rs
documentation
true
Rig-rs [Rig](http://rig.rs) is a Rust library for building scalable, modular, and ergonomic LLM-powered applications. It has full support for LLM completion and embedding workflows with minimal boiler plate. Rig supports Qdrant as a vectorstore to ingest and search for documents semantically. ## Installation ```con...
[ 0.0318860299885273, -0.022223861888051033, -0.06440610438585281, -0.038927990943193436, 0.06277208775281906, -0.035416796803474426, 0.08479003608226776, 0.004475738387554884, 0.041587237268686295, -0.021797649562358856, 0.020913956686854362, -0.05238945782184601, 0.03119548037648201, -0.02...
ea84ff8a-cb64-51d2-9cd6-3f2f28ffa442
/documentation/private-cloud/api-reference/
API Reference
qdrant.io/v1 > Resource Types > PersistentVolumeClaimTemplate
documentation
false
API Reference > qdrant.io/v1 > Resource Types > PersistentVolumeClaimTemplate _Appears in:_ - [Storage](#storage) | Field | Description | Default | Validation | | --- | --- | --- | --- | | `metadata` _[TemplateMetadata](#templatemetadata)_ | Refer to Kubernetes API documentation for fields of `metadata`. | | | | `s...
[ 0.01378522627055645, -0.03699926659464836, -0.02690882980823517, -0.061364490538835526, 0.04587816819548607, -0.07229458540678024, 0.05351080372929573, 0.07016987353563309, 0.04324141889810562, -0.06475969403982162, 0.01969321258366108, 0.0010813298868015409, 0.047389112412929535, -0.02940...
ea8a2b12-4cae-50f1-92e3-320639c31380
/documentation/ops-configuration/quotas/
Resource Quotas
Configuring Quotas
documentation
true
Resource Quotas > Configuring Quotas Quotas are disabled by default. You can configure quotas per node, or apply them to every node in a cluster via the API. To apply quotas to a node, configure `storage.quotas` in the node's configuration file: ```yaml storage: quotas: enabled: true max_disk_usage_percent...
[ 0.040345337241888046, -0.02496528811752796, -0.00784052163362503, -0.05323481559753418, 0.05903332680463791, -0.054729074239730835, 0.04185060039162636, 0.06064523011445999, 0.07554679363965988, -0.02484382688999176, -0.010925356298685074, 0.006421124562621117, 0.04449501633644104, -0.0732...
eaa15d3b-a90f-55c1-93bc-ea1a328743c4
/documentation/manage-data/quantization/
Quantization
Binary Quantization > Asymmetric Quantization
documentation
false
Quantization > Binary Quantization > Asymmetric Quantization *Available as of v1.15.0* The **Asymmetric Quantization** technique allows Qdrant to use different vector encoding algorithms for stored vectors and queries. A particularly interesting combination is binary stored vectors and Scalar quantized queries. With ...
[ 0.07429232448339462, -0.04707875847816467, -0.04600611329078674, -0.05061020329594612, 0.03876814618706703, -0.0011073659406974912, 0.01908804476261139, -0.013059122487902641, 0.1334010362625122, -0.016767434775829315, 0.024562370032072067, 0.002921671373769641, 0.08786801993846893, -0.002...
eab77534-0258-530f-9047-f71be88d5117
/documentation/search-precision/automate-filtering-with-llms/
LLM-Powered Filter Automation
Using Instructor to generate Qdrant filters
documentation
true
LLM-Powered Filter Automation > Using Instructor to generate Qdrant filters Instructor has some helper methods to decorate the LLM APIs, so you can interact with them as if you were using their normal SDKs. In case of Anthropic, you just pass an instance of `Anthropic` class to the `from_anthropic` function: ```pytho...
[ 0.04225429892539978, -0.03074204921722412, -0.05078781023621559, -0.021733548492193222, 0.05775683745741844, -0.011942469514906406, 0.050723131746053696, 0.02827197127044201, 0.058832235634326935, -0.004473830107599497, 0.02016107738018036, 0.031015882268548012, 0.05662815645337105, -0.013...
eabd2975-82d9-5640-a1c1-086c568e3541
/course/essentials/day-1/chunking-strategies/
Text Chunking Strategies
Chunking Strategies: The Shape Matters > 3. Paragraph-Based Chunking
course
true
Text Chunking Strategies > Chunking Strategies: The Shape Matters > 3. Paragraph-Based Chunking **The Approach:** Split on paragraph breaks, leveraging existing document structure. Example text with paragraphs: Paragraph 1: "HNSW (Hierarchical Navigable Small World) is a graph-based algorithm for approximate nearest...
[ 0.013163060881197453, -0.026693953201174736, -0.08538401871919632, -0.04154939204454422, 0.06470507383346558, -0.04010912775993347, 0.045074716210365295, 0.021841933950781822, 0.0348697230219841, -0.008706153370440006, 0.02378777787089348, -0.0027921011205762625, 0.043020378798246384, -0.0...
eac13d29-4be1-55e7-9959-87bf3d35e22f
/articles/agentic-rag/
What is Agentic RAG? Building Agents with Qdrant
What Agents Need from the Retrieval Layer
article
false
What is Agentic RAG? Building Agents with Qdrant > What Agents Need from the Retrieval Layer - **Measured quality and cost.** Track [retrieval quality](/documentation/search-evaluation/retrieval-relevance/), such as recall@k, and the cost per task, in tokens and end-to-end time. An agent that retrieves poorly takes mo...
[ -0.018960369750857353, -0.04365571215748787, -0.060893502086400986, -0.07475686818361282, 0.039048291742801666, -0.05184977129101753, 0.025388536974787712, 0.019888563081622124, 0.1133490577340126, -0.04064634442329407, 0.01977345161139965, 0.02825661189854145, 0.0659552812576294, -0.01144...
eadd845c-2c8d-58a1-99f7-1cdba2bffa3b
/documentation/tutorials-develop/hybrid-search-fastembed/
Build a Hybrid Search API
Upload data to Qdrant
documentation
true
Build a Hybrid Search API > Upload data to Qdrant 1. Install the official Python client to best interact with Qdrant. ```bash pip install "qdrant-client[fastembed]>=1.14.2" ``` > **Note:** This tutorial requires fastembed of version >=0.6.1. At this point, you should have startup records in the `startups_demo.json`...
[ 0.021313145756721497, -0.05207740142941475, -0.0518123023211956, -0.015872783958911896, 0.0850716307759285, -0.03712695837020874, 0.09375584870576859, 0.03619856759905815, 0.08077587932348251, -0.03352511301636696, 0.016225354745984077, 0.020481884479522705, 0.05877218768000603, -0.0283361...
eb30d084-f834-5108-895a-3a774f50a3e2
/articles/modern-sparse-neural-retrieval/
Modern Sparse Neural Retrieval: From DeepCT to SPLADE++
Sparse Neural Retrieval Evolution > Sparse Neural Retrieval on Relevance Objective
article
false
Modern Sparse Neural Retrieval: From DeepCT to SPLADE++ > Sparse Neural Retrieval Evolution > Sparse Neural Retrieval on Relevance Objective It’s much easier to define whether a document as a whole is relevant or irrelevant to a query. That’s why the [`DeepImpact`](https://arxiv.org/pdf/2104.12016) Sparse Neural Retr...
[ 0.06351617723703384, -0.06563454121351242, -0.04908473417162895, -0.017004521563649178, 0.07405997812747955, -0.01856941729784012, 0.07092754542827606, 0.030760394409298897, 0.06329546123743057, -0.04121566563844681, -0.013530716300010681, 0.03895903006196022, 0.04421047493815422, -0.04451...
eb3ca3c2-cf4f-5a5d-81c5-e6a17c459a62
/articles/bm42/
BM42: Attention-Based Sparse Embeddings for Hybrid Search
The best of both worlds
article
false
BM42: Attention-Based Sparse Embeddings for Hybrid Search > The best of both worlds As previously mentioned, `IDF` is the most important part of the BM25 formula. In fact it is so important, that we decided to build its calculation into the Qdrant engine itself. Check out our latest [release notes](https://github.com/...
[ 0.0441250205039978, -0.020003175362944603, -0.02661474235355854, -0.02675658091902733, 0.0492355152964592, -0.033436793833971024, 0.057103730738162994, 0.037135884165763855, 0.051790494471788406, -0.024766484275460243, 0.05291125550866127, 0.009033260866999626, 0.050226569175720215, -0.032...
eb4d0d3c-f4e7-53c8-bd83-3bfe536b4972
/course/beginners/module-2/core-data-model/
Core Data Model
Your Qdrant Cluster
course
true
Core Data Model > Your Qdrant Cluster To create a collection, you need a running Qdrant instance, or **cluster**. A cluster is a Qdrant deployment that stores your collections and handles requests. You can use Qdrant Cloud or run Qdrant yourself locally. Qdrant Cloud offers a free tier that takes about a minute to set...
[ 0.00667277118191123, -0.0019753577653318644, -0.04821379855275154, -0.05040505900979042, 0.06194916367530823, -0.03465910628437996, 0.07029293477535248, 0.03913305699825287, 0.07992003113031387, -0.057736363261938095, 0.03337721526622772, 0.0006985155050642788, 0.06342687457799911, -0.0100...
eb6ce737-7ee2-5e69-b111-859a4a119466
/documentation/security/
Security
TLS
documentation
true
Security > TLS *Available as of v1.2.0* TLS for encrypted connections can be enabled on your Qdrant instance to secure connections. Connections are unencrypted by default. This allows sniffing and MitM attacks. First make sure you have a certificate and private key for TLS, usually in `.pem` format. On your local m...
[ 0.05830955132842064, -0.07748545706272125, -0.046684734523296356, 0.0021818787790834904, 0.01113694254308939, -0.05995166301727295, 0.047126587480306625, 0.025636078789830208, 0.08039810508489609, -0.0065326206386089325, 0.007716526743024588, -0.03264292702078819, 0.058888018131256104, -0....
eb7a008c-c4f9-5c74-ab62-783cf72e96bf
/documentation/cloud/inference/
Inference
documentation
false
Inference [Inference](/documentation/inference/) is the process of creating vector embeddings from text, images, or other data types using a machine learning model. Qdrant Managed Cloud allows you to use inference directly in the cloud, without the need to set up and maintain your own inference infrastructure. You ca...
[ 0.06401882320642471, -0.04436836764216423, -0.04901111498475075, -0.0946514680981636, 0.061774492263793945, -0.006542702671140432, 0.1273932009935379, 0.06643953174352646, 0.06267476081848145, -0.007567231543362141, 0.02352163754403591, 0.006157535128295422, 0.0495198518037796, -0.03686700...
eb8abdc1-5025-592f-afe5-e7874430987a
/documentation/fastembed/
FastEmbed
Why is FastEmbed useful?
documentation
false
FastEmbed > Why is FastEmbed useful? - Light: Unlike other inference frameworks, such as PyTorch, FastEmbed requires very little external dependencies. Because it uses the ONNX runtime, it is perfect for serverless environments like AWS Lambda. - Fast: By using ONNX, FastEmbed ensures high-performance inference across...
[ 0.017425740137696266, -0.02680276706814766, -0.023616906255483627, -0.04875950142741203, 0.0672471895813942, -0.013801727443933487, 0.07572775334119797, 0.01367409247905016, 0.0038819191977381706, -0.017622215673327446, 0.017717648297548294, 0.014182872138917446, 0.029356585815548897, -0.0...
eb90867f-9c61-5b25-8ae3-5e0a5c7abf30
/documentation/tutorials-operations/incremental-embedding-updates/
Incremental Embedding Updates
Syncing with Documentation Changes
documentation
true
Incremental Embedding Updates > Syncing with Documentation Changes Your sync trigger could be a CI job on merge if your docs live in git, or a nightly cron job. The input of every sync with a documentation collection here is the **current full chunk list of the docs**. For a simple deterministic data prep pipeline it...
[ 0.027430785819888115, -0.05401766300201416, -0.04155854880809784, -0.03408024460077286, 0.055719614028930664, -0.0073080118745565414, 0.03817851096391678, -0.004144323989748955, 0.05093465745449066, -0.040071532130241394, 0.009148092940449715, 0.03918095678091049, 0.057564474642276764, -0....
ebacd537-fdc4-5e52-8a86-9fe9142738fe
/articles/gridstore-key-value-storage/
Introducing Gridstore: Qdrant's Custom Key-Value Store
Gridstore Architecture: Three Main Components
article
false
Introducing Gridstore: Qdrant's Custom Key-Value Store > Gridstore Architecture: Three Main Components Gridstore’s architecture is built around three key components that enable fast lookups and efficient space management: | Component | Description ...
[ 0.0028947500977665186, -0.012941757217049599, -0.04460986331105232, -0.03315301984548569, 0.07256629317998886, -0.025617552921175957, 0.037390973418951035, 0.0543951578438282, 0.047143857926130295, -0.0263155996799469, 0.024993259459733963, -0.005079720169305801, 0.05228438228368759, -0.00...
ebc7a540-62af-589d-8419-420811cd71dc
/documentation/edge/edge-quickstart/
Quickstart
Custom WAL Size
documentation
true
Quickstart > Custom WAL Size Qdrant Edge uses a Write-Ahead Log (WAL) to record every update before it's applied to storage. The WAL file is pre-allocated to 32 MB by default, inflating backup sizes and OS storage reports. To reduce the size, set `wal_options` on `EdgeConfig` when calling `new` or `load`. WAL options ...
[ 0.004938693717122078, -0.030484121292829514, -0.035157956182956696, -0.025383615866303444, 0.07255933433771133, -0.0846494659781456, 0.023706234991550446, 0.06268373876810074, 0.05923723801970482, -0.04081423953175545, 0.056734565645456314, 0.007724808994680643, 0.0310323778539896, -0.0386...
ebd52d38-8735-5b9f-b91e-c0a4f52d95da
/documentation/migrate-to-qdrant/from-redis/
From Redis
Run the Migration
documentation
true
From Redis > Run the Migration ```bash docker run --net=host --rm -it registry.cloud.qdrant.io/library/qdrant-migration redis \ --redis.index 'your-ft-index' \ --redis.addr 'localhost:6379' \ --qdrant.url 'https://your-instance.cloud.qdrant.io:6334' \ --qdrant.api-key 'your-qdrant-api-key' \ --qdra...
[ 0.03700106590986252, -0.037514183670282364, -0.029783399775624275, -0.03300708532333374, 0.07665906101465225, -0.07138492166996002, 0.0369485467672348, 0.04467155411839485, 0.060404278337955475, -0.03681270405650139, 0.046524111181497574, 0.007787453942000866, 0.037991441786289215, -0.0513...
ebfa1019-5593-5062-8a1c-0a20c67a90b7
/documentation/tutorials-develop/hybrid-search-fastembed/
Build a Hybrid Search API
Prepare sample dataset
documentation
true
Build a Hybrid Search API > Prepare sample dataset To conduct a hybrid search on startup descriptions, you must first encode the description data into vectors. Fastembed integration into qdrant client combines encoding and uploading into a single step. It also takes care of batching and parallelization, so you don't...
[ 0.031166324391961098, -0.05786343291401863, -0.05679469555616379, -0.054368194192647934, 0.07202424108982086, -0.027833230793476105, 0.04824439063668251, 0.045673251152038574, 0.03923732787370682, -0.015744294971227646, 0.029053237289190292, 0.02466527186334133, 0.09105920791625977, -0.033...
ebfe1484-0071-515f-9438-2215f70732bd
/documentation/fastembed/fastembed-colbert/
Working with ColBERT
Setup
documentation
true
Working with ColBERT > Setup Install `fastembed`. ```python pip install fastembed ``` Imports late interaction models for text embedding. ```python from fastembed import LateInteractionTextEmbedding ``` You can list which late interaction models are supported in FastEmbed. ```python LateInteractionTextEmbedding.l...
[ 0.04076981171965599, -0.021825412288308144, -0.026835545897483826, -0.050695449113845825, 0.06919757276773453, -0.043385911732912064, 0.0646241307258606, 0.04110609367489815, 0.01356226485222578, 0.006830074358731508, 0.03472308814525604, -0.007478646468371153, 0.05385703220963478, -0.0558...
ec0faabb-fbc9-5e9d-8615-3001624f810f
/documentation/migrate-to-qdrant/from-redis/
From Redis
Concept Mapping
documentation
false
From Redis > Concept Mapping | Redis | Qdrant | Notes | | :--- | :--- | :--- | | FT Index | Collection | One-to-one mapping | | Document | Point | Each document becomes a point | | Vector field | Vector | Named vectors are preserved | | Hash/JSON fields | Payload | Direct mapping | | Document key | Payload field | Sto...
[ 0.06908049434423447, -0.03890830650925636, -0.01401556096971035, -0.07110879570245743, 0.0613514706492424, -0.021161247044801712, 0.07155559957027435, 0.04697089269757271, 0.0683048889040947, -0.05808790773153305, 0.058249883353710175, 0.008100544102489948, 0.04812832549214363, -0.03982849...
ec17a2f3-730d-5cd0-b59a-128332460a4e
/documentation/data-synchronization/with-postgres/
Keeping Postgres and Qdrant in Sync
Tier 2: Transactional Outbox Pattern > Failure Modes
documentation
false
Keeping Postgres and Qdrant in Sync > Tier 2: Transactional Outbox Pattern > Failure Modes | Failure | Consequence | Mitigation | | :---- | :---- | :---- | | Qdrant is down | Events queue in `sync_outbox` | Worker retries with backoff on recovery | | Worker crashes | Events remain `pending` | Worker picks up on restar...
[ 0.04891928657889366, -0.031057344749569893, -0.03261111304163933, 0.0004975617048330605, 0.057488683611154556, -0.05604139342904091, 0.06321199238300323, 0.010750217363238335, 0.07137343287467957, -0.015796596184372902, 0.035972584038972855, 0.03389125317335129, 0.02440270036458969, -0.048...
ec1ab98b-9428-5c14-8333-5a8517f88146
/documentation/tutorials-search-engineering/food-discovery-recommendations/
Multimodal Food Search Demo
Using the demo
documentation
true
Multimodal Food Search Demo > Using the demo The Food Discovery Demo [is available online](https://food-discovery.qdrant.tech/), but if you prefer to run it locally, you can do it with Docker. The [README](https://github.com/qdrant/demo-food-discovery/blob/main/README.md) describes all the steps more in detail, but h...
[ 0.03315601497888565, -0.03798234462738037, -0.045729413628578186, -0.011868529953062534, 0.07489471137523651, -0.023157507181167603, 0.04803064838051796, 0.04829440265893936, 0.05993823707103729, -0.030391061678528786, 0.025900673121213913, 0.033356983214616776, 0.05533911660313606, -0.052...
ec27928a-3c7c-5289-880d-c14f07fe32f1
/articles/gridstore-key-value-storage/
Introducing Gridstore: Qdrant's Custom Key-Value Store
Gridstore Architecture: Three Main Components > 2. The Mask Layer Reuses Space
article
false
Introducing Gridstore: Qdrant's Custom Key-Value Store > Gridstore Architecture: Three Main Components > 2. The Mask Layer Reuses Space **The Mask Layer** helps Gridstore handle updates and deletions without the need for expensive data compaction. Instead of maintaining complex metadata for each block, Gridstore track...
[ 0.004864553455263376, -0.009355954825878143, -0.032735422253608704, -0.02552749402821064, 0.059501394629478455, -0.007946173660457134, -0.012796280905604362, 0.020479973405599594, 0.053736764937639236, -0.02602715976536274, 0.010114267468452454, 0.0026349732652306557, 0.07969027012586594, ...
ec45495f-f9bc-5387-a3e3-956f4447f89e
/documentation/tutorials-search-engineering/sparse-embeddings-ecommerce-part-3/
Fine-Tuning Sparse Embeddings for E-Commerce Search | Part 3: Evaluation and Hard Negatives
Production Latency
documentation
true
Fine-Tuning Sparse Embeddings for E-Commerce Search | Part 3: Evaluation and Hard Negatives > Production Latency A common concern: isn't running a transformer on every query slow? | Step | Latency | Note | |---|---|---| | Query encoding (SPLADE) | 10-20ms | Bottleneck | | Sparse retrieval (Qdrant) | <1ms | Negligible...
[ 0.038468096405267715, -0.04269140958786011, -0.06356731802225113, -0.05206594243645668, 0.02915782853960991, -0.024832619354128838, 0.05665839463472366, -0.008924558758735657, 0.0997999757528305, -0.028637319803237915, 0.044597215950489044, -0.008066998794674873, 0.062067385762929916, -0.0...
ec580090-ef07-5847-981c-8078e38109e7
/documentation/cloud-rbac/
Cloud RBAC
documentation
false
Cloud RBAC ## About Cloud RBAC Qdrant Cloud enables you to manage granular permissions for your cloud resources within the Qdrant Cloud console. This feature ensures that only authorized users have access to sensitive data and capabilities, covering the following areas: - Billing - Identity and Access Management - C...
[ 0.032577384263277054, -0.015021120198071003, -0.04962831735610962, -0.05048414319753647, 0.07238184660673141, -0.04479178041219711, 0.058100078254938126, 0.09670820832252502, 0.05549086630344391, -0.04797246679663658, 0.02988596260547638, 0.016470981761813164, 0.06590192019939423, -0.04290...
ec64cf6a-7b68-5d7d-ad6b-76ec0f4f7355
/documentation/ops-optimization/read-write-contention/
Read-Write Contention
Step 2: Try Smaller Batch Sizes
documentation
false
Read-Write Contention > Step 2: Try Smaller Batch Sizes When ingesting data into Qdrant, batching upserts is a best practice that significantly improves throughput by reducing the per-request overhead of individual insertions. The optimal batch size depends on your vector dimensions, payload size, and hardware. Reduc...
[ 0.008472951129078865, -0.03815066069364548, -0.057368457317352295, -0.04059090465307236, 0.03366971015930176, -0.029140694066882133, 0.02979743294417858, 0.02486453577876091, 0.0825556218624115, -0.016316834837198257, 0.022677583619952202, -0.012310270220041275, 0.04092779383063316, -0.012...
ec6e0972-dfb6-52ad-be7d-227e1e8a5fd5
/documentation/tutorials-operations/blue-green-deployment/
Blue-Green Cluster Deployment
Step 2: Load Data into the New Cluster > Option B: Use the Migration Tool
documentation
true
Blue-Green Cluster Deployment > Step 2: Load Data into the New Cluster > Option B: Use the Migration Tool If you prefer a streaming transfer over a one-shot restore, for example when you want to reshard as part of the move, use the [Qdrant Migration Tool](/documentation/migrate-to-qdrant/). ```shell docker run --rm -...
[ 0.04709043353796005, -0.090299591422081, -0.03634386137127876, -0.04042045772075653, 0.06481827050447464, -0.014457661658525467, 0.04782462865114212, 0.04560553655028343, 0.035274192690849304, -0.01113708782941103, 0.017870277166366577, -0.009381209500133991, 0.028552241623401642, -0.00133...
ec7ac173-d016-5d8a-8709-29141bc4b46f
/documentation/edge/edge-quickstart/
Quickstart
Create Facets
documentation
true
Quickstart > Create Facets To create facets on a payload field, use the [`facet`](/documentation/edge/edge-api/reading-data/#facet) method. ```python from qdrant_edge import FacetRequest facet_response = edge_shard.facet(FacetRequest(key="color", limit=10, exact=False)) ```
[ 0.0031045270152390003, -0.03284284844994545, -0.01067284680902958, -0.0814567282795906, 0.07004775106906891, -0.06157355010509491, -0.009189681150019169, 0.032041996717453, 0.060248397290706635, -0.047420281916856766, 0.044753264635801315, -0.007736638188362122, 0.0753185972571373, -0.0119...
ec881ed9-92f6-58d9-adfe-9dac0c5d2620
/documentation/tutorials-search-engineering/food-discovery-recommendations/
Multimodal Food Search Demo
General architecture
documentation
false
Multimodal Food Search Demo > General architecture We are happy to announce a refreshed version of our [Food Discovery Demo](https://food-discovery.qdrant.tech/). This time available as an open source project, so you can easily deploy it on your own and play with it. If you prefer to dive into the source code directl...
[ 0.013715765438973904, -0.026376912370324135, -0.04381958395242691, -0.01357471663504839, 0.10436074435710907, -0.03167853504419327, 0.09030789881944656, 0.06565672159194946, 0.031040215864777565, -0.0027807822916656733, 0.036569658666849136, 0.06863461434841156, 0.03222508355975151, -0.046...
ecac97f9-9e19-586d-a704-8e06705aac08
/documentation/manage-data/payload/
Payload
Create point with payload
documentation
true
Payload > Create point with payload REST API ([Schema](https://api.qdrant.tech/api-reference/points/upsert-points)) ```python from qdrant_client import QdrantClient, models client = QdrantClient(url="http://localhost:6333") client.upsert( collection_name="{collection_name}", points=[ models.PointStr...
[ 0.04680680111050606, -0.0032926492858678102, -0.0005373872700147331, -0.06371839344501495, 0.06504907459020615, -0.02884495072066784, 0.0016697710379958153, 0.018700264394283295, 0.03515097126364708, -0.016253583133220673, 0.07246319949626923, -0.016085484996438026, 0.03893493488430977, -0...
ecad1fd0-5d38-55a1-bc10-40172a9fde87
/documentation/cloud/encryption/
Encryption at Rest
Step 2: Create Your Encryption Key
documentation
false
Encryption at Rest > Step 2: Create Your Encryption Key Follow the instructions from Qdrant Support to complete the key setup for your cloud provider. ### AWS In AWS KMS, create a symmetric key in the same region as your cluster and grant Qdrant's account access to it using the identifier provided by Support. Copy t...
[ 0.032360680401325226, -0.026976782828569412, -0.0463721938431263, -0.024421289563179016, 0.04138313606381416, -0.020717913284897804, 0.052482545375823975, 0.036368630826473236, 0.07393500208854675, 0.004364007152616978, 0.0070143528282642365, -0.01106523908674717, 0.0677279531955719, -0.03...
eccb89f9-2f15-5437-82ea-c09af4c6b938
/articles/turboquant-quantization/
TurboQuant in Qdrant
What Qdrant Adds > Per-Coordinate Calibration (Anisotropy Compensation)
article
false
TurboQuant in Qdrant > What Qdrant Adds > Per-Coordinate Calibration (Anisotropy Compensation) Here, the per-coordinate `1/scale` collapses into the query, and the `⟨q, shift⟩` term is a single scalar that depends only on the query. Both are computed **once per query**. The hot path still scores the raw `b·D`-bit code...
[ 0.05511274188756943, -0.024840213358402252, -0.044423580169677734, -0.021744750440120697, 0.034548670053482056, -0.0008933098870329559, 0.024531232193112373, 0.02214900031685829, 0.10229857265949249, -0.024580515921115875, 0.030044354498386383, -0.01044271606951952, 0.04997904598712921, -0...
ecf8df77-b6fc-5a9c-81ea-a175d57e06d9
/course/essentials/day-5/universal-query-demo/
Demo: Universal Query for Hybrid Retrieval
Step 2: The Universal Query in Action > Add Fusion Stage
course
true
Demo: Universal Query for Hybrid Retrieval > Step 2: The Universal Query in Action > Add Fusion Stage Now we combine the two filtered candidate lists using Reciprocal Rank Fusion: ```python # Fusion stage - combines dense and sparse results fusion_query = models.Prefetch( prefetch=hybrid_query, query=models.F...
[ 0.033802781254053116, -0.018675945699214935, -0.06375135481357574, -0.052919235080480576, 0.06861117482185364, -0.021483298391103745, 0.06522106379270554, 0.04307816922664642, 0.06033904477953911, -0.030532823875546455, 0.0465206503868103, -0.004357394762337208, 0.06667105853557587, 0.0025...
ed11f52e-3784-5c0d-85b0-10ba0b462fbf
/documentation/private-cloud/api-reference/
API Reference
qdrant.io/v1 > Resource Types > GPU
documentation
false
API Reference > qdrant.io/v1 > Resource Types > GPU | Field | Description | Default | Validation | | --- | --- | --- | --- | | `groupsCount` _integer_ | GroupsCount is the amount of used vulkan "groups" of GPU.In other words, how many parallel points can be indexed by GPU.Optimal value might depend on the GPU model.Pr...
[ -0.01875118911266327, -0.01554142590612173, -0.06076190993189812, 0.003712922567501664, 0.10902930051088333, -0.021966755390167236, 0.058039866387844086, 0.06702801585197449, 0.07083305716514587, -0.05109228566288948, 0.08860793709754944, 0.04543282464146614, 0.0072391279973089695, -0.0384...
ed1b7c3b-365a-5845-8de0-cecf85af9b89
/documentation/edge/
Qdrant Edge
documentation
false
Qdrant Edge Qdrant Edge is in beta. The API and functionality may change in future releases. Qdrant Edge is a lightweight, embedded vector search engine for in-process retrieval with a minimal memory footprint and no background services. Qdrant Edge is designed for applications requiring low-latency vector search in ...
[ 0.01931919902563095, -0.046418800950050354, -0.0550430603325367, -0.022424373775720596, 0.08647177368402481, -0.04522988945245743, 0.10852482914924622, 0.014882405288517475, 0.07687206566333771, -0.02867281250655651, 0.021952513605356216, 0.03018520213663578, 0.025760343298316002, -0.01977...
ed264b25-9e8f-5897-92fc-ed3de4b94280
/course/essentials/day-2/collection-tuning-demo/
Demo: HNSW Performance Tuning
Performance Analysis
course
true
Demo: HNSW Performance Tuning > Performance Analysis Compare your results and see the effect of each optimization: ```python print("\n" + "=" * 60) print("FINAL PERFORMANCE SUMMARY") print("=" * 60) # Key metrics if unindexed_filter_time > 0 and indexed_filter_time > 0: index_speedup = unindexed_filter_time / in...
[ 0.03595013916492462, -0.04239322617650032, -0.03520052134990692, -0.03632728010416031, 0.051974378526210785, 0.0034945071674883366, 0.04037593677639961, 0.015717623755335808, 0.07080253213644028, -0.0035003230441361666, 0.031177517026662827, 0.02960699424147606, 0.07366659492254257, -0.021...
ed5b55a9-de70-5bb3-a322-47a4a44365b8
/documentation/tutorials-operations/prevent-unoptimized-usage/
Prevent Unoptimized Usage
Related Reading
documentation
false
Prevent Unoptimized Usage > Related Reading - A deep dive on the Qdrant optimizer and its impact on latency: [Tuning Qdrant Optimizer for Predictable Search Latency](/documentation/production-operations/predictable-search-latency/) - Both mechanisms compared side by side: [Query Indexed Data Only](/documentation/searc...
[ 0.02519514411687851, -0.03580106794834137, -0.06538659334182739, -0.02822493389248848, 0.05965708941221237, -0.01505167968571186, 0.04720950871706009, 0.01617988757789135, 0.07100985944271088, -0.01800667867064476, 0.04636643826961517, -0.013544009998440742, 0.061640121042728424, -0.023322...
ed9ecead-9812-5912-af1b-4031ab9e5ac9
/documentation/examples/rag-chatbot-vultr-dspy-ollama/
Private RAG Information Extraction Engine
Deploying Qdrant Hybrid Cloud on Vultr
documentation
false
Private RAG Information Extraction Engine > Deploying Qdrant Hybrid Cloud on Vultr All the services we are going to use in this tutorial will be running on [Vultr Kubernetes Engine](https://www.vultr.com/kubernetes/). That gives us a lot of flexibility in terms of scaling and managing the resources. Vultr manages the...
[ 0.03249730169773102, -0.002145295264199376, -0.07588312774896622, -0.016773389652371407, 0.07345154136419296, -0.0331357978284359, 0.10452485829591751, 0.06371255218982697, 0.07213980704545975, -0.030245041474699974, 0.03866684064269066, 0.021461954340338707, 0.05822383612394333, -0.019688...
edc2f111-5281-5667-a277-2e36afb75a7e
/documentation/tutorials-basics/reranking-hybrid-search/
Hybrid Search with Reranking
Conclusion
documentation
false
Hybrid Search with Reranking > Conclusion Reranking with late interaction models is a powerful tool that boosts the relevance of search results, especially when combined with hybrid search methods. While it can add some latency due to its complexity, applying it to a smaller, pre-filtered subset of results ensures bot...
[ 0.0042221457697451115, -0.02037099003791809, -0.09425589442253113, -0.0848582535982132, 0.08000162988901138, -0.032301515340805054, 0.08978114277124405, 0.010883219540119171, 0.09146192669868469, -0.0016555481124669313, 0.028458017855882645, -0.0018300939118489623, 0.07448183745145798, 0.0...
edc927a5-10a1-51ff-88ed-3694c7a2af22
/documentation/installation/
Installation
Client libraries
documentation
false
Installation > Client libraries In addition to the service, Qdrant provides a variety of client libraries for different programming languages. For a full list, see our [Client libraries](/documentation/interfaces/#client-libraries) documentation.
[ -0.007586054969578981, -0.033445511013269424, -0.04031888023018837, -0.02888442762196064, 0.0448472760617733, -0.06832268089056015, 0.08523966372013092, 0.04827255755662918, 0.08234596252441406, -0.03502088412642479, 0.028413891792297363, 0.0028318800032138824, 0.08180251717567444, -0.0377...
edd25405-0237-5f72-9346-72f3e25a55dc
/documentation/production-operations/memory-tiers/
Memory Tiers in Qdrant: What to Use and When
HNSW Inline Storage and Disk Space
documentation
false
Memory Tiers in Qdrant: What to Use and When > HNSW Inline Storage and Disk Space Test [HNSW inline storage](/documentation/ops-optimization/optimize/#inline-storage-in-hnsw-index) with quantization at your real vector dimensionality and graph connectivity before relying on it in production. Qdrant's inline HNSW stora...
[ 0.05672992020845413, -0.02228412590920925, -0.03328641504049301, -0.05318516120314598, 0.06054728105664253, -0.021685192361474037, 0.06713657826185226, 0.03742644190788269, 0.11552810668945312, -0.013984969817101955, 0.004711414687335491, 0.013524151407182217, 0.060511622577905655, -0.0061...
edff4863-da6b-5705-9bf1-64d38a248e8e
/documentation/tutorials-build-essentials/video-anomaly-edge-part-2/
Video Anomaly Detection Part 2: Edge-to-Cloud Pipeline
Mutable Shard Retention
documentation
false
Video Anomaly Detection Part 2: Edge-to-Cloud Pipeline > Mutable Shard Retention The mutable shard grows with every clip the edge processes. Between syncs, all those points are searched via brute-force (no HNSW index), so query latency degrades as the shard gets larger. If the cloud is unreachable for hours, the shard...
[ 0.0036966416519135237, -0.03828687593340874, -0.04230692237615585, -0.05442708730697632, 0.04315312206745148, -0.05365105718374252, 0.08475536853075027, 0.015817521139979362, 0.0496293269097805, -0.020079579204320908, 0.00883935671299696, 0.04160674288868904, 0.02703707478940487, -0.008342...
ee274a96-c270-5cb8-ab02-8319b9f62178
/documentation/private-cloud/api-reference/
API Reference
qdrant.io/v1 > Resource Types > NodePVCStatus
documentation
false
API Reference > qdrant.io/v1 > Resource Types > NodePVCStatus _Appears in:_ - [NodeStatus](#nodestatus) | Field | Description | Default | Validation | | --- | --- | --- | --- | | `storageClassName` _string_ | Name of the StorageClass used by the PVC | | Optional: \{\} | | `phase` _[PersistentVolumeClaimPhase](https...
[ 0.045624084770679474, -0.06568949669599533, -0.016881663352251053, -0.04222185164690018, 0.046169474720954895, -0.0420866459608078, 0.041991040110588074, 0.05138946324586868, 0.0454351082444191, -0.03987586498260498, 0.029441986232995987, -0.012532432563602924, 0.025974800810217857, -0.052...
ee326e3d-0dfa-5031-b046-a3e21d6cb20d
/documentation/scaling/vertical-scaling/
Vertical Scaling
RAM Sizing Guidelines
documentation
true
Vertical Scaling > RAM Sizing Guidelines RAM is the resource that most directly affects Qdrant's search performance, since search is fastest when vectors and indexes fit in memory. Exact RAM usage is difficult to predict precisely, but this formula gives a reasonable estimate for full-precision vectors kept in RAM: ...
[ 0.023447422310709953, -0.05473600700497627, -0.032392267137765884, -0.033912185579538345, 0.05094984173774719, -0.012088368646800518, 0.07095122337341309, 0.05171572417020798, 0.1212431937456131, -0.05509672313928604, 0.01916695199906826, 0.023812245577573776, 0.03182370960712433, -0.03918...
ee3b75aa-c3d1-58cd-83b2-a1293ec29c75
/articles/gridstore-key-value-storage/
Introducing Gridstore: Qdrant's Custom Key-Value Store
Gridstore in Production: Maintaining Data Integrity > Stability Through Idempotency: Recovering With WAL
article
false
Introducing Gridstore: Qdrant's Custom Key-Value Store > Gridstore in Production: Maintaining Data Integrity > Stability Through Idempotency: Recovering With WAL To guard against these risks, Qdrant relies on a [**Write-Ahead Log (WAL)**](/documentation/manage-data/storage/). Before committing an operation, Qdrant ens...
[ 0.01846909336745739, -0.024739304557442665, -0.014926202595233917, -0.03640178218483925, 0.04864335432648659, -0.02369772642850876, 0.03992759808897972, 0.009379392489790916, 0.06352321058511734, -0.004212041385471821, 0.052582595497369766, -0.012583077885210514, 0.04807398095726967, -0.03...
ee40c13f-e6dc-549c-823c-fecdd20fb63a
/articles/what-is-a-vector-database/
What is a Vector Database?
The Architecture of a Vector Database > Collections
article
false
What is a Vector Database? > The Architecture of a Vector Database > Collections A [collection](https://qdrant.tech/documentation/manage-data/collections/) is essentially a set of points that you can search over together. Within a collection, vectors of the same name must share the same dimensionality and be compared ...
[ 0.010953719727694988, -0.04960943013429642, -0.06681360304355621, -0.04630880802869797, 0.03895328566431999, -0.043752461671829224, 0.06596221029758453, 0.04689204692840576, 0.09215883165597916, -0.02334488369524479, 0.03631371632218361, 0.03585737571120262, 0.058872513473033905, 0.0047488...
ee673bfc-d5bc-5688-a0bc-0de5a6699287
/course/multi-vector-search/module-3/muvera/
MUVERA
What's Next
course
false
MUVERA > What's Next You've learned why MaxSim's asymmetry prevents HNSW from efficiently indexing multi-vector representations, and how MUVERA solves this with single-vector approximations. By combining MUVERA's fast retrieval with multi-vector reranking, you can achieve significant speedups while maintaining search ...
[ 0.05868950113654137, -0.04895244538784027, -0.04643045365810394, -0.0402069091796875, 0.0626569464802742, -0.05108537897467613, 0.05934316664934158, -0.009795304387807846, 0.057343751192092896, 0.00347249791957438, 0.013584061525762081, 0.059666477143764496, 0.053494229912757874, -0.029543...
ee7a3f5c-f6fa-5a74-8f03-1865328bffe5
/course/essentials/day-1/chunking-strategies/
Text Chunking Strategies
Chunking Strategies: The Shape Matters > 4. Sliding Window Chunking
course
true
Text Chunking Strategies > Chunking Strategies: The Shape Matters > 4. Sliding Window Chunking **The Approach:** Create overlapping chunks to maintain context continuity. Example text: "HNSW builds a multi-layer graph where each node represents a vector. The algorithm starts by inserting vectors into the bottom layer...
[ 0.006306531839072704, -0.06131090968847275, -0.07174237072467804, -0.05053368955850601, 0.07272772490978241, -0.029161861166357994, 0.06476577371358871, 0.012001316994428635, 0.05131155252456665, -0.02015003375709057, 0.02312667854130268, -0.031276583671569824, 0.06673745810985565, -0.0349...
ee82d271-6cf6-5c17-8aac-09b15c4e6758
/documentation/embeddings/mistral/
Mistral
Embed Queries
documentation
true
Mistral > Embed Queries Mistral models have no separate query mode. Embed the query with the same model you used for the documents: ```python query_vector = ( mistral_client.embeddings.create(model=MODEL, inputs=["What is Qdrant?"]) .data[0] .embedding ) assert query_vector is not None hits = client.quer...
[ 0.02886735461652279, -0.03404071927070618, -0.011967316269874573, 0.018880613148212433, 0.07367721199989319, -0.0055444915778934956, 0.06160321086645126, 0.03871867060661316, 0.012845694087445736, -0.015033934265375137, 0.027639372274279594, -0.01987472176551819, 0.06542900204658508, -0.02...
ee9f5992-1829-5670-974a-f5e075eeb985
/course/multi-vector-search/module-1/use-cases-multi-vector/
Use Cases for Multi-Vector Search
When NOT to Use Multi-Vector Search
course
false
Use Cases for Multi-Vector Search > When NOT to Use Multi-Vector Search Multi-vector search isn't always necessary. Two scenarios where single-vector embeddings are preferable: **Simple, broad queries**: When you're searching for general topical relevance rather than verifying specific requirements, single-vector emb...
[ 0.014696457423269749, -0.056453023105859756, -0.05402228981256485, -0.05065489560365677, 0.046241238713264465, -0.04041784629225731, 0.051321085542440414, 0.015189649537205696, 0.0782516747713089, -0.0390104204416275, 0.014958621002733707, 0.01718231290578842, 0.061553917825222015, -0.0305...
eea29096-ac55-500c-bc67-9c2ea9ab40de
/documentation/manage-data/bulk-upload/
Bulk Upload
Create Payload Indexes Before Ingesting Data
documentation
false
Bulk Upload > Create Payload Indexes Before Ingesting Data If your collection uses payload indexes, [create them](/documentation/manage-data/indexing/#create-a-payload-index) before you start uploading points. Qdrant builds extra HNSW links for each payload index to optimize filtered vector search quality. If you add ...
[ 0.03265773504972458, -0.035845573991537094, -0.02706620842218399, -0.027332132682204247, 0.07750198245048523, -0.037210140377283096, 0.05880089849233627, 0.045234162360429764, 0.07232210785150528, -0.0337902195751667, 0.042672738432884216, 0.043854814022779465, 0.020160414278507233, -0.022...
eea492de-9f50-5e17-b92a-8981b70af8b4
/documentation/examples/rag-chatbot-vultr-dspy-ollama/
Private RAG Information Extraction Engine
Implementing the information extraction pipeline > Deploying the extraction pipeline
documentation
false
Private RAG Information Extraction Engine > Implementing the information extraction pipeline > Deploying the extraction pipeline Vultr gives us a lot of flexibility in terms of deploying the applications. Perfectly, we would use the Kubernetes cluster we set up earlier to run it. The deployment is as simple as runnin...
[ 0.014389585703611374, -0.05053381249308586, -0.046868711709976196, -0.007509894669055939, 0.11133235692977905, -0.010453358292579651, 0.06725950539112091, 0.037459846585989, 0.04971995949745178, -0.0065026553347706795, 0.05058083310723305, 0.032966308295726776, 0.05529998615384102, -0.0113...
eeb02d6a-bc70-52e2-8104-0c11b5588e63
/documentation/tutorials-develop/hybrid-search-fastembed/
Build a Hybrid Search API
Run Qdrant in Docker
documentation
true
Build a Hybrid Search API > Run Qdrant in Docker Next, you need to manage all of your data using a vector engine. Qdrant lets you store, update or delete created vectors. Most importantly, it lets you search for the nearest vectors via a convenient API. > **Note:** Before you begin, create a project directory and a v...
[ 0.03575999662280083, -0.042619556188583374, -0.029601076617836952, -0.009841701947152615, 0.05089099705219269, -0.016622208058834076, 0.07703176140785217, 0.010575628839433193, 0.07720321416854858, -0.04997476562857628, 0.01702708564698696, 0.011755629442632198, 0.0859287828207016, -0.0234...
eebeb0ce-0173-5775-be66-f66fb7a4598c
/documentation/search-precision/reranking-semantic-search/
Reranking for Better Search
Conclusion
documentation
false
Reranking for Better Search > Conclusion Reranking is a powerful way to boost the relevance and precision of search results in RAG systems. It comes with a trade-off: every query makes an extra call to the reranker, which adds latency and, with a hosted model like Cohere Rerank, usage-based costs. This guide demonstr...
[ 0.0033171905670315027, -0.05347226932644844, -0.08027977496385574, -0.07044916599988937, 0.06450946629047394, -0.0479297898709774, 0.06330043077468872, 0.043764423578977585, 0.08401259034872055, -0.02815609984099865, 0.008260853588581085, 0.01553873997181654, 0.07554085552692413, -0.003914...
eeccce60-7083-520b-8ea9-d0b2502d7497
/documentation/manage-data/quantization/
Quantization
Setting Up Quantization in Qdrant > Searching with Quantization
documentation
true
Quantization > Setting Up Quantization in Qdrant > Searching with Quantization Once you have configured quantization for a collection, you don't need to do anything extra to search with quantization. Qdrant will automatically use quantized vectors if they are available. However, there are a few options that you can u...
[ 0.04734199866652489, -0.0727161318063736, -0.02226170152425766, -0.030328046530485153, 0.06354587525129318, -0.00930624920874834, 0.03604405000805855, -0.009222215041518211, 0.093520388007164, -0.04381201043725014, 0.03528270497918129, 0.01855481043457985, 0.05234905704855919, -0.028555342...
eee31286-0bea-58db-a491-6b7c63fd2e51
/documentation/hybrid-cloud/platform-deployment-options/
Deployment Platforms
OVHcloud
documentation
false
Deployment Platforms > OVHcloud [Service Managed Kubernetes](https://www.ovhcloud.com/en-in/public-cloud/kubernetes/), powered by OVH Public Cloud Instances, a leading European cloud provider. With OVHcloud Load Balancers and disks built in. OVHcloud Managed Kubernetes provides high availability, compliance, and CNCF ...
[ 0.01274081040173769, -0.030403394252061844, -0.043024029582738876, -0.049342572689056396, 0.05236094072461128, -0.05538365617394447, 0.11541718244552612, 0.015060470439493656, 0.07255656272172928, -0.0417456179857254, 0.020412931218743324, -0.0070816827937960625, 0.03545206040143967, -0.03...
eee657ef-dc5a-5e5b-8e2e-8a0368afffe3
/documentation/private-cloud/api-reference/
API Reference
qdrant.io/v1 > Resource Types > RestoreDestination
documentation
false
API Reference > qdrant.io/v1 > Resource Types > RestoreDestination _Appears in:_ - [QdrantClusterRestoreSpec](#qdrantclusterrestorespec) | Field | Description | Default | Validation | | --- | --- | --- | --- | | `name` _string_ | Name of the destination cluster | | | | `namespace` _string_ | Namespace of the destin...
[ 0.06895621120929718, -0.03393355384469032, 0.006943379063159227, -0.024284837767481804, 0.06837622821331024, -0.05596541613340378, 0.013772867619991302, 0.036330558359622955, 0.028192901983857155, -0.0473477840423584, 0.011888984590768814, 0.02976982109248638, 0.0506572350859642, 0.0016103...
eefe09ad-9099-5878-a9fe-0b2a1441b2e9
/documentation/data-synchronization/with-postgres/
Keeping Postgres and Qdrant in Sync
The Universal Safety Net: Reconciliation
documentation
true
Keeping Postgres and Qdrant in Sync > The Universal Safety Net: Reconciliation Every sync architecture drifts eventually. The reconciliation script is what catches the residue: ```py async def reconcile(fix: bool = False) -> ReconcileResult: pg_ids = set(await get_all_article_ids_from_postgres()) qdrant_ids =...
[ 0.05309521034359932, -0.035797663033008575, -0.020719224587082863, -0.02976333163678646, 0.0399828664958477, -0.02872566506266594, 0.08532290160655975, 0.009230628609657288, 0.06505550444126129, -0.0009499585721641779, 0.05936114490032196, 0.012540169060230255, 0.0559053011238575, 0.000486...
ef7fff7c-3b1f-5e80-8870-de3880a1c412
/documentation/tutorials-basics/search-beginners/
Semantic Search 101
2. Set up a Client Connection
documentation
true
Semantic Search 101 > 2. Set up a Client Connection First, install the Qdrant Client for your preferred programming language: ```python pip install qdrant-client ``` This library allows you to interact with Qdrant from code. Next, create a client connection to your Qdrant cluster using the endpoint and API key. ``...
[ 0.040644869208335876, -0.07275722920894623, -0.031003274023532867, -0.02221527509391308, 0.047230929136276245, -0.029391027987003326, 0.07812250405550003, 0.046211790293455124, 0.04698171094059944, -0.035311292856931686, 0.016078956425189972, 0.020713578909635544, 0.062091439962387085, -0....
ef815a77-0c76-54d5-a4b0-3e6f74fbef52
/documentation/tutorials-basics/search-beginners/
Semantic Search 101
3. Create a Collection
documentation
true
Semantic Search 101 > 3. Create a Collection All data in Qdrant is organized within [collections](/documentation/manage-data/collections/). Since you're storing books, let's create a collection named `my_books`. ```python COLLECTION_NAME="my_books" client.create_collection( collection_name=COLLECTION_NAME, v...
[ 0.03237088397145271, -0.02918780967593193, -0.036403339356184006, -0.04093918949365616, 0.07783251255750656, -0.037643205374479294, 0.06563787907361984, 0.0956243947148323, 0.10477267950773239, -0.026359111070632935, 0.04192069172859192, 0.03499199450016022, 0.06890726089477539, -0.0021450...
ef8614b6-88fd-5d20-ba27-bfced0b54f2d
/documentation/migrate-to-qdrant/from-opensearch/
From OpenSearch
What You Need from OpenSearch
documentation
false
From OpenSearch > What You Need from OpenSearch - **OpenSearch URL** — the HTTP endpoint - **Index name** — the index containing your vectors - **Credentials** — username/password or API key
[ -0.005250793881714344, -0.07971134036779404, -0.04548042267560959, -0.018834156915545464, 0.05089502036571503, -0.014927512966096401, 0.04179694876074791, 0.029412755742669106, 0.0629628598690033, -0.025699589401483536, 0.024158313870429993, 0.018927350640296936, 0.06238750368356705, -0.05...
ef874851-b8c4-5cfc-85ce-ea9964e6348e
/documentation/snapshots/
Snapshots
Storage
documentation
false
Snapshots > Storage Created, uploaded and recovered snapshots are stored as `.snapshot` files. By default, they're stored on the [local file system](#local-file-system). You may also configure to use an [S3 storage](#s3) service for them. As of v1.20.0, you can also use [Google Cloud Storage](#google-cloud-storage) or...
[ -0.003683802904561162, 0.009534497745335102, -0.02577912248671055, -0.0730583444237709, 0.025056524202227592, 0.00351022114045918, 0.04391522705554962, 0.029989372938871384, 0.06129603087902069, -0.040110401809215546, 0.01962120831012726, 0.017246540635824203, 0.051358163356781006, -0.0165...
ef99bc84-46e9-507a-a9ae-6886804b421a
/course/essentials/day-1/embedding-models/
Points, Vectors and Payloads
Points: The Core Entity
course
false
Points, Vectors and Payloads > Points: The Core Entity Points are the central entity that Qdrant operates with. A point is a record consisting of three components: - **Unique ID** (64-bit unsigned integer or UUID) - **Vector** (dense, sparse, or multivector) - **Optional Payload** (metadata) ![Creating an embedding]...
[ 0.03122793883085251, -0.034499119967222214, -0.020682143047451973, -0.07038319110870361, 0.05215024575591087, -0.039002593606710434, 0.06691180914640427, 0.08126959204673767, 0.12378182262182236, -0.0582931712269783, 0.02603914961218834, 0.027606865391135216, -0.010710377246141434, -0.0057...
efa5b9d0-f5ce-5a1f-9262-40415d80512b
/documentation/tutorials-develop/search-as-you-type/
Semantic Search As You Type
Step 2: Create the Collections and Index the Documents
documentation
true
Semantic Search As You Type > Step 2: Create the Collections and Index the Documents You need two collections with the same vector size. `site` holds the content you search, and `prefix_cache` holds embeddings of short prefixes. In `site`, each point is a heading or a paragraph, with a `tag` that says which. The `tex...
[ 0.025842959061264992, -0.0517272874712944, -0.035091400146484375, -0.039145711809396744, 0.048019733279943466, -0.031343184411525726, 0.0362456738948822, 0.04431324079632759, 0.051266249269247055, -0.017840489745140076, 0.013102882541716099, 0.045329779386520386, 0.033966463059186935, -0.0...
efc56b0f-ecb6-5704-a7e4-8c886276da8f
/course/beginners/module-3/filtering/
Filtering: Works with Any Retrieval Method
Filtering a Hybrid Query
course
true
Filtering: Works with Any Retrieval Method > Filtering a Hybrid Query Put the filter inside every `Prefetch`, so each retriever searches only the valid subset: ```python def filtered_hybrid_search(query_text, query_filter, limit=4): return client.query_points( collection_name="products", prefetch=...
[ 0.04277205094695091, -0.013654138892889023, -0.007820900529623032, -0.02163947932422161, 0.06479368358850479, -0.025270376354455948, -0.00040661069215275347, -0.0016868810635060072, 0.05786352977156639, -0.0032929773442447186, 0.0075390697456896305, 0.040424879640340805, 0.05843820050358772,...
efd82318-2b99-55cf-bff5-b6c755885071
/documentation/migrate-to-qdrant/from-solr/
From Apache Solr
Concept Mapping
documentation
false
From Apache Solr > Concept Mapping | Solr | Qdrant | Notes | | :--- | :--- | :--- | | Collection | Collection | One-to-one mapping | | Document | Point | Each document becomes a point | | Dense vector field | Vector | Named vectors are preserved | | Non-vector fields | Payload | Direct mapping | | Document ID (`id` fi...
[ 0.08408427238464355, -0.05706632882356644, -0.04711392521858215, -0.06962171941995621, 0.05673207342624664, -0.027890188619494438, 0.06852243095636368, 0.06799975037574768, 0.06003280356526375, -0.05418388172984123, 0.045641761273145676, 0.015003232285380363, 0.024673955515027046, -0.01810...
efdc38de-3747-51ec-9ba0-69245fb245cc
/articles/turboquant-quantization/
TurboQuant in Qdrant
At a Glance
article
false
TurboQuant in Qdrant > At a Glance Recall, HNSW (`m=16`, `ef_construct=128`), on four representative datasets: [arxiv-instructorxl-768](https://huggingface.co/datasets/Qdrant/arxiv-titles-instructorxl-embeddings), [dbpedia-gemini](https://huggingface.co/datasets/nirantk/dbpedia-entities-google-palm-gemini-embedding-00...
[ 0.05391484126448631, -0.00006817243411205709, -0.06665114313364029, -0.042231474071741104, 0.041901689022779465, -0.015684625133872032, 0.03652489557862282, 0.05189647153019905, 0.051085345447063446, -0.038341447710990906, 0.030128350481390953, 0.016897739842534065, 0.05010613054037094, -0...
f00af6ec-dad4-56f2-8875-e044269f6bd7
/articles/product-quantization/
Product Quantization for Vector Search
Measuring the distance
article
false
Product Quantization for Vector Search > Measuring the distance Vector search relies on the distances between the points. Enabling Product Quantization slightly changes the way it has to be calculated. The query vector is divided into chunks, and then we figure the overall distance as a sum of distances between the ...
[ 0.05508588254451752, -0.052847933024168015, -0.0475858673453331, -0.015095478855073452, 0.09667110443115234, 0.0016558842035010457, 0.051809877157211304, 0.06328004598617554, 0.11318179219961166, -0.028535038232803345, 0.04875906929373741, 0.01399956364184618, 0.041926633566617966, -0.0081...
f042a90d-53bf-5db9-be9c-dbd437a49862
/documentation/search/text-search/
Text Search
Combining Semantic and Lexical Search
documentation
false
Text Search > Combining Semantic and Lexical Search Semantic and lexical search are complementary techniques. When you don't know in advance whether a user is looking for a concept or an exact keyword, you can run both in a single query and merge the results. Qdrant supports this through [hybrid search](/documentation...
[ 0.03707919642329216, -0.03510671108961105, -0.05487801134586334, -0.022551225498318672, 0.04971251264214516, -0.00987657718360424, 0.09599652141332626, -0.018786074593663216, 0.05653184652328491, -0.0055865636095404625, 0.049332499504089355, 0.023586507886648178, 0.0869361013174057, -0.008...
f059693e-5146-597e-972a-b497b8ea424b
/documentation/frameworks/honeyhive/
HoneyHive
Basic Integration Example
documentation
true
HoneyHive > Basic Integration Example The following example demonstrates a complete RAG pipeline with HoneyHive tracing for Qdrant operations. We'll break down each component step by step. ### Initialize Clients and Setup First, set up the necessary clients and configuration for HoneyHive, OpenAI, and Qdrant: ```py...
[ 0.0347459502518177, -0.05900502949953079, -0.04867716506123543, -0.07615222036838531, 0.061976946890354156, -0.05261855572462082, 0.05124357342720032, 0.0522468276321888, 0.048387330025434494, -0.03024473786354065, -0.01724383234977722, 0.028315957635641098, 0.035312019288539886, -0.040998...
f0609743-8423-5c6d-84ed-30c672595cac
/articles/minicoil-en-es/
miniCOIL EN-ES: Sparse Neural Retrieval Across the Language Barrier
Training miniCOIL EN-ES > Four Quadrants (not Qdrant's), Five Points
article
false
miniCOIL EN-ES: Sparse Neural Retrieval Across the Language Barrier > Training miniCOIL EN-ES > Four Quadrants (not Qdrant's), Five Points For both triplets (same- and cross-language), `margin` is not a fixed hyperparameter: it is the gap the teacher itself measures on that triplet, `margin = d_teacher(anchor, negativ...
[ 0.06331966817378998, -0.023056261241436005, -0.042778339236974716, -0.011667164042592049, 0.06053538993000984, -0.002630633767694235, 0.07294166833162308, 0.05613807588815689, 0.058658257126808167, -0.02029391936957836, 0.013313394971191883, -0.013068675994873047, 0.07004322111606598, -0.0...
f06e879b-938d-5e94-81f6-9fd20a66f85a
/documentation/inference/inference-bm25/
BM25
documentation
true
BM25 We recommend running BM25 on a full-text index rather than with the BM25 model. A full-text index with scoring enabled works on text already stored in the payload and doesn't require generating embeddings. [BM25](https://en.wikipedia.org/wiki/Okapi_BM25) (Best Matching 25) is a ranking function for text search. ...
[ 0.03945987671613693, -0.026692915707826614, -0.02166268602013588, -0.03755035623908043, 0.05105921998620033, -0.060560837388038635, 0.08242186903953552, 0.03721921890974045, 0.05747266486287117, -0.008691747672855854, 0.0706460177898407, 0.029365111142396927, 0.02642917074263096, -0.045869...
f087ae1b-e1d2-567e-835d-865c08df530a
/documentation/tutorials-basics/search-beginners-local/
Semantic Search 101
Next Steps
documentation
false
Semantic Search 101 > Next Steps Congratulations, you have just created your very first search engine! Trust us, the rest of Qdrant is not that complicated, either. For your next tutorial, try [building your own hybrid search service](/documentation/tutorials-develop/hybrid-search-fastembed/) or take the free [Qdrant ...
[ 0.031297292560338974, -0.04075849428772926, -0.01840835064649582, -0.012856817804276943, 0.04479071870446205, -0.048961348831653595, 0.05165048688650131, 0.0052493722178041935, 0.05501696839928627, -0.022163797169923782, 0.060026343911886215, 0.02292473055422306, 0.07838645577430725, -0.06...
f092705a-f838-5fce-85e0-5be1e1041a12
/documentation/data-synchronization/with-postgres/
Keeping Postgres and Qdrant in Sync
Choosing Your Tier
documentation
false
Keeping Postgres and Qdrant in Sync > Choosing Your Tier ![Choosing your sync tier](/documentation/data-synchronization/tier_descision_tree.png) These tiers aren't permanent decisions. Start with Tier 1. When you hit its limits — Qdrant outages generating too much drift, write latency becoming noticeable — move to Ti...
[ -0.011422799900174141, -0.033101774752140045, -0.03489912673830986, -0.023317085579037666, 0.04761713743209839, -0.05121789127588272, 0.02642832323908806, 0.03362618759274483, 0.07574838399887085, -0.012697865255177021, 0.030267035588622093, 0.024595022201538086, 0.047330379486083984, -0.0...
f0a10c20-b127-5158-938b-61129c10ac79
/articles/immutable-data-structures/
Immutable Data Structures in Qdrant
Downsides and How to Compensate
article
false
Immutable Data Structures in Qdrant > Downsides and How to Compensate While immutable data structures are great for read-heavy operations, they come with trade-offs: - **Higher update costs:** Immutable structures are less efficient for updates. The amortized time complexity might be the same as mutable structures, b...
[ 0.04102856665849686, -0.02782948687672615, -0.05638032406568527, -0.05298611894249916, 0.045140430331230164, -0.048366330564022064, 0.08956543356180191, 0.009420757181942463, 0.08939939737319946, -0.029078587889671326, 0.00855933502316475, 0.006639581639319658, 0.044428661465644836, -0.024...