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 |
|---|---|---|---|---|---|---|---|
9b5f3c0f-5ea0-5461-a2a0-208693d813d6 | /documentation/tutorials-search-engineering/using-multivector-representations/ | Multivectors and Late Interaction | Why Indexing Every Vector by Default is a Problem | documentation | true | Multivectors and Late Interaction > Why Indexing Every Vector by Default is a Problem
In multivector representations (such as those used by Late Interaction models like ColBERT), a single logical document results in hundreds of token-level vectors. Indexing each of these vectors individually with HNSW in Qdrant can le... | [
0.0343962199985981,
-0.036296676844358444,
-0.04783528670668602,
-0.03303791210055351,
0.049441587179899216,
-0.04238457232713699,
0.059824101626873016,
0.05780982971191406,
0.07876908779144287,
-0.015734774991869926,
0.010412602685391903,
0.033273469656705856,
0.040135014802217484,
-0.010... |
9b668b7d-2983-5dce-8a49-aebb39fd6137 | /documentation/tutorials-operations/create-snapshot/ | Snapshots | Prerequisites | documentation | true | Snapshots > Prerequisites
Let's assume you already have a running Qdrant instance or a cluster. If not, you can follow the [installation guide](/documentation/installation/) to set up a local Qdrant instance or use [Qdrant Cloud](https://cloud.qdrant.io/) to create a cluster in a few clicks.
Once the cluster is runni... | [
-0.011780960485339165,
-0.023997541517019272,
-0.029750589281320572,
-0.05835819989442825,
0.03906138986349106,
-0.048779573291540146,
0.029454097151756287,
0.02343185804784298,
0.06835165619850159,
-0.011441376991569996,
0.032822366803884506,
0.015091531910002232,
0.06690414249897003,
-0.... |
9b854557-150a-51a2-9ede-3c5ccc7cfd70 | /documentation/search/filtering/ | Filtering | Filtering Conditions > Exact Match Conditions > Match Any | documentation | true | Filtering > Filtering Conditions > Exact Match Conditions > Match Any
*Available as of v1.1.0*
In case you want to check if the stored value is one of multiple values, you can use the Match Any condition.
Match Any works as a logical OR for the given values. It can also be described as a `IN` operator.
You can apply... | [
-0.000812335463706404,
-0.043014656752347946,
-0.049549028277397156,
-0.033630501478910446,
0.04257456958293915,
0.008528257720172405,
0.02125483565032482,
0.03087705932557583,
0.05549253523349762,
-0.015591703355312347,
0.010328225791454315,
0.03547415882349014,
0.06310058385133743,
-0.01... |
9bbd5522-58eb-5ac1-8faf-45d027041e4b | /documentation/deploy-intro/ | Deploy Qdrant | Operations | documentation | false | Deploy Qdrant > Operations
- [Configuration](/documentation/ops-configuration/index.md) — Customize Qdrant via config files and environment variables; runtime administration; GPU-accelerated indexing.
- [Monitoring & Telemetry](/documentation/ops-monitoring/index.md) — Monitor Qdrant with Prometheus and Grafana via bu... | [
0.01033801306039095,
-0.044040437787771225,
-0.04725750535726547,
-0.019061850383877754,
0.08869374543428421,
0.008420524187386036,
0.06976620107889175,
0.048143960535526276,
0.051872123032808304,
-0.01924578845500946,
0.048782553523778915,
-0.02676878310739994,
0.05371904373168945,
-0.040... |
9be36f2d-25a3-54cb-81e9-bad42b4d424a | /documentation/snapshots/ | Snapshots | Restore snapshot > Recover during start-up | documentation | true | Snapshots > Restore snapshot > Recover during start-up
This method cannot be used in a multi-node deployment and cannot be used in Qdrant Cloud.
If you have a single-node deployment, you can recover any collection at start-up and it will be immediately available.
Restoring snapshots is done through the Qdrant CLI at ... | [
0.03698454424738884,
-0.06589961796998978,
0.028618142008781433,
-0.031340084969997406,
0.07620050758123398,
-0.04693685472011566,
0.05744237080216408,
-0.011603712104260921,
0.05614577978849411,
-0.043884918093681335,
0.042300641536712646,
0.04537925869226456,
0.06677363067865372,
-0.0084... |
9be6851e-fcc5-5cad-b9fa-7a1fa80cecc5 | /documentation/cloud-quickstart/ | Cloud Quickstart | 1. Create a Cloud Cluster | documentation | false | Cloud Quickstart > 1. Create a Cloud Cluster
1. Register for a [Cloud account](https://cloud.qdrant.io/signup) with your email, Google or Github credentials.
2. Under **Create a Free Cluster**, enter a cluster name and select your preferred cloud provider and region. Click **Create Free Cluster**.
3. Copy your **API k... | [
0.0032579998951405287,
-0.01225217804312706,
-0.04780818894505501,
-0.037655990570783615,
0.07344651967287064,
-0.016713758930563927,
0.0301371943205595,
0.02305961772799492,
0.05775228142738342,
-0.05382251366972923,
0.02221720665693283,
0.006679263897240162,
0.04770006984472275,
-0.03960... |
9bedf209-8d4c-5c7d-9fe5-9a795fe07963 | /documentation/overview/ | Overview | Welcome! | documentation | false | Overview > Welcome!
Whether you’re getting started with Qdrant Open-Source or Cloud, this brief primer will help you with understanding an overview of the platform. **It’s highly recommended you read this overview before starting your development with Qdrant\!** | [
0.009840969927608967,
-0.03311458230018616,
-0.04949484393000603,
-0.07370194792747498,
0.0453357957303524,
-0.030935633927583694,
0.058381397277116776,
0.043658576905727386,
0.05924646183848381,
-0.011542830616235733,
0.02124999463558197,
-0.002602106425911188,
0.04742496460676193,
-0.034... |
9c109574-3a3f-5f08-a057-fde4ff4b19df | /documentation/search/text-search/text-filtering/ | Text Filtering | Filtering on Multiple Exact Strings | documentation | true | Text Filtering > Filtering on Multiple Exact Strings
You can provide multiple filter clauses. For example, to find all books co-authored by Larry Niven and Jerry Pournelle, use the following filter:
```python
from qdrant_client import QdrantClient, models
client = QdrantClient(
url="https://xyz-example.qdrant.io... | [
0.025690848007798195,
-0.0073631457053124905,
-0.05626741424202919,
0.0034698976669460535,
0.07280714809894562,
-0.022928250953555107,
0.038741230964660645,
0.05414535477757454,
0.042643483728170395,
0.0026480224914848804,
0.029674243181943893,
0.030217880383133888,
0.11286838352680206,
-0... |
9c1243e9-1433-5dd7-8b7d-4423fe5b3f67 | /documentation/frameworks/langchain/ | LangChain | Using an existing collection | documentation | true | LangChain > Using an existing collection
To get an instance of `langchain_qdrant.QdrantVectorStore` without loading any new documents or texts, you can use the `QdrantVectorStore.from_existing_collection()` method.
```python
doc_store = QdrantVectorStore.from_existing_collection(
embeddings=embeddings,
collec... | [
0.04064854606986046,
-0.03660367429256439,
-0.022002361714839935,
-0.03562679886817932,
0.06042863801121712,
-0.053364381194114685,
0.02237337827682495,
0.016310149803757668,
0.04237306863069534,
-0.0407429113984108,
-0.009800378233194351,
0.04871901497244835,
0.053701288998126984,
-0.0172... |
9c605372-30e8-575a-81b9-bfe9b161f012 | /documentation/embeddings/qdrant-cloud-inference/ | Qdrant Cloud Inference | Setup | documentation | false | Qdrant Cloud Inference > Setup
If inference isn't enabled on your cluster yet, see [Enabling/Disabling Inference](/documentation/cloud/inference/#enablingdisabling-inference).
Create the client with your cluster URL and API key. In Python, set `cloud_inference=True`; without it, the Python client embeds locally with ... | [
0.04939635470509529,
-0.022993817925453186,
-0.009254354983568192,
-0.038532085716724396,
0.07354392111301422,
-0.018375277519226074,
0.07775342464447021,
0.038220103830099106,
0.03935679420828819,
-0.024409141391515732,
0.017355453222990036,
0.026402361690998077,
0.05580250918865204,
-0.0... |
9c64d046-ba04-5e14-a254-60b544ab57e8 | /course/essentials/day-7/jina/ | Integrating with Jina AI | Jina AI Multimodal Embeddings > Core Architecture | course | false | Integrating with Jina AI > Jina AI Multimodal Embeddings > Core Architecture
Jina AI's embedding system offers several key capabilities:
- **Multimodal Support**: Jina Embeddings v4 supports both text and images on document and query sides
- **Unified Vector Space**: All data types embedded in the same vector space, ... | [
0.01531072799116373,
-0.039727307856082916,
-0.0556984581053257,
-0.0501263253390789,
0.07455868273973465,
-0.019260987639427185,
0.06954158842563629,
0.01893291063606739,
0.05127844586968422,
-0.014643904753029346,
0.05840151384472847,
-0.02701999619603157,
0.05971173569560051,
-0.0265516... |
9c7b4079-472a-50bb-b31b-089255b8a5f3 | /articles/dataset-quality/ | Detecting Dataset Errors with Similarity Search | Furniture web-marketplace > Diversity search | article | false | Detecting Dataset Errors with Similarity Search > Furniture web-marketplace > Diversity search
Since pre-trained models have only general knowledge about the data, they can still leave some misplaced items undetected.
You might find yourself in a situation when the model focuses on non-important features, selects a lo... | [
0.08810654282569885,
-0.04675858095288277,
-0.05098014697432518,
-0.00046269624726846814,
0.0896952897310257,
-0.0019344527972862124,
0.05005969852209091,
0.04322084039449692,
0.06991613656282425,
-0.007170419674366713,
0.02811277098953724,
0.02083130180835724,
0.04086463153362274,
-0.0080... |
9cac1e20-f3d7-500c-9607-8e44bd404048 | /documentation/faq/qdrant-fundamentals/ | Qdrant Fundamentals | Collections > Can I switch to a different embedding model without recreating my collection? | documentation | false | Qdrant Fundamentals > Collections > Can I switch to a different embedding model without recreating my collection?
Yes, if you're using named vectors:
1. [Add a new named vector](/documentation/manage-data/collections/#update-vector-schema) for the new model
2. Re-embed points in the background
3. Remove the old named... | [
0.07352104038000107,
-0.03376439958810806,
-0.012029587291181087,
-0.045563023537397385,
0.04906713590025902,
-0.04884594306349754,
0.04515475034713745,
0.02252056822180748,
0.03746704384684563,
-0.02875581569969654,
0.030523797497153282,
0.049927737563848495,
0.067303866147995,
-0.0405274... |
9cb9a324-c761-51b4-88a0-ed50753c46bf | /documentation/overview/vector-search/ | Understanding Vector Search in Qdrant | A Brief History of Search | documentation | false | Understanding Vector Search in Qdrant > A Brief History of Search
Human memory is unreliable. Thus, as long as we have been trying to collect ‘knowledge’ in written form, we had to figure out how to search for relevant content without rereading the same books repeatedly. That’s why some brilliant minds introduced the ... | [
0.028063228353857994,
-0.05529202148318291,
-0.05128017067909241,
-0.008516869507730007,
0.046677082777023315,
-0.0482979379594326,
0.043271105736494064,
0.024092787876725197,
0.10254116356372833,
-0.02216573804616928,
0.028827328234910965,
0.033711135387420654,
0.03327653184533119,
-0.031... |
9cc06774-47d8-5426-b752-fb10a448c27b | /documentation/data-management/redpanda/ | Redpanda Connect | documentation | true | Redpanda Connect

[Redpanda Connect](https://www.redpanda.com/connect) is a declarative data-agnostic streaming service designed for efficient, stateless processing steps. It offers transaction-based resiliency with back pressure, ensuring a... | [
0.025492504239082336,
-0.05019601061940193,
-0.06290871649980545,
-0.050892941653728485,
0.01780625246465206,
-0.0670008435845375,
0.09429436922073364,
0.03557073697447777,
0.03730688989162445,
-0.04134051501750946,
0.04068993777036667,
-0.003784003434702754,
0.022224081680178642,
-0.01446... | |
9cf28b5f-d664-5bca-a4c1-58023c4b9c08 | /documentation/search-precision/reranking-semantic-search/ | Reranking for Better Search | documentation | false | Reranking for Better Search
In Retrieval-Augmented Generation (RAG) systems, irrelevant or missing information can throw off your model's ability to produce accurate, meaningful outputs. One of the best ways to ensure you're feeding your language model the most relevant, context-rich documents is through reranking. It... | [
0.012640225701034069,
-0.035926856100559235,
-0.07925011962652206,
-0.04694647714495659,
0.08592261373996735,
-0.04369267821311951,
0.05744316428899765,
0.04336676746606827,
0.06722977757453918,
-0.02246021293103695,
0.015238415449857712,
0.0070459540002048016,
0.07130805402994156,
-0.0327... | |
9d146749-cb14-5f57-b030-7b59c11643d1 | /documentation/embeddings/mixedbread/ | Mixedbread | Mixedbread API (Legacy) | documentation | true | Mixedbread > Mixedbread API (Legacy)
Mixedbread's Python and TypeScript SDKs still include the direct embeddings API:
- Python: `pip install mixedbread`
- TypeScript: `npm install @mixedbread/sdk`
The client reads your API key from the `MXBAI_API_KEY` environment variable. As in Mixedbread's own example, this code a... | [
0.04653635993599892,
-0.021079231053590775,
0.006240117363631725,
-0.027800805866718292,
0.06779097020626068,
-0.06197609379887581,
0.049173902720212936,
0.05980581417679787,
0.025709981098771095,
-0.0418141707777977,
0.037299931049346924,
0.01999506726861,
0.042925138026475906,
-0.0262804... |
9d4f50b3-8ec1-5b79-8625-9a3f41263a90 | /documentation/tutorials-develop/neural-search/ | Build a Semantic Search API | documentation | false | Build a Semantic Search API
| Time: 30 min | Level: Beginner | Output: [GitHub](https://github.com/qdrant/examples/tree/master/neural-search) | [](https://githubtocolab.com/qdrant/examples/blob/master/neural-search/neural_search.ipynb) |
| ---... | [
0.017855660989880562,
-0.07636051625013351,
-0.04919769614934921,
-0.019528863951563835,
0.08210691809654236,
-0.006717291660606861,
0.05055880546569824,
0.0590304471552372,
0.06114525720477104,
-0.035434722900390625,
0.05485687777400017,
0.01859845407307148,
0.09513351321220398,
-0.050163... | |
9d5282a7-12b4-520c-8920-af6a0744caf3 | /articles/io_uring/ | Faster Disk I/O for Vector Search Using io_uring | And what about Qdrant? > Qdrant 1.19 and later | article | true | Faster Disk I/O for Vector Search Using io_uring > And what about Qdrant? > Qdrant 1.19 and later
Set `storage.performance.io_uring` to `auto`. Qdrant then uses io\_uring for
every structure with a `cold` [memory placement](/documentation/ops-configuration/memory-tiers/#async-io),
such as original vectors and payload,... | [
0.05374365299940109,
-0.0463944710791111,
-0.029425887390971184,
-0.006895987316966057,
0.08532195538282394,
-0.006537439301609993,
0.04612952843308449,
0.00476816575974226,
0.07750990986824036,
-0.012240355834364891,
-0.00042262437636964023,
-0.0071691726334393024,
0.05355008691549301,
0.... |
9d836fca-2c26-595e-9564-05a6ac6f40bb | /documentation/production-checklist/ | Production Checklist | 1. Distributed Deployment & Sharding | documentation | false | Production Checklist > 1. Distributed Deployment & Sharding
Architect for scale from day one. Retrofitting these patterns onto an existing deployment is costly.
- **Ensure you have enough shards to scale.**
Qdrant [scales horizontally](/documentation/scaling/distributed_deployment/) through [sharding](/documentation... | [
0.028349462896585464,
-0.039538636803627014,
-0.0648939236998558,
-0.05083034560084343,
0.06798725575208664,
-0.045118290930986404,
0.03593013063073158,
0.03064200095832348,
0.0664062350988388,
-0.03663874790072441,
0.0451202392578125,
-0.0001074176179827191,
0.031469423323869705,
-0.04174... |
9d9348d1-ff5b-5d4a-bda5-5ee0f4aa0aa4 | /documentation/cloud/inference/ | Inference | Embedding Models | documentation | false | Inference > Embedding Models
Clusters on Qdrant Managed Cloud can access embedding models that are hosted on Qdrant Cloud and models that are hosted externally by other providers.
### Qdrant-Hosted Models
The following models are available:
#### Dense Models
The `qdrant/clip-vit-b-32-text` and `qdrant/clip-vit-b-3... | [
0.05434001237154007,
-0.03473725542426109,
-0.03218308091163635,
-0.042068347334861755,
0.05325143411755562,
-0.03920351713895798,
0.059161338955163956,
0.08137306571006775,
0.047959327697753906,
-0.04666011407971382,
0.049951571971178055,
-0.014586634933948517,
0.045418113470077515,
-0.03... |
9d9abca6-d6fd-509b-9c73-995e6d49dc68 | /course/essentials/day-7/camel/ | Integrating with Camel AI | CAMEL Auto-Retrieval Architecture > Vector Retrieval Demonstration | course | false | Integrating with Camel AI > CAMEL Auto-Retrieval Architecture > Vector Retrieval Demonstration
When you query "What is Qdrant?" with a Qdrant website link, the system:
- Retrieves relevant content with similarity scores
- Includes metadata for context understanding
- Provides comprehensive answers based on the retriev... | [
0.035753991454839706,
-0.06404011696577072,
-0.06607798486948013,
-0.048318926244974136,
0.03788207471370697,
-0.013236215338110924,
0.07988052070140839,
0.022949116304516792,
0.06598864495754242,
-0.036368418484926224,
0.045925285667181015,
-0.013296669349074364,
0.058361757546663284,
-0.... |
9da49832-384e-598c-887b-06f11f101b26 | /documentation/cloud/backups/ | Backup Clusters | Backup Considerations | documentation | false | Backup Clusters > Backup Considerations
Backups are incremental for AWS and GCP clusters. For example, if you have two backups, backup number 2
contains only the data that changed since backup number 1. This reduces the
total cost of your backups.
For Azure clusters, backups are based on total disk usage. The cost is... | [
-0.04188006371259689,
-0.01564314402639866,
-0.0350482203066349,
-0.05939101427793503,
0.014794430695474148,
0.019717194139957428,
0.04365691542625427,
0.034813057631254196,
0.07145489752292633,
-0.0000728262311895378,
0.05309754237532616,
0.021334515884518623,
0.014135402627289295,
-0.030... |
9ddbba01-a27b-50fb-81b9-857fe12163e3 | /documentation/fastembed/fastembed-rerankers/ | Reranking with FastEmbed | Conclusion | documentation | false | Reranking with FastEmbed > Conclusion
Rerankers refine search results by reordering retrieved candidates through deeper semantic analysis. For efficiency, they should be applied **only to a small subset of retrieved results**.
Balance speed and accuracy in search by leveraging the power of rerankers! | [
0.006426294334232807,
-0.05991150811314583,
-0.055825408548116684,
-0.06557503342628479,
0.09405995160341263,
-0.035643745213747025,
0.05681854113936424,
0.010675939731299877,
0.07391616702079773,
-0.011882305145263672,
0.007829244248569012,
0.04522040858864784,
0.08091920614242554,
-0.009... |
9de6d588-6b45-5947-9553-b198cd0c9de7 | /documentation/edge/edge-quickstart/ | Quickstart | Configure the Edge Shard | documentation | true | Quickstart > Configure the Edge Shard
An Edge Shard is configured with a definition of the dense and sparse vectors that can be stored in the Edge Shard, similar to how you would configure a Qdrant collection.
Set up a configuration by creating an instance of `EdgeConfig`. For example:
```python
from qdrant_edge imp... | [
0.04418126493692398,
-0.03822915628552437,
-0.06548673659563065,
-0.05323747918009758,
0.025518249720335007,
-0.05454472079873085,
0.10132592916488647,
0.07355136424303055,
0.08988970518112183,
-0.03406462445855141,
0.03992466256022453,
0.008273172192275524,
0.03730262070894241,
-0.0171708... |
9de898ba-d520-5d56-93a8-47f1cdfcfdb5 | /documentation/tutorials-search-engineering/turbo4-multivector-search/ | Compressed Multivector Search | Wrapping Up | documentation | false | Compressed Multivector Search > Wrapping Up
`turbo4` is a general-purpose datatype, not a special case for single dense vectors: this collection stores it on a per-token ColBERT multivector, side by side with a BM25 sparse vector, and queries both through one Query API call. Set `datatype=models.Datatype.TURBO4` on an... | [
0.04066888242959976,
-0.021111050620675087,
-0.06483016908168793,
-0.046284373849630356,
0.052273012697696686,
-0.0051045604050159454,
0.06696298718452454,
0.04431302845478058,
0.06192509084939957,
-0.04128700867295265,
0.041366174817085266,
0.018267789855599403,
0.02798180840909481,
-0.01... |
9e07754e-40d1-572f-a1e3-3d48542aebd2 | /documentation/tutorials-search-engineering/branch-aware-search/ | Branch-Aware Search | Step 4: Branch Off the Root | documentation | true | Branch-Aware Search > Step 4: Branch Off the Root
A fork records where it started: the parent and the parent's latest `seq`. It writes no points; its view begins as whatever the parent has at that `seq`.
```text
root ●──●──● seq 0–2
└── A ● A forks at root seq 2; A seq 0: add roadmap.md
`... | [
0.02427864633500576,
-0.06590350717306137,
-0.0269040297716856,
-0.02023993991315365,
0.06478258967399597,
-0.017693711444735527,
0.008381170220673084,
0.04445177689194679,
0.06997624039649963,
-0.015491447411477566,
0.040435172617435455,
0.06156349554657936,
0.07057732343673706,
-0.016274... |
9e0b1682-4c91-589e-8a97-06959d99fcb3 | /documentation/search-patterns/vector-search-filtering/ | Filtering in Vector Search: When to Use What | Let the Query Planner Choose | documentation | false | Filtering in Vector Search: When to Use What > Let the Query Planner Choose
For each filtered query, the [query planner](/documentation/search/search/#query-planning) estimates the **filter cardinality**, the number of points that satisfy the filter, from the payload indexes. It plans each segment of the collection se... | [
0.06240048259496689,
-0.03782470151782036,
-0.044086799025535583,
-0.025418387725949287,
0.04546334594488144,
-0.005847779102623463,
0.04969660937786102,
0.028874091804027557,
0.09166465699672699,
-0.022863104939460754,
0.05276479944586754,
0.02482043020427227,
0.04371408000588417,
-0.0295... |
9e17964c-9b68-5e31-9963-5cca21abfca4 | /course/multi-vector-search/module-3/evaluating-pipelines/ | Evaluating Search Pipelines | Ground Truth: Defining "Correct" Retrieval | course | true | Evaluating Search Pipelines > Ground Truth: Defining "Correct" Retrieval
To measure retrieval quality, you need **relevance judgments** (called **qrels**) - data that defines which documents are relevant for which queries.
### What Are Qrels?
Each qrel is a triplet: a query, a document, and a relevance score indicat... | [
0.029896244406700134,
-0.02990630827844143,
-0.041713085025548935,
-0.053706295788288116,
0.048924390226602554,
-0.007402416318655014,
0.06592442095279694,
0.033747874200344086,
0.06170728802680969,
-0.042359836399555206,
0.02388906478881836,
-0.018992049619555473,
0.042967867106199265,
-0... |
9e43d6ea-fdd7-5b99-bce0-3e8c3ebe5244 | /course/essentials/day-7/camel/ | Integrating with Camel AI | CAMEL Auto-Retrieval Architecture > Auto-Retrieval Process | course | false | Integrating with Camel AI > CAMEL Auto-Retrieval Architecture > Auto-Retrieval Process
The CAMEL Auto-Retrieval workflow follows these key steps:
1. **Environment Setup**: Install necessary libraries and configure your chosen large model API (supports various platforms including "gamma" and others).
2. **Vector Data... | [
0.02046431414783001,
-0.056410323828458786,
-0.05550072342157364,
-0.030942382290959358,
0.08376690745353699,
-0.027163328602910042,
0.04072478786110878,
0.0023736231960356236,
0.04204709082841873,
-0.016260674223303795,
0.054209716618061066,
-0.021437902003526688,
0.03756038472056389,
-0.... |
9e4c821c-3a3f-577a-8498-64efe0d48ff1 | /documentation/ops-optimization/read-write-contention/ | Read-Write Contention | Step 8: Scale Horizontally with Replicas | documentation | false | Read-Write Contention > Step 8: Scale Horizontally with Replicas
The previous steps all operate within a fixed hardware budget: they reallocate CPU between the optimizer and queries. If you've exhausted those options and latency is still too high, the answer is more capacity.
[Adding nodes to your cluster](/documenta... | [
-0.015256138518452644,
-0.038303811103105545,
-0.08740899711847305,
-0.02497805468738079,
0.07027296721935272,
-0.03205736353993416,
0.0557769276201725,
0.06349104642868042,
0.06361937522888184,
-0.02754637412726879,
0.02532704547047615,
0.0028565877582877874,
0.023312073200941086,
-0.0091... |
9e541ecb-0688-5378-9864-2a077ec2f691 | /documentation/tutorials-operations/gpu-accelerated-hnsw-indexing/ | GPU-Accelerated HNSW Indexing | Creating a Collection and Uploading Data | documentation | true | GPU-Accelerated HNSW Indexing > Creating a Collection and Uploading Data
### Install Dependencies
Install `qdrant-client` to interact with the new cluster, and `huggingface-hub` and `polars` to download and process the dataset.
```bash
pip install -q qdrant-client huggingface-hub polars
```
### Initialize the Clien... | [
0.03620050847530365,
-0.04206793010234833,
-0.05643787980079651,
-0.02904178947210312,
0.06920024007558823,
-0.056167375296354294,
0.05026024580001831,
0.08947978168725967,
0.05560010299086571,
-0.029169009998440742,
0.0449507050216198,
0.028291236609220505,
0.04482341930270195,
-0.0254671... |
9e5deed1-054a-556d-a1ac-37874994681c | /documentation/tutorials-search-engineering/slicing-filter/ | Deterministic Collection Slicing | Related Reading | documentation | false | Deterministic Collection Slicing > Related Reading
- [Filtering](/documentation/search/filtering/) for the full set of filter clauses `slice` combines with.
- [Scroll Points](/documentation/manage-data/points/#scroll-points) for paging through a collection.
- [Random Sampling](/documentation/search/search/#random-samp... | [
0.04667113721370697,
0.014647205360233784,
-0.07086573541164398,
-0.06453246623277664,
0.061807580292224884,
-0.02809995412826538,
0.10081604868173599,
0.04173096641898155,
0.08980362862348557,
-0.02945893630385399,
0.03301570564508438,
0.04057757556438446,
0.059736598283052444,
0.02210214... |
9e5e94b1-ecb1-5dae-81d0-5329f1acc58e | /documentation/tutorials-operations/data-streaming-kafka-qdrant/ | Kafka Streaming into Qdrant | Integration and Pipeline Importance | documentation | false | Kafka Streaming into Qdrant > Integration and Pipeline Importance
The integration of these components forms a powerful and efficient data streaming pipeline. The **Qdrant Sink Connector** ensures that the data flowing through Kafka is continuously ingested into Qdrant without any manual intervention. This real-time in... | [
0.0036891677882522345,
-0.04138978570699692,
-0.05967700481414795,
-0.02228284627199173,
0.04384297877550125,
-0.02670595981180668,
0.08619808405637741,
0.029856257140636444,
0.06132392957806587,
-0.0012647532857954502,
-0.0029400077182799578,
0.004915047902613878,
0.012505033053457737,
-0... |
9e62dae8-d8ad-5586-9569-bf45aa9ba40e | /course/ | Qdrant Academy | Upcoming Courses > Advanced Level | course | false | Qdrant Academy > Upcoming Courses > Advanced Level
Advanced courses are recommended for those that have completed the Beginner and Intermediate Level Modules first, and provide guidance for the ultimate level of mastery with Qdrant. In these modules, you'll learn of common anti-patterns and pitfalls, nuances and best ... | [
0.015923285856842995,
-0.017400197684764862,
-0.04516597464680672,
-0.0331122912466526,
0.0366649255156517,
-0.03961871564388275,
0.02698829211294651,
0.05059858411550522,
0.08317071199417114,
-0.049852706491947174,
0.050674330443143845,
0.006094839423894882,
0.0557766929268837,
-0.0539037... |
9e8c6e04-3f86-5314-831c-098f65ad1896 | /documentation/edge/edge-api/configuration/ | Configuration | EdgeConfig | documentation | true | Configuration > EdgeConfig
```python
EdgeConfig(
vectors: Optional[Union[EdgeVectorParams, Dict[str, EdgeVectorParams]]] = None,
sparse_vectors: Optional[Dict[str, EdgeSparseVectorParams]] = None,
on_disk_payload: Optional[bool] = None,
hnsw_config: Optional[HnswIndexConfig] = None,
quantization_co... | [
0.029299169778823853,
-0.06076494976878166,
-0.06157917156815529,
-0.020623356103897095,
0.05485536903142929,
-0.03462604433298111,
0.07809922844171524,
0.06808781623840332,
0.052979230880737305,
-0.02598654106259346,
0.04982805997133255,
0.002500720787793398,
0.012375254184007645,
-0.0407... |
9e929d56-a9d4-5914-bc1c-a181c1043702 | /documentation/search-tuning/before-tuning-a-qdrant-collection/ | What to Check Before Tuning a Qdrant Collection | Make Sure Your Labels Can Detect a Gain | documentation | true | What to Check Before Tuning a Qdrant Collection > Make Sure Your Labels Can Detect a Gain
[Retrieval relevance](/documentation/search-evaluation/retrieval-relevance/) covers building a labeled set. Its size decides whether any retrieval tuning is visible to you at all.
A labeled set is large enough when it can distin... | [
0.02635941654443741,
-0.02811185084283352,
-0.060938749462366104,
-0.005958838853985071,
0.041219450533390045,
-0.05120677128434181,
0.06824592500925064,
0.04543818533420563,
0.061356786638498306,
-0.04003191739320755,
0.018381832167506218,
0.020742297172546387,
0.050264809280633926,
-0.00... |
9e9b5ead-f0f8-5091-9a17-54be4a779572 | /documentation/observability/datadog/ | Datadog | documentation | true | Datadog

[Datadog](https://www.datadoghq.com/) is a cloud-based monitoring and analytics platform that offers real-time monitoring of servers, databases, and numerous other tools and services. It provides visibility into the performance of applic... | [
-0.006317922845482826,
-0.009396849200129509,
-0.053863875567913055,
-0.009353982284665108,
0.041321318596601486,
-0.037024129182100296,
0.09182289242744446,
0.03442949801683426,
0.051164038479328156,
-0.046970997005701065,
0.008836023509502411,
0.01456552091985941,
0.028106961399316788,
-... | |
9ead3f7b-7bc1-5c28-8114-02a64fdf594e | /documentation/security/ | Security | Hardening | documentation | false | Security > Hardening
* Block Qdrant's external network access. This can help mitigate [server side request forgery attacks](https://owasp.org/www-community/attacks/Server_Side_Request_Forgery), like via the [snapshot recovery API](https://api.qdrant.tech/api-reference/snapshots/recover-from-snapshot). Single-node Qdra... | [
0.02424273081123829,
-0.04125427454710007,
-0.04938257113099098,
-0.0031324271112680435,
0.041658155620098114,
-0.046854790300130844,
0.07514123618602753,
0.0611237995326519,
0.07420666515827179,
-0.01926598697900772,
0.021268265321850777,
0.03295508399605751,
0.02964399941265583,
-0.04457... |
9eada2dc-a618-5c51-908c-ec9b33d6b1e2 | /documentation/search-evaluation/retrieval-relevance/ | Measuring Retrieval Relevance | Generating Queries > 3. LLM-Based Synthetic Generation | documentation | true | Measuring Retrieval Relevance > Generating Queries > 3. LLM-Based Synthetic Generation
An LLM can generate plausible queries for documents sampled from your corpus. This scales cheaply to thousands of pairs, but synthetic queries are typically easier to retrieve than real user queries, which inflates offline scores. F... | [
-0.0009981909533962607,
-0.03644150123000145,
-0.05946648120880127,
-0.05579047277569771,
0.05283696576952934,
-0.02786220982670784,
0.051147930324077606,
0.020307769998908043,
0.07805314660072327,
-0.05348348617553711,
0.018640363588929176,
-0.01162062305957079,
0.05318168178200722,
-0.01... |
9eb30ab9-38dc-552e-825e-6ae8781a64dd | /documentation/embeddings/mixedbread/ | Mixedbread | Embed Queries | documentation | true | Mixedbread > Embed Queries
For retrieval, `mxbai-embed-large-v1` expects each query to start with the prompt `Represent this sentence for searching relevant passages: `. Documents take no prompt. `encode_query` adds the prompt from the model's configuration, so pass the query text as it is:
```python
query_vector = m... | [
0.030430978164076805,
-0.05661085620522499,
-0.015492419712245464,
-0.08865978568792343,
0.07184959203004837,
-0.03701488673686981,
0.028221050277352333,
0.048904675990343094,
0.024109112098813057,
-0.017754700034856796,
0.02019559033215046,
0.01082873996347189,
0.052424073219299316,
-0.04... |
9ebc971c-76ee-52ae-9c12-ecf3d4b0eee8 | /course/multi-vector-search/module-3/muvera/ | MUVERA | course | false | MUVERA
Module 3
MUVERA (Multi-Vector Retrieval with Approximation) solves a fundamental problem: MaxSim's asymmetry makes traditional indexing methods like HNSW ineffective. MUVERA enables fast approximate search for multi-vector representations.
Understanding MUVERA is key to scaling multi-vector search to millions... | [
0.07679947465658188,
-0.028393259271979332,
-0.050412263721227646,
-0.054104555398225784,
0.07190585136413574,
-0.01801651157438755,
0.05630239471793175,
0.016035284847021103,
0.04226433485746384,
0.032691337168216705,
0.011960727162659168,
0.05655158683657646,
0.055493392050266266,
-0.017... | |
9ed4dd3a-1ad6-5a9e-a3ed-655bca05e2e6 | /documentation/embeddings/mixedbread/ | Mixedbread | Qdrant Cloud Inference | documentation | true | Mixedbread > Qdrant Cloud Inference
On Qdrant Cloud, Qdrant hosts `mixedbread-ai/mxbai-embed-large-v1` as a paid model. Instead of a vector, send a `Document` with the model name `mixedbread-ai/mxbai-embed-large-v1` in place of the snippet's `<the-model-to-use>` placeholder. The collection's vector size must be 1024.
... | [
0.04856935515999794,
-0.028044234961271286,
-0.014054335653781891,
-0.06707698106765747,
0.08096025139093399,
-0.037003155797719955,
0.09109701961278915,
0.048082444816827774,
0.05230723321437836,
-0.04221830144524574,
0.012599196285009384,
0.004948139656335115,
0.0457763671875,
-0.0505561... |
9ee75ce6-772c-5f5f-8e0d-8ab5e5c2972a | /articles/hyperbolic-embeddings-qdrant/ | Why Hyperbolic Embeddings Matter for Hierarchical Data | Serving Them with Qdrant > What About a Production 512-Dimensional Model? | article | false | Why Hyperbolic Embeddings Matter for Hierarchical Data > Serving Them with Qdrant > What About a Production 512-Dimensional Model?
Everything above uses a five dimensional embedding we trained on one taxonomy. A fair question is what changes if you reach for a pretrained 512 dimensional hyperbolic model instead, such ... | [
0.044895611703395844,
0.01026733685284853,
-0.03660248592495918,
-0.0397535040974617,
0.06701890379190445,
-0.020133689045906067,
0.05579407885670662,
0.06914951652288437,
0.09407773613929749,
-0.03265202045440674,
0.0110298041254282,
0.006141564808785915,
0.0755816102027893,
-0.0496476814... |
9f099b9c-8414-5489-8ef9-0df094864791 | /documentation/ops-configuration/memory-tiers/ | Memory Tiers | Example | documentation | true | Memory Tiers > Example
This example configures a collection so the vectors are cached in RAM, the HNSW vector index is cold, the quantized vectors are pinned, and the payload and the ID tracker are cached:
```python
from qdrant_client import QdrantClient, models
client.create_collection(
collection_name="{collec... | [
0.031035900115966797,
-0.04718845337629318,
-0.04574211686849594,
-0.04721638932824135,
0.04679480940103531,
-0.01127068791538477,
0.050090234726667404,
0.06204219534993172,
0.10970313102006912,
-0.00794316828250885,
-0.01615402288734913,
0.052784502506256104,
0.01996789500117302,
-0.00810... |
9f0ed849-90ea-57fb-a1ea-a4b06ab4240c | /documentation/manage-data/indexing/ | Indexing | Payload Index > Parameterized Index > Principal Index | documentation | true | Indexing > Payload Index > Parameterized Index > Principal Index
*Available as of v1.11.0*
Similar to the tenant index, the principal index is used to optimize storage for faster search, assuming that the search request is primarily filtered by the principal field.
A good example of a use case for the principal inde... | [
-0.005609605461359024,
-0.03763139620423317,
-0.04061633348464966,
-0.01496936846524477,
0.057313356548547745,
-0.02635788917541504,
0.0184206273406744,
0.008475338108837605,
0.11195118725299835,
-0.04098653793334961,
0.04836806282401085,
0.014568706974387169,
0.0546283945441246,
-0.016876... |
9f159a4a-7c8f-59ce-b979-dbe931f36611 | /articles/sparse-vectors/ | Understanding SPLADE and Sparse Vectors | Why SPLADE works: term expansion | article | false | Understanding SPLADE and Sparse Vectors > Why SPLADE works: term expansion
Consider a query "solar energy advantages". SPLADE might expand this to include terms like "renewable," "sustainable," and "photovoltaic," which are contextually relevant but not explicitly mentioned. This process is called term expansion, and ... | [
0.05123066157102585,
-0.01515941321849823,
-0.05291474610567093,
-0.05199401080608368,
0.060822293162345886,
-0.034044161438941956,
0.05225107818841934,
0.054793208837509155,
0.07336270809173584,
-0.04166605696082115,
0.01060431357473135,
-0.009362737648189068,
0.004630153998732567,
0.0159... |
9f260482-4be8-5084-a0af-480533d94223 | /documentation/ops-configuration/running-with-gpu/ | Running with GPU | NVIDIA GPUs | documentation | true | Running with GPU > NVIDIA GPUs
### Prerequisites
To use Docker with NVIDIA GPU support, ensure the following are installed on your host:
- Latest NVIDIA drivers
- [nvidia-container-toolkit](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/install-guide.html)
Most AI or CUDA images on Amazon/G... | [
0.037724435329437256,
-0.07207633554935455,
-0.06399676203727722,
-0.0020231897942721844,
0.04859799146652222,
-0.02615528181195259,
0.07510498911142349,
0.04570094123482704,
0.06072832643985748,
-0.02927962876856327,
0.02177520841360092,
-0.015364042483270168,
0.02992256172001362,
-0.0400... |
9f450cd7-ea01-5678-8545-345a6e93b200 | /documentation/cloud/configure-cluster/ | Configure Clusters | Collection Defaults | documentation | false | Configure Clusters > Collection Defaults
You can set default values for the configuration of new collections in your cluster. These defaults will be used when creating a new collection, unless you override them in the collection creation request.
You can configure the default *Replication Factor*, the default *Write ... | [
0.03610290214419365,
-0.05246372148394585,
-0.03407792001962662,
-0.04250577837228775,
0.0511319525539875,
-0.0435950793325901,
0.0165919978171587,
0.05938917025923729,
0.06004133075475693,
-0.028265507891774178,
0.011458691209554672,
0.005765294190496206,
0.04657311737537384,
-0.027591267... |
9f53a9d9-485a-5acd-9521-1ba2239602c7 | /documentation/migrate-to-qdrant/from-qdrant/ | From Qdrant | Run the Migration | documentation | true | From Qdrant > Run the Migration
### Between Two Instances
```bash
docker run --net=host --rm -it registry.cloud.qdrant.io/library/qdrant-migration qdrant \
--source.url 'http://source-instance:6334' \
--source.api-key 'source-api-key' \
--source.collection 'your-collection' \
--target.url 'https://you... | [
0.02328251115977764,
-0.017027463763952255,
-0.05445031076669693,
0.014731709845364094,
0.09065050631761551,
-0.029519999399781227,
0.07096713036298752,
0.0497049018740654,
0.04974588751792908,
-0.033139996230602264,
0.033266764134168625,
-0.0003398092230781913,
0.011013581417500973,
-0.02... |
9f762055-776c-5dd0-8f9d-9e79d5bb0c83 | /documentation/snapshots/ | Snapshots | Restore full storage snapshot | documentation | true | Snapshots > Restore full storage snapshot
Restoring snapshots can only be done through the Qdrant CLI at startup time.
For example:
```bash
./qdrant --storage-snapshot /snapshots/full-snapshot-2022-07-18-11-20-51.snapshot
``` | [
0.020976558327674866,
-0.04032556712627411,
0.019285716116428375,
-0.052287507802248,
0.03753872215747833,
-0.012404086999595165,
0.03156710043549538,
-0.024896033108234406,
0.09742310643196106,
-0.025269711390137672,
0.04070386663079262,
0.04171635955572128,
0.06832241266965866,
-0.009228... |
9fc36573-969a-5dd2-ba2d-11b3a14af597 | /course/multi-vector-search/ | Multi-Vector Search Course | What you'll learn | course | false | Multi-Vector Search Course > What you'll learn
- Late interaction paradigm and MaxSim distance metric
- ColBERT for text and ColPali for visual documents
- Multi-stage retrieval with prefetch and reranking
- Quantization and pooling techniques for memory optimization
- MUVERA indexing for billion-scale search
- Evalua... | [
0.05035174638032913,
-0.02670332044363022,
-0.01840854622423649,
-0.013529437594115734,
0.08630610257387161,
-0.02888781577348709,
0.05504981800913811,
0.027003690600395203,
0.06438345462083817,
-0.015071545727550983,
0.004575581289827824,
0.02544855885207653,
0.02965357154607773,
-0.02842... |
9fcb4a24-cfa5-53a3-89f3-ac24a2ac34a3 | /articles/dedicated-vector-search/ | Why Vector Search Needs a Dedicated Database | Objections to a Dedicated Service > "My current database is the most seamless option" | article | false | Why Vector Search Needs a Dedicated Database > Objections to a Dedicated Service > "My current database is the most seamless option"
It is, at first. Over time, a dedicated service gives you flexibility that an all-in-one database does not:
- **Independent upgrades.** You do not have to update the primary database ea... | [
-0.01260872557759285,
-0.05687617510557175,
-0.054540351033210754,
-0.013525976799428463,
0.01531943678855896,
-0.007467402145266533,
0.022854482755064964,
0.014297306537628174,
0.05283216014504433,
-0.01672975905239582,
0.030939429998397827,
0.04186256602406502,
0.07738994061946869,
-0.03... |
9fd74759-09f6-5d87-967e-7a97ab32c818 | /documentation/cloud/create-cluster/ | Create a Cluster | Free Clusters | documentation | false | Create a Cluster > Free Clusters
Qdrant Cloud offers two types of clusters: **Free** and **Standard**.
Free tier clusters are perfect for prototyping and testing. You don't need a credit card to join.
A free tier cluster only includes 1 single node with the following resources:
| Resource | Value |
|------------|... | [
0.011344842612743378,
-0.020688971504569054,
-0.047602131962776184,
-0.03868579864501953,
0.048469435423612595,
-0.043097130954265594,
0.03233439847826958,
0.0403367280960083,
0.09047431498765945,
-0.04679114744067192,
0.01127617433667183,
0.014102227054536343,
0.02521296590566635,
-0.0362... |
9feec475-2aa2-50f9-a440-56d4bafafe35 | /documentation/production-operations/bulk-data-import/ | Bulk Uploading Data to Qdrant | Throughput Choices > Use Multiple Shards for Larger Uploads | documentation | true | Bulk Uploading Data to Qdrant > Throughput Choices > Use Multiple Shards for Larger Uploads
_Dense & sparse vectors_
Each shard has its own write-ahead log and update worker, so a collection with multiple shards can process writes in parallel. Sharding matters most when you spread a collection across two or more node... | [
-0.0022998396307229996,
-0.02325238659977913,
-0.054337821900844574,
-0.01981491409242153,
0.07907908409833908,
-0.04371176287531853,
0.06690860539674759,
0.0377175435423851,
0.07236387580633163,
-0.04181661456823349,
0.020789187401533127,
0.03344593197107315,
0.011252613738179207,
-0.0409... |
9ff38d80-d49b-52e1-b6f1-af82040bd790 | /articles/what-is-vector-quantization/ | What is Vector Quantization? | Distributing Resources Between Disk & Memory > Speeding Up Rescoring with io_uring | article | true | What is Vector Quantization? > Distributing Resources Between Disk & Memory > Speeding Up Rescoring with io_uring
When dealing with large collections of quantized vectors, frequent disk reads are required to retrieve both original and compressed data for rescoring operations. While `mmap` helps with efficient I/O by r... | [
0.053296130150556564,
-0.06068155914545059,
-0.030914505943655968,
-0.024765072390437126,
0.08544004708528519,
-0.029832474887371063,
0.06718241423368454,
0.03526534140110016,
0.08052650839090347,
-0.009593494236469269,
-0.0010714050149545074,
0.013344439677894115,
0.03298123925924301,
-0.... |
9ffcb8b9-4302-5a18-aa6c-b0576fdd1c58 | /articles/gridstore-key-value-storage/ | Introducing Gridstore: Qdrant's Custom Key-Value Store | Gridstore in Production: Maintaining Data Integrity | article | false | Introducing Gridstore: Qdrant's Custom Key-Value Store > Gridstore in Production: Maintaining Data Integrity
Gridstore’s architecture introduces multiple interdependent structures that must remain in sync to ensure data integrity:
- **The Data Layer** holds the data and associates each key with its location in storage... | [
0.01513638999313116,
0.005371924955397844,
-0.05815094709396362,
-0.022845685482025146,
0.04926258698105812,
-0.0040717292577028275,
0.03926386684179306,
0.022394176572561264,
0.05175482854247093,
-0.015101365745067596,
0.021291952580213547,
-0.0053035542368888855,
0.08353480696678162,
-0.... |
a016a7c8-11bf-5e62-b92d-eaa5daca3318 | /course/multi-vector-search/module-2/colpali-family/ | ColPali Family Overview | The Impact of Bidirectional Attention > Bidirectional Attention: Optimized for Embeddings | course | true | ColPali Family Overview > The Impact of Bidirectional Attention > Bidirectional Attention: Optimized for Embeddings
**Bidirectional attention**, as used in encoder models like BERT, allows each token to attend to the entire input sequence - both past and future tokens. This creates richer, more contextually informed r... | [
0.036972012370824814,
-0.012397242709994316,
-0.031222233548760414,
-0.009703227318823338,
0.05877116695046425,
0.0037131912540644407,
0.06934218853712082,
0.05825861170887947,
0.046934083104133606,
-0.020760169252753258,
0.042108967900276184,
-0.029958130791783333,
0.0407584086060524,
-0.... |
a03df98e-2e4c-5c84-aa8a-b75c2d589a7f | /documentation/frameworks/rocketride/ | RocketRide | How retrieval works | documentation | false | RocketRide > How retrieval works
The example uses two Qdrant nodes pointed at the same collection:
- The ingestion node receives document chunks after the embedding step and writes them to Qdrant.
- The query node receives an embedded question, retrieves matching chunks, and sends both the question and retrieved cont... | [
0.02817947044968605,
-0.05480889603495598,
-0.04802794009447098,
-0.034062519669532776,
0.06938924640417099,
-0.03021690435707569,
0.0376463383436203,
0.03389585018157959,
0.04908024147152901,
-0.018458722159266472,
0.03230952471494675,
-0.0049546449445188046,
0.042490508407354355,
-0.0308... |
a08506bc-3794-5b4d-8ade-2d20bd25f116 | /documentation/tutorials-search-engineering/food-discovery-recommendations/ | Multimodal Food Search Demo | Different search modes > Location-based search | documentation | true | Multimodal Food Search Demo > Different search modes > Location-based search
Last but not least, location plays an important role in the food discovery process. You are definitely looking for something you can find nearby, not on the other
side of the globe. Therefore, your current location can be toggled as a filteri... | [
0.025541329756379128,
-0.050257015973329544,
-0.040286146104335785,
-0.005408321972936392,
0.07187903672456741,
-0.0018732944736257195,
0.07748467475175858,
0.03805247321724892,
0.07114449888467789,
-0.01723713055253029,
0.0029061990790069103,
0.040098607540130615,
0.05518238991498947,
-0.... |
a08ff75f-0a83-5269-b892-2e028f4c347f | /articles/search-feedback-loop/ | Relevance Feedback in Information Retrieval | Query Refinement > Query As Vector > Gradient Descent-Based Methods | article | false | Relevance Feedback in Information Retrieval > Query Refinement > Query As Vector > Gradient Descent-Based Methods
The efficient way of doing so on-the-fly remained an open question until the introduction of a **gradient-descent-based Rocchio’s method generalization**: [`Test-Time Optimization of Query Representations ... | [
0.02427327260375023,
-0.0218814704567194,
-0.04224224016070366,
-0.045406799763441086,
0.048022348433732986,
-0.019299600273370743,
0.06974317133426666,
0.04735309258103371,
0.06500087678432465,
-0.010558183304965496,
0.01591898314654827,
0.01843547448515892,
0.03831672668457031,
-0.062907... |
a0928c68-0ce8-590f-8efd-b5446123a8d7 | /documentation/tutorials-build-essentials/video-anomaly-edge-part-1/ | Video Anomaly Detection Part 1: Architecture, Twelve Labs, and NVIDIA VSS | Application Demo | documentation | false | Video Anomaly Detection Part 1: Architecture, Twelve Labs, and NVIDIA VSS > Application Demo
Before we begin coding, check out the project repository and live demo to get familiarized with what we'll be building.
**GitHub**: [qdrant/video-anomaly-edge](https://github.com/qdrant/video-anomaly-edge)
**Live Demo**: [qd... | [
0.05031871050596237,
-0.04970374330878258,
-0.05556880310177803,
-0.04637378454208374,
0.06163199618458748,
-0.0010281543945893645,
0.044944554567337036,
0.03835127502679825,
0.07089291512966156,
-0.04554812237620354,
0.0655616894364357,
0.03157223388552666,
0.06381753087043762,
-0.0370652... |
a0a0eead-e812-58d8-87d4-45bba2b4ce9c | /documentation/fastembed/fastembed-minicoil/ | Working with miniCOIL | When to use miniCOIL | documentation | false | Working with miniCOIL > When to use miniCOIL
When exact keyword matches in the retrieved results are a requirement, and all matches should be ranked based on the contextual meaning of keywords.
If results should be similar by meaning but are expressed differently, with no overlapping keywords, you should use dense em... | [
0.009731068275868893,
-0.05584690719842911,
-0.04252426326274872,
-0.06234930083155632,
0.08850608766078949,
-0.07986399531364441,
0.07584735006093979,
0.060784563422203064,
0.07438203692436218,
-0.015415608882904053,
0.02728687785565853,
0.014191504567861557,
0.08534126728773117,
-0.00476... |
a0a18dfa-bafd-5ae0-b75c-8f2048a917b8 | /documentation/manage-data/vectors/ | Vectors | Datatypes > TurboQuant Datatypes > Turbo16 | documentation | true | Vectors > Datatypes > TurboQuant Datatypes > Turbo16
*Available as of v1.20.0*
The `turbo16` datatype stores each vector as a 16-bit representation per dimension instead of a 32-bit floating-point number. This shrinks vector storage to about half of its original size.
To use `turbo16`, specify the `turbo16` datatype... | [
0.03554490581154823,
-0.03504762426018715,
-0.04043648764491081,
-0.03580407425761223,
0.03673574700951576,
0.00406255666166544,
0.02668886072933674,
0.0248507559299469,
0.07693269848823547,
-0.022340834140777588,
0.04959971457719803,
0.0060261148028075695,
0.055923014879226685,
-0.0082716... |
a0e229d4-4755-5946-a7b1-895359dfe1e6 | /articles/immutable-data-structures/ | Immutable Data Structures in Qdrant | How Does Immutability Help? | article | false | Immutable Data Structures in Qdrant > How Does Immutability Help?
A large part of the immutable advantage comes from the fact that we know the exact data we need to put into the structure even before we start building it.
The simplest example is a sorted array: we would know exactly how many elements we have to put in... | [
0.06875298917293549,
-0.029547879472374916,
-0.07714622467756271,
-0.027859177440404892,
0.080994613468647,
-0.05104893073439598,
0.06358811259269714,
0.042950473725795746,
0.07878386229276657,
-0.011108548380434513,
-0.013621583580970764,
0.01918855682015419,
0.026684535667300224,
-0.0339... |
a0f6bef4-5dd6-5016-9be5-86bd6d2e3135 | /course/essentials/day-3/hybrid-search/ | Hybrid Search and the Universal Query API | The Challenge: Different Users, Different Search Needs | course | false | Hybrid Search and the Universal Query API > The Challenge: Different Users, Different Search Needs
The reality is that your users exist across a spectrum: from precise keyword searchers to vague natural language describers, and forcing a single search approach means disappointing part of your audience. Rather than com... | [
-0.002448262646794319,
-0.03619851544499397,
-0.05536486208438873,
-0.027600832283496857,
0.07241133600473404,
-0.04954458400607109,
0.07158058881759644,
0.04723425582051277,
0.06937673687934875,
-0.027703192085027695,
-0.0029589880723506212,
0.03488698974251747,
0.0685686394572258,
-0.026... |
a107da92-88a6-5695-a0ec-95d87e13de88 | /documentation/tutorials-search-engineering/multi-representation-search/ | Multi-Representation Search | Setup | documentation | true | Multi-Representation Search > Setup
Install the Python packages used throughout the tutorial:
```bash
pip install qdrant-client datasets
```
This tutorial uses Qdrant Cloud Inference to generate embeddings server-side. The free tier covers this tutorial's footprint. Core BM25 runs on any Qdrant instance, but dense C... | [
0.04869096353650093,
-0.021025413647294044,
-0.007224485743790865,
-0.0784514918923378,
0.06229132041335106,
-0.048140060156583786,
0.10269414633512497,
0.06373830884695053,
0.06246953830122948,
-0.02569410391151905,
0.0346369631588459,
0.011568423360586166,
0.033384960144758224,
-0.040131... |
a1088504-0351-540b-9596-ded2ed7daa85 | /documentation/tutorials-search-engineering/using-multivector-representations/ | Multivectors and Late Interaction | What are Multivector Representations? | documentation | false | Multivectors and Late Interaction > What are Multivector Representations?
In most vector engines, each document is represented by a single vector - an approach that works well for short texts but often struggles with longer documents. Single vector representations perform pooling of the token-level embeddings, which o... | [
0.026383375748991966,
-0.04316338896751404,
-0.038645509630441666,
-0.050382886081933975,
0.06034557521343231,
-0.033370811492204666,
0.0871901661157608,
0.05886557698249817,
0.05763592571020126,
-0.01019324455410242,
0.0242341049015522,
0.007980789989233017,
0.03018975630402565,
-0.025325... |
a14d99cf-6793-57c6-b0b6-9e7eb4f54a86 | /course/essentials/day-7/llamaindex/ | Integrating with LlamaIndex | LlamaIndex Agent Development Framework > Core Architecture | course | false | Integrating with LlamaIndex > LlamaIndex Agent Development Framework > Core Architecture
LlamaIndex's agent framework offers several key components:
- **Multi-Deployment Support**: Local Qdrant, Qdrant Cloud, and Llama Cloud integration options
- **Agent Workflows**: Custom workflow development with events and steps
... | [
0.010670360177755356,
-0.028323626145720482,
-0.06851085275411606,
-0.04468916729092598,
0.08705545961856842,
-0.0330197848379612,
0.08064541965723038,
0.016279716044664383,
0.06696458905935287,
-0.04815263673663139,
0.053762245923280716,
0.027945755049586296,
-0.014428278431296349,
-0.000... |
a153bb0f-15b9-589e-b0e1-2538a6c703f8 | /documentation/tutorials-operations/blue-green-deployment/ | Blue-Green Cluster Deployment | Step 2: Load Data into the New Cluster > What Happens to Blue Cluster Writes During This Step | documentation | false | Blue-Green Cluster Deployment > Step 2: Load Data into the New Cluster > What Happens to Blue Cluster Writes During This Step
Loading the green cluster does not affect blue's ability to serve reads and writes. The blue cluster keeps running at full capacity throughout Step 2, because every option above (ingestion scri... | [
0.023376064375042915,
-0.045381784439086914,
-0.058465663343667984,
-0.02775951288640499,
0.0880630761384964,
-0.005909980274736881,
0.02314789406955242,
0.009224223904311657,
0.02741938643157482,
-0.020867031067609787,
0.0034155177418142557,
0.008392664603888988,
0.020626632496714592,
-0.... |
a1667b45-cd25-5074-b25f-538c52ada64f | /course/beginners/module-1/what-is-search/ | What Is Search? | course | false | What Is Search?
Module 1
Search is the act of finding the right information out of everything you have, given a question. You type "car repair" into a box, and something has to decide which of your thousands of documents, products, or messages actually answers that.
Every search system, no matter how it's built inte... | [
0.0317385233938694,
-0.07216738909482956,
-0.08072517812252045,
-0.03190293163061142,
0.08028805255889893,
0.0034860465675592422,
0.024003999307751656,
0.021030472591519356,
0.08309688419103622,
-0.0362522266805172,
0.03607920929789543,
0.032249610871076584,
0.07163325697183609,
-0.0214997... | |
a178943d-4ad6-57c5-861e-1b6b6ffc5ee4 | /documentation/overview/ | Overview | Scaling Considerations > Scaling | documentation | false | Overview > Scaling Considerations > Scaling
Vertical scaling has natural limits \- eventually, you'll hit the maximum capacity of available hardware, and single-node deployments lack redundancy. Optimize scaling with **sharding**, **replication**, and **segment configuration** options.
#### Sharding
Qdrant uses shar... | [
0.01965676248073578,
-0.028357500210404396,
-0.07136475294828415,
-0.03799105063080788,
0.07428398728370667,
-0.02805984765291214,
0.0679757297039032,
0.04173767566680908,
0.08379819244146347,
-0.024861803278326988,
0.019452642649412155,
-0.011943826451897621,
0.023699287325143814,
-0.0345... |
a195a9c5-d82d-5bff-8d08-2ffa1d1e73b5 | /course/essentials/day-2/what-is-hnsw/ | HNSW Indexing Fundamentals | Configuring HNSW > Graph Connectivity: m | course | true | HNSW Indexing Fundamentals > Configuring HNSW > Graph Connectivity: m
The `m` parameter controls the maximum number of connections per node in the graph.
- Higher `m`: results in a denser graph where each vector is connected to more neighbors, which improves search accuracy because the graph has more pathways to trav... | [
0.010358276776969433,
-0.06044480577111244,
-0.06971395760774612,
0.012934541329741478,
0.03857402130961418,
-0.04101449251174927,
0.050263773649930954,
0.06742008030414581,
0.0512920618057251,
0.004914030898362398,
0.018445435911417007,
0.007134770043194294,
0.04433916509151459,
-0.001949... |
a1b497c6-326f-5ee3-a553-9620c356aa86 | /documentation/cloud-rbac/permission-reference/ | Permission Reference | Hybrid Cloud | documentation | false | Permission Reference > Hybrid Cloud
Permissions for Hybrid Cloud environments.
| Permission | Description |
|------------|------------|
| `read:hybrid_cloud_environments` | View Hybrid Cloud environment details. |
| `write:hybrid_cloud_environments` | Modify Hybrid Cloud environment settings. |
| `delete:hybrid_cloud... | [
0.015031447634100914,
0.02869371883571148,
-0.04932485148310661,
-0.11191076785326004,
0.06309853494167328,
-0.0012689458671957254,
0.03070809505879879,
0.06739656627178192,
0.05370153859257698,
-0.046329315751791,
0.010508009232580662,
-0.012967342510819435,
0.08631749451160431,
-0.025837... |
a1cf8cf5-f6b1-50ab-9a6b-28828a28ec16 | /articles/what-is-a-vector-database/ | What is a Vector Database? | Common Applications of Vector Search | article | false | What is a Vector Database? > Common Applications of Vector Search
Vector search enables a wide range of applications by allowing systems to retrieve information based on meaning rather than exact matches. Some of the most common use cases include:
| **Use Case** | **How It Works** ... | [
-0.007912423461675644,
-0.06411196291446686,
-0.06128767132759094,
-0.056720294058322906,
0.08176669478416443,
-0.02245953306555748,
0.03817296773195267,
0.023363206535577774,
0.07118638604879379,
-0.04612526297569275,
0.021392853930592537,
0.011232574470341206,
0.0596223920583725,
-0.0327... |
a1d34741-ce57-5bee-a2c3-e98c6293d94f | /documentation/ops-configuration/memory-tiers/ | Memory Tiers | Optimizing for Disk-Based Retrieval > Inline Storage | documentation | false | Memory Tiers > Optimizing for Disk-Based Retrieval > Inline Storage
*Available as of v1.16.0*
Avoid putting the HNSW vector index in the `cold` tier. If you must store it on disk and use quantization, consider enabling [inline storage](/documentation/ops-optimization/optimize/#inline-storage-in-hnsw-index). This redu... | [
0.02951839193701744,
-0.04801187291741371,
-0.0037991476710885763,
-0.03511010482907295,
0.03599750995635986,
-0.025044765323400497,
0.05086157098412514,
0.052055723965168,
0.07896038889884949,
0.012438041158020496,
0.0007408287492580712,
0.03237040713429451,
0.04634029418230057,
-0.024938... |
a229d739-79f1-5d10-9d77-2690c059dc52 | /course/beginners/module-0/qdrant-cloud/ | Qdrant Setup | Other Ways to Connect | course | true | Qdrant Setup > Other Ways to Connect
You can also reach your cluster directly over the web, without Python. This is handy for a quick test:
```bash
# Using the api-key header
curl -X GET https://xyz-example.eu-central.aws.cloud.qdrant.io:6333/collections \
--header 'api-key: <your-api-key>'
# Using the Authorizati... | [
0.03998350352048874,
-0.04538111761212349,
-0.044188518077135086,
-0.006632880307734013,
0.0575164370238781,
-0.046926066279411316,
0.03645651042461395,
0.017713813111186028,
0.046304214745759964,
-0.030138861387968063,
0.022596988826990128,
0.03517971187829971,
0.05114532634615898,
-0.009... |
a22da742-ca9a-5d06-9032-19ff0009331e | /articles/what-is-rag-in-ai/ | Understanding Retrieval-Augmented Generation (RAG) | Where is RAG being used? | article | false | Understanding Retrieval-Augmented Generation (RAG) > Where is RAG being used?
Because of their more knowledgeable and contextual responses, we can find RAG models being applied in many areas today, especially those that need factual accuracy and knowledge depth.
### Real-World Applications
**Question answering:** Th... | [
0.001222478924319148,
-0.012072903104126453,
-0.0348048098385334,
-0.06707587838172913,
0.06968547403812408,
-0.023115728050470352,
0.03443806618452072,
0.025339500978589058,
0.06038303300738335,
-0.06882969290018082,
0.03348343074321747,
0.019670994952321053,
0.08866588771343231,
-0.01516... |
a272088e-f64c-5c24-96fc-b394ff9d2fc9 | /documentation/tutorials-search-engineering/index-dynamic-payloads/ | Indexing Payloads of Random Shape | Keep Value Types for Ranges | documentation | true | Indexing Payloads of Random Shape > Keep Value Types for Ranges
The string array handles exact match, but strings have no numeric ordering, so it rules out range queries like `price` below 500. Split attributes by value type into parallel arrays, one per type: strings, numbers, and booleans, each indexed with the sche... | [
0.02819804847240448,
-0.012811850756406784,
-0.04446309804916382,
-0.019473591819405556,
0.06446325778961182,
-0.03969060629606247,
0.03521151840686798,
0.011063620448112488,
0.07477155327796936,
-0.022619565948843956,
0.026186665520071983,
0.01953754760324955,
0.06860996782779694,
0.01002... |
a27a44c3-dae2-5e91-80b6-1dc8eba1b428 | /documentation/tutorials-build-essentials/agentic-rag-camelai-discord/ | Discord RAG Bot | Understand the RAG Flow | documentation | false | Discord RAG Bot > Understand the RAG Flow
Select **Index documents** or **Answer a question** to trace the two paths. Hover over a step, focus it with the keyboard, or select it to keep its explanation visible.
 | `pinecone` | Yes |
| [Weaviate](/documentation/migrate-to-qdrant/from-weaviate/) | `weaviate` | No (must pre-create) |
| [Milvus](/documentati... | [
0.04496770724654198,
-0.05091447755694389,
-0.028384098783135414,
-0.0488952100276947,
0.05122130364179611,
-0.02245407924056053,
0.054117027670145035,
0.027046432718634605,
0.02726989984512329,
-0.04661234840750694,
0.04407954961061478,
0.016381394118070602,
0.05019749701023102,
-0.026345... |
a2f7a093-978f-5539-8261-8b2fb9ba53e2 | /documentation/embeddings/cohere/ | Cohere | Reduce Vector Size | documentation | true | Cohere > Reduce Vector Size
Embed v5 and v4 models accept an `output_dimension` parameter (`outputDimension` in TypeScript) that returns a shorter vector. Set the collection's `size` to the same value, and pass the same `output_dimension` when you embed queries:
```python
response = co.embed(
model="embed-v5.0-pr... | [
0.06326638907194138,
-0.005498351529240608,
-0.006252800580114126,
-0.07160881161689758,
0.07574326545000076,
-0.011797657236456871,
0.056639861315488815,
0.06552919000387192,
0.07072824239730835,
-0.017589358612895012,
0.05138577148318291,
0.03096524253487587,
0.048498064279556274,
-0.017... |
a2fa9ce2-fe3f-5be5-9b31-3311b098df1b | /documentation/frameworks/vanna-ai/ | Vanna.AI | documentation | true | Vanna.AI
[Vanna](https://vanna.ai/) is a Python package that uses retrieval augmentation to help you generate accurate SQL queries for your database using LLMs.
Vanna works in two easy steps - train a RAG "model" on your data, and then ask questions which will return SQL queries that can be set up to automatically ru... | [
0.03572623059153557,
-0.04261287301778793,
-0.02122802659869194,
-0.04755719006061554,
0.060393333435058594,
-0.05248452350497246,
0.06059315800666809,
0.05255650356411934,
0.019625771790742874,
-0.02060130052268505,
0.048386987298727036,
0.01385222002863884,
0.039615217596292496,
-0.03627... | |
a310a643-2db4-542b-9c7c-481de580af72 | /documentation/examples/rag-chatbot-vultr-dspy-ollama/ | Private RAG Information Extraction Engine | Deploying Qdrant Hybrid Cloud on Vultr > Installing the necessary packages | documentation | true | Private RAG Information Extraction Engine > Deploying Qdrant Hybrid Cloud on Vultr > Installing the necessary packages
We are going to need a couple of Python packages to run our application. They might be installed together with the
`dspy-ai` package and `qdrant` extra:
```shell
pip install dspy-ai dspy-qdrant
``` | [
0.027382545173168182,
-0.02661452256143093,
-0.02655448205769062,
-0.05660494044423103,
0.09076500684022903,
-0.041440218687057495,
0.05257531628012657,
0.048071764409542084,
0.0706682950258255,
-0.03162391856312752,
0.04687710851430893,
0.03562159091234207,
0.09704276919364929,
-0.0332097... |
a3205d70-87e3-5ba3-b248-26c6cca25433 | /documentation/production-operations/managed-cloud-prometheus/ | Managed Cloud Prometheus Monitoring | Step 2: Configure Prometheus to Scrape Qdrant Metrics | documentation | true | Managed Cloud Prometheus Monitoring > Step 2: Configure Prometheus to Scrape Qdrant Metrics
To monitor Qdrant, you need to configure Prometheus to scrape metrics from the Qdrant database. You can do this by creating a `ScrapeConfig` resource in your Kubernetes cluster. The API key to authenticate at your Qdrant databa... | [
0.040898289531469345,
-0.001991917844861746,
-0.035470325499773026,
0.0030942184384912252,
0.04767070338129997,
-0.03300566226243973,
0.074141725897789,
0.06749988347291946,
0.050346147269010544,
-0.023121504113078117,
0.05701278895139694,
0.004900866188108921,
0.12005472183227539,
-0.0201... |
a32f9bc2-3af1-579b-a347-fd9364e14c93 | /course/multi-vector-search/module-3/quantization-techniques/ | Vector Quantization Techniques | Quantization: Compressing Without Losing Quality > Scalar Quantization: The Reliable Default | course | false | Vector Quantization Techniques > Quantization: Compressing Without Losing Quality > Scalar Quantization: The Reliable Default
**Scalar quantization** converts float32 values to 8-bit integers (uint8), reducing memory by **4x**.
**How it works:**
1. Find the min and max values across all vector components
2. Map the f... | [
0.03404741361737251,
-0.047547511756420135,
-0.05109941586852074,
-0.026953430846333504,
0.04955393821001053,
0.007187388837337494,
0.008353912271559238,
0.032320812344551086,
0.11098238825798035,
-0.02052135393023491,
0.045778822153806686,
-0.005574447568506002,
0.042278651148080826,
-0.0... |
a336d066-3c1f-5282-853a-90a4b98fd801 | /course/beginners/module-2/chunking-strategies/ | Chunking Strategies | course | false | Chunking Strategies
Module 2
Embedding models have a maximum token limit. `all-MiniLM-L6-v2` from Module 1 takes 256 tokens, larger models take 8,000 or more, and anything past the limit is dropped without an error. Check your model's card for its limit.
Fitting isn't the only reason to split. A chunk is the unit th... | [
0.029529109597206116,
-0.03746306523680687,
-0.07027776539325714,
-0.028698498383164406,
0.05808964744210243,
-0.035634301602840424,
0.04916118457913399,
0.017083682119846344,
0.07014908641576767,
-0.04193666949868202,
0.03066958673298359,
-0.020547637715935707,
0.05586154758930206,
-0.052... | |
a33bbea5-1959-5325-9013-59168f0299e0 | /articles/dedicated-vector-search/ | Why Vector Search Needs a Dedicated Database | Vector Search > Pick Any Two | article | false | Why Vector Search Needs a Dedicated Database > Vector Search > Pick Any Two
Distributed systems are perfect for scalability — horizontal scaling in these systems allows you to add more machines as needed. In the world of distributed systems, one well-known principle — the **CAP theorem** — illustrates that you cannot ... | [
-0.010118439793586731,
-0.06059540808200836,
-0.0633993148803711,
-0.017838675528764725,
0.05046723410487175,
-0.025163525715470314,
0.05707982927560806,
0.03126440942287445,
0.07404962182044983,
0.009014565497636795,
0.0024475816171616316,
0.06113239377737045,
0.0632137879729271,
-0.04623... |
a33f2549-558b-5a38-847b-1876a4356248 | /documentation/tutorials-build-essentials/semantic-cache/ | Semantic Caching for RAG | Set Up Cache Lookup > Optional: Resolve Hard Hits with Jev | documentation | false | Semantic Caching for RAG > Set Up Cache Lookup > Optional: Resolve Hard Hits with Jev
Here's what Jev said on the two hard cases:
| New question | Cached question | Cosine | Jev | Decision |
| --- | --- | --- | --- | --- |
| "Is there a fee to send something back if I want my money back?" | "Do I pay for return shipp... | [
0.02518996223807335,
-0.03793531283736229,
-0.060502614825963974,
-0.05934977903962135,
0.03590106964111328,
-0.0067754327319562435,
0.02456781640648842,
0.019190505146980286,
0.06332813203334808,
-0.026161812245845795,
0.04208998754620552,
0.019532471895217896,
0.050430506467819214,
-0.02... |
a33f36ef-1d4c-5bdc-b39a-657de29c5fd7 | /documentation/tutorials-build-essentials/video-anomaly-edge-part-2/ | Video Anomaly Detection Part 2: Edge-to-Cloud Pipeline | Why Qdrant Edge | documentation | true | Video Anomaly Detection Part 2: Edge-to-Cloud Pipeline > Why Qdrant Edge
The cloud tier runs a full Qdrant cluster. The edge tier cannot. NVIDIA Jetson devices have limited memory, no guaranteed internet connectivity, and need sub-millisecond kNN lookups without the overhead of a client-server architecture.
[Qdrant E... | [
0.02913954295217991,
-0.034860946238040924,
-0.06381065398454666,
-0.02564075030386448,
0.060441724956035614,
-0.012612023390829563,
0.07912928611040115,
0.010744074359536171,
0.07532599568367004,
-0.048455242067575455,
0.02468874864280224,
0.029532460495829582,
0.018168672919273376,
-0.01... |
a34bc098-7d07-594e-966c-f39fb1f92984 | /articles/discovery-search/ | Discovery Search in Qdrant | article | false | Discovery Search in Qdrant
When Christopher Columbus and his crew sailed to cross the Atlantic Ocean, they were not looking for the Americas. They were looking for a new route to India because they believed the Earth was small enough to reach India by sailing west. They didn't know anything about a new continent, but ... | [
0.048638228327035904,
-0.02832166664302349,
-0.03887137025594711,
-0.02154514752328396,
0.045663218945264816,
-0.01841500587761402,
0.033944271504879,
0.00019869774405378848,
0.05731712654232979,
-0.036108437925577164,
0.01827288791537285,
0.006972198840230703,
0.08246764540672302,
-0.0055... | |
a34c594e-aa4d-5c6e-83ae-6c4e876b59dd | /articles/vector-similarity-beyond-search/ | Vector Similarity: Going Beyond Full-Text Search | Recommendations > Distance-Based Recommendations | article | true | Vector Similarity: Going Beyond Full-Text Search > Recommendations > Distance-Based Recommendations
Another approach uses the distances from each candidate to the positive and negative examples to create exclusion areas.
In this technique, we perform searches near the positive examples while excluding the points that ... | [
0.06276612728834152,
-0.062238387763500214,
-0.034500863403081894,
-0.0019029347458854318,
0.06461741030216217,
-0.021885447204113007,
0.08159691840410233,
0.029670359566807747,
0.07970740646123886,
-0.03889862075448036,
0.04052778705954552,
0.035265758633613586,
0.04224586486816406,
0.007... |
a386af27-b4e0-51bf-8570-d38ef4200886 | /documentation/snapshots/ | Snapshots | Restore snapshot > Recover from an uploaded file | documentation | true | Snapshots > Restore snapshot > Recover from an uploaded file
The snapshot file can also be uploaded as a file and restored using the [recover from uploaded snapshot](https://api.qdrant.tech/master/api-reference/snapshots/recover-from-uploaded-snapshot). This endpoint accepts the raw snapshot data in the request body. ... | [
0.04881880059838295,
-0.03391917422413826,
-0.01695849932730198,
-0.07277698814868927,
0.04827987030148506,
0.007810887414962053,
0.06207536906003952,
0.012729021720588207,
0.05987812951207161,
-0.030252350494265556,
0.02766343019902706,
0.05882316827774048,
0.09691141545772552,
-0.0068937... |
a398e37e-e2f9-58ff-be31-34a5f82661ce | /documentation/examples/hybrid-search-llamaindex-jinaai/ | Chat With Product PDF Manuals Using Hybrid Search | Check Embeddings, Inference, and Qdrant | documentation | true | Chat With Product PDF Manuals Using Hybrid Search > Check Embeddings, Inference, and Qdrant
Use `llama_index.llms.openrouter`, provided by the
`llama-index-llms-openrouter` package. The wrapper sends chat-completion
requests to OpenRouter. No `CustomLLM` subclass is needed.
These small live requests check the embeddi... | [
0.038047585636377335,
-0.06303226947784424,
-0.019581379368901253,
-0.03914603590965271,
0.057466354221105576,
-0.028431396931409836,
0.11111774295568466,
0.07544779032468796,
0.031847286969423294,
0.012585301883518696,
0.010546534322202206,
-0.0005708473036065698,
-0.002041606465354562,
-... |
a3a4eff2-36fd-5036-a50e-8a80be3f89f4 | /course/essentials/day-2/filterable-hnsw/ | Combining Vector Search and Filtering | Performance Optimization Tips | course | false | Combining Vector Search and Filtering > Performance Optimization Tips
1. **Index early:** Create payload indexes before HNSW build.
1. **Index the right fields:** Create payload indexes for all filtered fields. Prefer high-selectivity fields if memory is limited.
3. **Test filter combinations:** Complex multi-field fi... | [
0.026930615305900574,
-0.04975530505180359,
-0.05180481821298599,
-0.0234747976064682,
0.07382098585367203,
-0.011880177073180676,
0.054956935346126556,
0.017130877822637558,
0.06894551217556,
-0.022583121433854103,
0.042345885187387466,
0.04013378173112869,
0.038564395159482956,
-0.052023... |
a3a997e1-8d6a-506c-ab84-86ae2f34c145 | /documentation/migration-guidance/search-quality/ | Search Quality | Tier 3: Gold-Standard Evaluation (When You Have Labeled Data) > Tier 3 Pass/Fail Criteria | documentation | false | Search Quality > Tier 3: Gold-Standard Evaluation (When You Have Labeled Data) > Tier 3 Pass/Fail Criteria
Tier 3 targets are domain-specific. Here are starting points:
| Metric | Good | Acceptable | Investigate |
| ----- | ----- | ----- | ----- |
| NDCG@10 | ≥0.70 | 0.50 to 0.70 | <0.50 |
| MRR | ≥0.60 | 0.40 to 0.6... | [
0.012802401557564735,
-0.04233323782682419,
-0.04493379592895508,
-0.01079044584184885,
0.05062069743871689,
-0.024513034150004387,
0.08736269176006317,
0.04288671538233757,
0.08843301236629486,
-0.03240545094013214,
-0.001295194262638688,
0.010697922669351101,
0.040916334837675095,
-0.018... |
a3a9f616-2723-5b55-b700-c0f473522f4f | /course/multi-vector-search/module-2/visual-interpretability/ | Visual Interpretability of ColPali | Aggregated MaxSim Visualization | course | true | Visual Interpretability of ColPali > Aggregated MaxSim Visualization
Sometimes you want to see which patches contribute most to the overall score, regardless of which query token they matched. This **aggregated MaxSim view** shows the document-level relevance:
```python
def compute_maxsim_contribution(query_vectors, ... | [
0.03974798321723938,
-0.0331018902361393,
-0.04638011008501053,
-0.05953472480177879,
0.03811023756861687,
-0.026797130703926086,
0.03772994875907898,
0.029161982238292694,
0.017986003309488297,
-0.030806751921772957,
0.05705326050519943,
-0.002841152250766754,
0.09436493366956711,
-0.0330... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.