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 |
|---|---|---|---|---|---|---|---|
241c1b6d-85de-58c3-a04f-d698e81a34f6 | /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
The `parallel` parameter enables data-parallelism instead of built-in ONNX parallelism.
Additionally, you can specify ids for each document, if you want to use them later to update or delete documents.
If you don't specify ids, they will be generated automatically.
Y... | [
0.027812117710709572,
-0.0499303862452507,
-0.04763256385922432,
-0.03293653577566147,
0.07080380618572235,
-0.004124700091779232,
0.06723740696907043,
0.026572521775960922,
0.06995728611946106,
-0.027944110333919525,
0.023443488404154778,
0.02757197618484497,
0.05624343827366829,
-0.02159... |
242efb03-e892-5987-9688-2caab22c7f42 | /documentation/migration-guidance/diagnosing-discrepancies/ | Diagnosing Discrepancies | Configuration-Level Issues > Quantization Differences | documentation | false | Diagnosing Discrepancies > Configuration-Level Issues > Quantization Differences
If your source system uses one quantization scheme and Qdrant uses another (or none), scores will differ. This is expected and doesn't indicate data corruption.
| Source Quantization | Qdrant Quantization | Expected Impact |
| ----- | --... | [
0.049854591488838196,
-0.04439177364110947,
-0.02271214686334133,
-0.035105641931295395,
0.03798414021730423,
-0.009375590831041336,
0.07537759840488434,
0.06269556283950806,
0.07999753206968307,
-0.01923062466084957,
0.034460339695215225,
0.01885465905070305,
0.0477055199444294,
-0.018105... |
2457875d-0187-516e-bde2-47ba8d580a5d | /documentation/private-cloud/api-reference/ | API Reference | qdrant.io/v1 > Resource Types > QdrantCloudRegionSpec | documentation | false | API Reference > qdrant.io/v1 > Resource Types > QdrantCloudRegionSpec
QdrantCloudRegionSpec defines the desired state of QdrantCloudRegion
_Appears in:_
- [QdrantCloudRegion](#qdrantcloudregion)
| Field | Description | Default | Validation |
| --- | --- | --- | --- |
| `id` _string_ | Id specifies the unique identif... | [
0.042652495205402374,
-0.019831590354442596,
-0.010918296873569489,
-0.061125993728637695,
0.10218658298254013,
-0.04191555455327034,
0.03810692951083183,
0.08457627147436142,
0.04270986467599869,
-0.03263717144727707,
0.04374413192272186,
0.05032079294323921,
0.04769313335418701,
-0.02408... |
246cbab1-f573-5f2b-9e7d-8a581ccd69d9 | /course/essentials/day-0/qdrant-cloud/ | Qdrant Setup | Other ways to connect | course | true | Qdrant Setup > Other ways to connect
You can also send your key in the `Authorization` header:
```bash
# Using api-key header
curl -X GET https://xyz-example.eu-central.aws.cloud.qdrant.io:6333/collections \
--header 'api-key: <your-api-key>'
# Using Authorization header
curl -X GET https://xyz-example.eu-centra... | [
0.018521305173635483,
-0.054586563259363174,
-0.0211173165589571,
-0.015324631705880165,
0.037456367164850235,
-0.038902413100004196,
0.019753847271203995,
0.06541749089956284,
0.052487798035144806,
-0.00640694797039032,
0.019955748692154884,
-0.006285803392529488,
0.08563553541898727,
-0.... |
247a51a5-2046-554d-9d68-137cb4a0de27 | /articles/filtered-vector-search-acorn/ | Filtered Vector Search: What ACORN Fixes, and What Fixes ACORN | ACORN on a Normal Collection | article | false | Filtered Vector Search: What ACORN Fixes, and What Fixes ACORN > ACORN on a Normal Collection
When the planner stays on the graph, ACORN is expensive: 5.4x latency on the 20% filter and 6.7x on the 4% intersection. When it routes most queries to the payload index instead, ACORN is nearly free: the 1% intersection gain... | [
0.024030188098549843,
-0.058915991336107254,
-0.04822292551398277,
-0.0051483516581356525,
0.08503276109695435,
0.001919348374940455,
0.02942611835896969,
0.006910668686032295,
0.06301449239253998,
-0.00296186120249331,
0.06330549716949463,
0.018052347004413605,
0.04375035688281059,
0.0021... |
249706c2-9236-59a0-8bf1-3efc066d7329 | /course/essentials/day-3/pitstop-project/ | Project: Building a Hybrid Search Engine | Setup | course | false | Project: Building a Hybrid Search Engine > Setup
### Prerequisites
* Qdrant Cloud cluster (URL + API key)
* Python 3.9+ (or Google Colab)
* Packages: `qdrant-client`, `sentence-transformers`
### Models
* Dense encoder: `sentence-transformers/all-MiniLM-L6-v2` (384-dim)
### Dataset
* A small domain dataset (e.g., ... | [
0.038961172103881836,
-0.014686618000268936,
-0.06973292678594589,
-0.023476801812648773,
0.06362152099609375,
-0.045270368456840515,
0.08327558636665344,
0.035040244460105896,
0.05804208666086197,
-0.04628884419798851,
0.0406264029443264,
0.0077005838975310326,
0.07657770067453384,
-0.029... |
2498d7eb-0e47-5079-9543-f86a455c0e67 | /documentation/search/search-relevance/ | Search Relevance | documentation | false | Search Relevance
By default, Qdrant ranks search results based on vector similarity scores. However, you may wish to consider additional factors when ranking results. Qdrant offers several tools to help you accomplish this. | [
0.05550580471754074,
-0.0862223356962204,
-0.06261010468006134,
-0.04248759150505066,
0.03394230827689171,
-0.022933045402169228,
0.09521625936031342,
0.01444387249648571,
0.09505515545606613,
-0.012311930768191814,
0.054774705320596695,
0.018469786271452904,
0.06757236272096634,
-0.003669... | |
249cfaaf-02bf-5cca-9686-3c5b4eda016d | /documentation/search/low-latency-search/ | Low-Latency Search | Scale Horizontally with Replicas | documentation | false | Low-Latency Search > Scale Horizontally with Replicas
Qdrant can be deployed in a [distributed configuration](/documentation/scaling/distributed_deployment/). In distributed mode, multiple instances of Qdrant, called peers, operate as a single entity, called a cluster. Data is stored in [collections](/documentation/ma... | [
0.01638106070458889,
-0.018227469176054,
-0.0719854012131691,
-0.0040996987372636795,
0.0730472058057785,
-0.05397089570760727,
0.06081821769475937,
0.06302545219659805,
0.06532268226146698,
-0.02285175770521164,
0.015143719501793385,
0.003582422388717532,
0.03886685520410538,
-0.022483278... |
24a4dc7f-7c42-5682-943f-4b2fb4663054 | /documentation/private-cloud/api-reference/ | API Reference | qdrant.io/v1 > Resource Types > QdrantClusterScheduledSnapshotList | documentation | false | API Reference > qdrant.io/v1 > Resource Types > QdrantClusterScheduledSnapshotList
QdrantClusterScheduledSnapshotList contains a list of QdrantCluster
| Field | Description | Default | Validation |
| --- | --- | --- | --- |
| `apiVersion` _string_ | `qdrant.io/v1` | | |
| `kind` _string_ | `QdrantClusterScheduledSnap... | [
0.022914713248610497,
-0.013797586783766747,
-0.01558239571750164,
-0.061334628611803055,
0.04859837517142296,
-0.060237810015678406,
0.06531418114900589,
0.06542862206697464,
0.0567600280046463,
-0.04823140799999237,
0.06341312825679779,
0.006462614983320236,
0.03532784804701805,
-0.00470... |
24e49938-33bd-5dc9-b5ae-16ec9ff178e5 | /documentation/ops-optimization/read-write-contention/ | Read-Write Contention | Step 7: Lower the Maximum Segment Size | documentation | true | Read-Write Contention > Step 7: Lower the Maximum Segment Size
Lowering the maximum segment size can reduce query throughput. This is a deliberate tradeoff: improved read latency under write load at the cost of overall query throughput.
Qdrant builds an [HNSW index](/documentation/manage-data/indexing/#vector-index) ... | [
0.013896888121962547,
-0.0465976782143116,
-0.0595780685544014,
-0.02166161872446537,
0.03184624761343002,
-0.04312938079237938,
0.031623031944036484,
0.04960726574063301,
0.0793011412024498,
-0.026384105905890465,
0.020321713760495186,
0.004069776274263859,
0.043862804770469666,
-0.014785... |
25048d94-5c23-5a0f-a7d2-c8e5eed0e169 | /documentation/tutorials-build-essentials/video-anomaly-edge-part-2/ | Video Anomaly Detection Part 2: Edge-to-Cloud Pipeline | Edge-to-Cloud Escalation > Escalation Flow | documentation | true | Video Anomaly Detection Part 2: Edge-to-Cloud Pipeline > Edge-to-Cloud Escalation > Escalation Flow
1. **Edge detection**: DeepStream inference produces an embedding, Qdrant Edge kNN scores above `ESCALATION_THRESHOLD` (0.06)
2. **Queue**: Metropolis IoT Gateway queues the clip for secure cloud transport (persists acr... | [
0.05920638144016266,
-0.01558141503483057,
-0.06983129680156708,
-0.028128204867243767,
0.051109664142131805,
-0.010716401971876621,
0.06659377366304398,
0.02966594137251377,
0.08238759636878967,
-0.04037283733487129,
0.03256624564528465,
-0.004738376010209322,
0.04144998639822006,
-0.0068... |
254e63d3-78c6-5f80-a6c2-2f251552ab35 | /course/essentials/day-5/universal-query-demo/ | Demo: Universal Query for Hybrid Retrieval | Real ArXiv Dataset Integration | course | true | Demo: Universal Query for Hybrid Retrieval > Real ArXiv Dataset Integration
Here's how you could populate the collection with real arXiv data:
```python
# ! pip install arxiv
import arxiv
arxiv_client = arxiv.Client()
search = arxiv.Search(
query="transformer AND multimodal",
max_results=2,
sort_by=arxi... | [
0.06217173859477043,
0.018027540296316147,
-0.05706790089607239,
-0.060946401208639145,
0.1046079695224762,
-0.03704626113176346,
0.024556253105401993,
0.02631715126335621,
0.03818163275718689,
-0.03281551972031593,
0.05001158267259598,
0.04087114334106445,
0.08235068619251251,
-0.00717782... |
2559e487-1b8d-5f3d-b3bd-46bdc0096e55 | /course/multi-vector-search/module-3/final-project/ | Final Project: Build Your Own Multi-Vector Search System | Suggested Approach | course | false | Final Project: Build Your Own Multi-Vector Search System > Suggested Approach
These hints are optional. Feel free to chart your own path.
**Start simple.** Get a basic multi-vector search working before adding optimizations. It's easier to measure the impact of changes when you have a baseline.
**Create ground truth... | [
0.027724217623472214,
-0.09327885508537292,
-0.03918527066707611,
0.0018239590572193265,
0.06863299012184143,
0.00024312461027875543,
0.054132018238306046,
0.010338877327740192,
0.05658989027142525,
-0.03154371306300163,
0.03426406532526016,
0.012528783641755581,
0.05653459206223488,
-0.06... |
25655ad6-6a64-5150-ae4b-032ae195c0a4 | /documentation/frameworks/rocketride/ | RocketRide | Configure the Qdrant node | documentation | true | RocketRide > Configure the Qdrant node
For a self-hosted Qdrant server, use the `local` profile. This is the ingestion-side Qdrant node from the example:
```json
{
"id": "qdrant_1",
"provider": "qdrant",
"config": {
"profile": "local",
"local": {
"host": "${ROCKETRIDE_QDRANT_HOST}",
"port": ... | [
0.07411414384841919,
-0.01945045404136181,
-0.04060584306716919,
-0.03418857976794243,
0.0667342022061348,
-0.05215907841920853,
0.07453162968158722,
0.06958985328674316,
0.06465044617652893,
-0.025311008095741272,
0.01842738315463066,
-0.022122330963611603,
0.0471651516854763,
-0.04373918... |
25710276-8be6-5e45-946b-1e2e8cbb360e | /documentation/migrate-to-qdrant/from-pgvector/ | From pgvector | After Migration: Keeping Postgres and Qdrant in Sync | documentation | false | From pgvector > After Migration: Keeping Postgres and Qdrant in Sync
If you continue using Postgres as your source of truth alongside Qdrant, you'll need a sync strategy. The [Data Synchronization Guide](/documentation/data-synchronization/) covers three approaches from simple dual-writes to production-grade Change Da... | [
0.0417732298374176,
-0.053063683211803436,
0.000341649167239666,
-0.040582116693258286,
0.006307522766292095,
-0.021988166496157646,
0.07590912282466888,
0.01620728150010109,
0.06040377542376518,
0.003927589859813452,
0.022149546071887016,
0.032267872244119644,
0.05466047301888466,
-0.0122... |
25a37028-0e86-5d44-bb61-f6d71f6a58d1 | /articles/memory-consumption/ | Minimal RAM to Serve 1M Vectors | Let's run some benchmarks > All in Memory | article | false | Minimal RAM to Serve 1M Vectors > Let's run some benchmarks > All in Memory
In the first experiment, we tested how well our system performs when all vectors are stored in memory.
We tried using different amounts of memory, ranging from 1512mb to 1024mb, and measured the number of requests per second (rps) that our sys... | [
0.009518766775727272,
-0.04691716656088829,
-0.02714838832616806,
-0.013518738560378551,
0.0634794533252716,
-0.043747518211603165,
0.030378451570868492,
0.04270944371819496,
0.10054775327444077,
-0.030423574149608612,
-0.0011287188390269876,
0.012811458669602871,
0.06984123587608337,
-0.0... |
25b4f716-4841-5228-a176-55ea53826e29 | /documentation/examples/graphrag-qdrant-neo4j/ | GraphRAG with Qdrant and Neo4j | Conclusion | documentation | false | GraphRAG with Qdrant and Neo4j > Conclusion
This example uses vector search to select paragraphs, expands their entities' graph neighborhoods, and passes the resulting context to an LLM for answer generation.
This implementation template we've explored offers a foundation for your projects. You can adapt and customiz... | [
0.006351896096020937,
-0.08391536772251129,
-0.060093916952610016,
-0.05100029334425926,
0.08740357309579849,
-0.01723039522767067,
0.062257591634988785,
0.0676092579960823,
0.036188527941703796,
-0.018637672066688538,
-0.019385209307074547,
0.028491515666246414,
0.036084987223148346,
-0.0... |
25c52c0b-7b73-5d53-849d-3de0e13785b0 | /course/multi-vector-search/module-3/final-project/ | Final Project: Build Your Own Multi-Vector Search System | Project Requirements | course | false | Final Project: Build Your Own Multi-Vector Search System > Project Requirements
Your project should demonstrate:
### 1. Working Multi-Vector Search
**Use ColModernVBERT** to index your documents and retrieve results using MaxSim scoring. The system should return relevant documents for natural language queries. Alter... | [
0.040160320699214935,
-0.041109129786491394,
-0.03766576200723648,
-0.01193696353584528,
0.07165712118148804,
-0.01966330036520958,
0.04251802712678909,
0.014193516224622726,
0.06834296137094498,
-0.029506418853998184,
0.014480669051408768,
0.018721476197242737,
0.04850219190120697,
-0.023... |
25db499a-9170-572a-98ae-3364b49d847c | /documentation/cloud/cluster-monitoring/ | Monitor Clusters | Alerts | documentation | false | Monitor Clusters > Alerts
If your cluster is highly available you may avoid total downtime but the situation is still unstable. Without high availability you should expect disruption until your cluster has been scaled up.
**What can I do to resolve this?**
Ensuring your cluster is highly available will mitigate the ... | [
0.031223705038428307,
-0.02557372860610485,
-0.07125923037528992,
-0.037712227553129196,
0.08256052434444427,
-0.03819839656352997,
0.04426794499158859,
0.046251021325588226,
0.0770605131983757,
-0.05160738155245781,
0.00372037454508245,
0.00035810936242341995,
0.06255851686000824,
-0.0377... |
25f982b0-1e5b-5b66-b4b6-3a5ca995956b | /documentation/fastembed/fastembed-minicoil/ | Working with miniCOIL | Setup | documentation | true | Working with miniCOIL > Setup
Install `qdrant-client` integration with `fastembed`.
```python
pip install "qdrant-client[fastembed]"
```
Then, initialize the Qdrant client. You could use for experiments [a free cluster](https://qdrant.tech/documentation/cloud-quickstart/#3-connect-to-qdrant-cloud) in Qdrant Cloud or... | [
0.05386415123939514,
-0.04950057342648506,
-0.022256428375840187,
-0.011051647365093231,
0.08876980096101761,
-0.04907272756099701,
0.07359198480844498,
0.023505210876464844,
0.02482146956026554,
-0.005695159547030926,
0.034619010984897614,
0.005102310795336962,
0.07159271091222763,
-0.026... |
25fde8e1-d3c0-5034-8bb3-48238d18539e | /documentation/hybrid-cloud/hybrid-cloud-cluster-creation/ | Create a Cluster | Authentication to your Qdrant Clusters | documentation | true | Create a Cluster > Authentication to your Qdrant Clusters
By default, clusters in Hybrid Cloud are only exposed through a Kubernetes ClusterIP Service inside the Kubernetes network and not accessible to the outside, and no API key is configured. If you choose to expose the database internally or externally, you must c... | [
0.04531780630350113,
-0.0052641513757407665,
-0.05034174770116806,
-0.03798987716436386,
0.0443793348968029,
-0.059397269040346146,
0.08586286753416061,
0.07293776422739029,
0.04718218371272087,
-0.041633035987615585,
0.04371162876486778,
0.018250737339258194,
0.07632788270711899,
-0.03639... |
261a6e4d-2eda-50c9-a812-1ef7fc6e0611 | /documentation/edge/edge-api/configuration/ | Configuration | WAL Options | documentation | true | Configuration > WAL Options
*Rust only*
Qdrant Edge records every update in a write-ahead log before applying it to storage. `WalOptions` is available in Rust only, and is set through `EdgeConfig.wal_options`.
```rust
pub struct WalOptions {
pub segment_capacity: usize,
pub segment_queue_len: usize,
pub ... | [
0.02181064337491989,
-0.031256962567567825,
-0.03321019932627678,
-0.03608220815658569,
0.06384493410587311,
-0.0572369210422039,
0.026642970740795135,
0.05603181943297386,
0.057775914669036865,
-0.014608372002840042,
0.05723514035344124,
-0.005644398741424084,
0.02165442518889904,
-0.0313... |
2620b2eb-3b12-5022-9d14-bd93c102bb42 | /articles/what-is-vector-quantization/ | What is Vector Quantization? | Performance of Quantized vs. Non-Quantized Data | article | true | What is Vector Quantization? > Performance of Quantized vs. Non-Quantized Data
Qdrant uses the quantized vectors by default if they are available. If you want to evaluate how quantization affects your search results, you can temporarily disable it to compare results from quantized and non-quantized searches. To do thi... | [
0.06752046942710876,
-0.05766870826482773,
-0.05459664389491081,
-0.026164203882217407,
0.05057500675320625,
0.0111619858071208,
0.0632529929280281,
0.009191369637846947,
0.09945502877235413,
-0.053492121398448944,
0.03208768740296364,
0.02158624865114689,
0.04686490073800087,
-0.038157418... |
262714a1-816b-519e-b25c-a4c19955c990 | /course/multi-vector-search/module-0/installing-dependencies/ | Installing Dependencies | Installing Dependencies | course | true | Installing Dependencies > Installing Dependencies
With your virtual environment activated, install the required libraries:
**Using uv (recommended):**
```bash
uv pip install "qdrant-client>=1.16.2" "fastembed>=0.8.0"
```
**Using Poetry:**
```bash
poetry add "qdrant-client>=1.16.2" "fastembed>=0.8.0"
```
**Using p... | [
0.035281114280223846,
-0.04700563848018646,
-0.02728925086557865,
-0.023789331316947937,
0.06709983199834824,
-0.04080897197127342,
0.05658343806862831,
0.04249504581093788,
0.07992172241210938,
-0.02984452247619629,
0.0449354462325573,
0.01579056866466999,
0.04710083082318306,
-0.04783481... |
26321446-69a5-5153-a431-5ad16474dc02 | /documentation/scaling/distributed_deployment/ | Distributed Deployment | Sharding > Choosing the Right Number of Shards | documentation | true | Distributed Deployment > Sharding > Choosing the Right Number of Shards
When you create a collection, Qdrant splits the collection into `shard_number` shards. If left unset, `shard_number` is set to the number of nodes in your cluster when the collection was created. The `shard_number` cannot be changed without recrea... | [
0.013780897483229637,
-0.019932523369789124,
-0.05327492579817772,
-0.017259962856769562,
0.06278133392333984,
-0.07152091711759567,
0.02280285209417343,
0.05863611400127411,
0.05636895075440407,
-0.04206695780158043,
0.03610682114958763,
0.025459907948970795,
0.04635009542107582,
-0.03794... |
2638e263-923b-5689-8c16-cfd9ff8f8065 | /documentation/private-cloud/private-cloud-setup/ | Setup Private Cloud | Requirements > Required artifacts | documentation | false | Setup Private Cloud > Requirements > Required artifacts
Container images:
- `registry.cloud.qdrant.io/qdrant/qdrant`
- `registry.cloud.qdrant.io/qdrant/operator`
- `registry.cloud.qdrant.io/qdrant/cluster-manager`
- `registry.cloud.qdrant.io/qdrant/qdrant-cluster-exporter`
Open Containers Initiative (OCI) Helm chart... | [
0.013577593490481377,
-0.0365535169839859,
-0.015955429524183273,
-0.05255480110645294,
0.035995543003082275,
-0.051748570054769516,
0.030293701216578484,
0.05491086095571518,
0.038092877715826035,
-0.06861811131238937,
0.03443728759884834,
0.01470274105668068,
0.08096638321876526,
-0.0632... |
26498e36-055b-5956-b74e-e6d217089765 | /documentation/examples/graphrag-qdrant-neo4j/ | GraphRAG with Qdrant and Neo4j | Architecture Overview | documentation | false | GraphRAG with Qdrant and Neo4j > Architecture Overview
The architecture has two main components: **Ingestion** and **Retrieval & Generation**. Ingestion processes raw data into structured knowledge and vector representations, while Retrieval and Generation enable efficient querying and response generation.
Let’s star... | [
-0.027943897992372513,
-0.020504945889115334,
-0.0497819185256958,
-0.03397270292043686,
0.08133459836244583,
-0.03767647594213486,
0.06511184573173523,
0.0720018669962883,
0.0554879754781723,
-0.03616945073008537,
0.0352468304336071,
0.04984712228178978,
0.06155822053551674,
-0.0031602019... |
266e9e1b-b802-564d-9f8c-e1a3d451469f | /documentation/cloud-security/ | Cloud Security | Software Bill of Materials (SBOM) and Container Image Security | documentation | true | Cloud Security > Software Bill of Materials (SBOM) and Container Image Security
We provide a Software Bill of Materials (SBOM) for all Qdrant clusters and Qdrant Cloud components running on your infrastructure in Qdrant Hybrid Cloud and Qdrant Private Cloud. An SBOM is attached to every released container image. You c... | [
0.049469899386167526,
-0.00550887593999505,
-0.03881318122148514,
-0.07263905555009842,
0.02172759734094143,
-0.05192599445581436,
0.10694307833909988,
0.03564704209566116,
0.05514518916606903,
-0.005673161242157221,
0.046860985457897186,
-0.0021079936996102333,
0.08426439017057419,
-0.026... |
26b67e4f-8336-5a7d-8e9e-3dc60582a310 | /documentation/manage-data/collections/ | Collections | Update Collection > Update Vector Parameters | documentation | true | Collections > Update Collection > Update Vector Parameters
*Available as of v1.4.0*
To update vector parameters using the collection update API, you must always specify a vector name. If your collection does not have named vectors, use an empty ("") name.
Qdrant 1.4 adds support for updating more collection paramete... | [
0.045071449130773544,
-0.061174508184194565,
-0.014743136242032051,
-0.035761758685112,
0.05266430601477623,
-0.03823428973555565,
0.05152444541454315,
0.05777762085199356,
0.0784502923488617,
-0.034784022718667984,
0.013102068565785885,
0.0516270250082016,
0.04751063138246536,
-0.04954626... |
26c3bfd6-de16-54f4-a52f-dd4a24f510ce | /articles/muvera-embeddings/ | MUVERA: Making Multivectors More Performant | How MUVERA Embeddings are created > Final random projection | article | false | MUVERA: Making Multivectors More Performant > How MUVERA Embeddings are created > Final random projection
The original paper suggests applying an additional random projection to the final FDE to further reduce its
dimensionality. It is an optional step that can be used if the resulting FDE is still too large. This fi... | [
0.07745682448148727,
-0.04427257180213928,
-0.015764104202389717,
-0.08810752630233765,
0.0712290108203888,
-0.019062848761677742,
0.06740571558475494,
0.052109528332948685,
0.08717764914035797,
-0.0075532663613557816,
0.03667738288640976,
0.0533241406083107,
0.03897389769554138,
-0.025859... |
26d34994-85bb-514e-a9fb-38afaa92dd2d | /articles/minicoil-en-es/ | miniCOIL EN-ES: Sparse Neural Retrieval Across the Language Barrier | Same Word, Different Language > Making Sparse Vectors Polyglots | article | false | miniCOIL EN-ES: Sparse Neural Retrieval Across the Language Barrier > Same Word, Different Language > Making Sparse Vectors Polyglots
We want cross-lingual matching with no translation step and no reliance on dense retrieval alone.
Instead, we want to reuse the miniCOIL recipe: keep BM25's reliable signals, a word's ... | [
0.04130145534873009,
-0.036616042256355286,
-0.025796853005886078,
-0.05155773460865021,
0.06415067613124847,
-0.056148458272218704,
0.0495406910777092,
0.034327078610658646,
0.05959995836019516,
-0.033214468508958817,
0.019882166758179665,
0.011730958707630634,
0.08763979375362396,
-0.010... |
26dd78ca-e5ef-589b-9880-3625236aed7e | /documentation/tutorials-search-engineering/pdf-retrieval-at-scale/ | Multivector Document Retrieval | Embedding and Mean Pooling | documentation | true | Multivector Document Retrieval > Embedding and Mean Pooling
We'll use a function that generates multivector representations and their mean pooled versions of each PDF page (aka image) in batches.
For complete understanding, it's important to consider the following specifics of **ColPali** and **ColQwen**:
**ColPali:*... | [
0.0003270847664680332,
-0.008094141259789467,
-0.03241443261504173,
-0.0354749895632267,
0.08413584530353546,
-0.03511784225702286,
0.026363136246800423,
0.08485352247953415,
0.03887912258505821,
0.00851601641625166,
0.029608935117721558,
0.011628848500549793,
0.04176465794444084,
-0.03490... |
270306fc-890c-5f47-bc58-382bf35fb920 | /articles/minicoil-en-es/ | miniCOIL EN-ES: Sparse Neural Retrieval Across the Language Barrier | Constructing a Retriever | article | false | miniCOIL EN-ES: Sparse Neural Retrieval Across the Language Barrier > Constructing a Retriever
Obtaining per-concept mini-vectors is only halfway to multilingual sparse neural retrieval. The other half is turning them into one sparse vector that, used in retrieval, will rank multilingual results by their relevance. | [
0.04128362610936165,
-0.034287501126527786,
-0.004226210527122021,
-0.05830850824713707,
0.04859388992190361,
-0.018468815833330154,
0.08258869498968124,
0.052969712764024734,
0.07771104574203491,
-0.024723084643483162,
-0.0012750899186357856,
0.07257922738790512,
0.05654198303818703,
-0.0... |
272d2bc2-5c0b-55e7-a5ca-9f2149e88696 | /documentation/search-patterns/choose-embedding-model/ | How to Choose an Embedding Model: Evaluation & Tradeoffs | Balancing All Aspects | documentation | false | How to Choose an Embedding Model: Evaluation & Tradeoffs > Balancing All Aspects
Remember that this doesn't have to be a one-time decision. As your application evolves, you might need to revisit your choice of the embedding model. Qdrant's architecture makes it relatively easy to migrate to a different model if needed... | [
0.03404933959245682,
-0.043286655098199844,
-0.03424998000264168,
-0.021264901384711266,
0.037113845348358154,
-0.06108133867383003,
0.10758242011070251,
0.04915756732225418,
0.05013437569141388,
-0.01665693148970604,
0.015436581335961819,
0.02411006949841976,
0.053152017295360565,
-0.0501... |
2736f603-5352-5860-b9a1-a29b83184242 | /articles/dedicated-vector-search/ | Why Vector Search Needs a Dedicated Database | Vector Index > Filterable Index | article | false | Why Vector Search Needs a Dedicated Database > Vector Index > Filterable Index
Vector search wasn't historically designed for filtering — imposing strict constraints on results. It's inherently fuzzy; every document is, to some extent, both similar and dissimilar to any query — there's no binary "*fits/doesn't fit*" s... | [
0.027730943635106087,
-0.05086969956755638,
-0.06461871415376663,
-0.02816322259604931,
0.04674011468887329,
-0.030077725648880005,
0.0354805588722229,
-0.005449838936328888,
0.08533064275979996,
-0.0010151114547625184,
0.04887612909078598,
0.040415991097688675,
0.07950853556394577,
-0.025... |
274cccac-43b7-587d-a1dc-6315c6fd3639 | /documentation/tutorials-build-essentials/semantic-cache/ | Semantic Caching for RAG | Set Up Cache Lookup > Hard Hits vs Threshold | documentation | true | Semantic Caching for RAG > Set Up Cache Lookup > Hard Hits vs Threshold
At a high threshold with zero wrong hits on our golden set, our cache deals with exact repeats and some light rewordings. That is useful, but could we do better? It'd be nice to also cover hard cases: paraphrases and questions that differ by one w... | [
0.02582200989127159,
-0.03086247853934765,
-0.04675289988517761,
-0.013372316025197506,
0.03778655081987381,
0.003636781359091401,
0.013624421320855618,
0.02097475156188011,
0.05520862713456154,
-0.04049140214920044,
0.0244321096688509,
0.012150568887591362,
0.039392828941345215,
-0.031238... |
27572b50-5e69-515a-b3cc-346faf0beae3 | /documentation/tutorials-search-engineering/sparse-embeddings-ecommerce-part-1/ | Fine-Tuning Sparse Embeddings for E-Commerce Search | Part 1: Why Sparse Embeddings Beat BM25 | Enter Sparse Embeddings | documentation | false | Fine-Tuning Sparse Embeddings for E-Commerce Search | Part 1: Why Sparse Embeddings Beat BM25 > Enter Sparse Embeddings
[Sparse embeddings](https://qdrant.tech/articles/sparse-vectors/) take a fundamentally different approach. Instead of compressing text into a small, dense vector, they project it onto a large vocabul... | [
0.03124837949872017,
-0.04772990942001343,
-0.04804064705967903,
-0.0682847872376442,
0.03647748380899429,
-0.03788122534751892,
0.06587342917919159,
0.039852574467659,
0.05875178053975105,
-0.01944534294307232,
0.005778032820671797,
0.048656169325113297,
0.04326236993074417,
-0.0308029856... |
27781840-56a1-5fbb-a08f-059ee4eef1b2 | /documentation/edge/edge-data-synchronization-patterns/ | Data Synchronization Patterns | Update a Server Collection from an Edge Shard | documentation | true | Data Synchronization Patterns > Update a Server Collection from an Edge Shard
To synchronize data from an Edge Shard to a server collection, implement a dual-write mechanism in your application. When you add or update a point in the Edge Shard, simultaneously store it in a server collection using the Qdrant client.
I... | [
0.0175496693700552,
-0.019067537039518356,
-0.043874092400074005,
-0.021263811737298965,
0.06267449259757996,
-0.06220724806189537,
0.07318704575300217,
0.038748886436223984,
0.09293943643569946,
0.0018031104700639844,
0.032589949667453766,
0.03874628618359566,
0.026367949321866035,
-0.010... |
27796bad-dc36-534e-ae01-90ec0d7e171d | /documentation/tutorials-develop/neural-search/ | Build a Semantic Search API | Create a Qdrant Cloud Cluster | documentation | true | Build a Semantic Search API > Create a Qdrant Cloud Cluster
Next, you need to manage all of your data using a vector search engine. Qdrant lets you store, update or delete created vectors. Most importantly, it lets you search for the nearest vectors via a convenient API.
1. Register for a [Qdrant Cloud account](https... | [
0.05114361643791199,
-0.023744942620396614,
-0.041940320283174515,
-0.0191178135573864,
0.07515885680913925,
-0.01465716864913702,
0.058077625930309296,
0.008611934259533882,
0.06368951499462128,
-0.05697620287537575,
0.029561052098870277,
0.013623734004795551,
0.07477182894945145,
-0.0380... |
277be591-ef20-5040-b0e9-28457323756d | /documentation/migration-guidance/data-integrity/ | Data Integrity | 1. Vector Count Verification | documentation | true | Data Integrity > 1. Vector Count Verification
The simplest check: does the number of vectors in Qdrant match your source system?
```py
from qdrant_client import QdrantClient
client = QdrantClient("localhost", port=6333)
# Get collection info
collection_info = client.get_collection("your_collection")
qdrant_count = ... | [
0.042384058237075806,
-0.061836883425712585,
-0.046863164752721786,
-0.03598674759268761,
0.0401526540517807,
0.006142281461507082,
0.07576672732830048,
0.05396757647395134,
0.07619138807058334,
-0.04857899248600006,
0.015636788681149483,
0.01915271207690239,
0.013949165120720863,
-0.02531... |
27bac3c7-2e39-5f79-801a-6d0709c026d3 | /course/beginners/module-5/references-and-further-reading/ | References and Further Reading | course | false | References and Further Reading
Module 5
- [Named Vectors](/documentation/manage-data/vectors/#named-vectors): declaring more than one vector per point and querying a named one with `using`.
- [Hybrid Queries](/documentation/search/hybrid-queries/): prefetch semantics, Reciprocal Rank Fusion with weights, Distribution... | [
0.029750268906354904,
-0.005161873530596495,
-0.06884252279996872,
-0.02749680168926716,
0.07041241973638535,
-0.04618941247463226,
0.052959050983190536,
0.03975002095103264,
0.08864881843328476,
-0.02363620325922966,
0.04254336655139923,
0.026435451582074165,
0.04091503843665123,
-0.05070... | |
27c97f7a-76a9-5151-99b9-9fff9f9b80b0 | /documentation/private-cloud/api-reference/ | API Reference | qdrant.io/v1 > Resource Types > AuditConfig | documentation | false | API Reference > qdrant.io/v1 > Resource Types > AuditConfig
AuditConfig specifies the audit logging configuration for Qdrant.
_Appears in:_
- [QdrantConfiguration](#qdrantconfiguration)
| Field | Description | Default | Validation |
| --- | --- | --- | --- |
| `enabled` _boolean_ | Enabled specifies whether to enabl... | [
0.0034915972501039505,
-0.04652087390422821,
-0.0049156672321259975,
-0.03615441545844078,
0.05826185643672943,
-0.07310415059328079,
-0.0043484303168952465,
0.09918871521949768,
0.06080571934580803,
-0.026865607127547264,
-0.0028119180351495743,
0.03346395120024681,
0.03851461783051491,
-... |
27ce599f-0a50-548f-a03a-45599a71ea35 | /documentation/agentic-tools/skills/ | Agent Skills | The Qdrant Advisor | documentation | true | Agent Skills > The Qdrant Advisor
The recommended way to use skills is the Qdrant Advisor, a meta-skill that gives your agent live access to the entire skill hierarchy.
Instead of shipping static content of its own, it searches [skills.qdrant.tech](https://skills.qdrant.tech) at runtime and loads only the guidance rel... | [
0.013935989700257778,
-0.04855623468756676,
-0.053554583340883255,
-0.04597438871860504,
0.05484474450349808,
-0.04823628440499306,
0.03236794471740723,
0.03568274527788162,
0.06912949681282043,
-0.05464167892932892,
0.04103389009833336,
0.0009169174009002745,
0.06858143955469131,
-0.05477... |
27f1bf2f-acb4-5118-8656-635f67dfd227 | /documentation/tutorials-search-engineering/branch-aware-search/ | Branch-Aware Search | Related Reading | documentation | false | Branch-Aware Search > Related Reading
- [Filtering](/documentation/search/filtering/) for the full set of filter clauses, including the [nested object filter](/documentation/search/filtering/#nested-object-filter) that binds `by` and `seq` on the same record.
- [Payload Indexing](/documentation/manage-data/indexing/#p... | [
0.06266894936561584,
-0.034486670047044754,
-0.04921582713723183,
-0.06699854880571365,
0.055299192667007446,
-0.023921722546219826,
0.0562884621322155,
0.0790584608912468,
0.09081028401851654,
-0.029599027708172798,
0.03176377713680267,
0.058154162019491196,
0.05662115663290024,
-0.004400... |
286b7589-46d5-587f-94ad-a1a496dc7531 | /articles/what-are-embeddings/ | Vector Embeddings Explained: How They Work in ML & Search | How do embeddings work? | article | false | Vector Embeddings Explained: How They Work in ML & Search > How do embeddings work?
Embeddings are created through neural networks. They capture complex relationships and semantics into [dense vectors](https://www1.se.cuhk.edu.hk/~seem5680/lecture/semantics-with-dense-vectors-2018.pdf) which are more suitable for mach... | [
0.03611447289586067,
-0.030897529795765877,
-0.06896578520536423,
-0.03222441300749779,
0.04112810641527176,
-0.026919063180685043,
0.06499969214200974,
0.048640016466379166,
0.10290869325399399,
-0.008418320678174496,
0.022277984768152237,
0.03094061091542244,
0.044344350695610046,
-0.033... |
286efada-923a-50c8-b0cf-0c42ff457936 | /documentation/cloud-cli/ | Qdrant Cloud CLI | Configuration | documentation | true | Qdrant Cloud CLI > Configuration
`qcloud` can be configured in three ways, listed here from lowest to highest precedence:
### Config File
The config file is stored at `~/.config/qcloud/config.yaml` by default. You can override the location with the `--config` flag. It stores named contexts so you can switch between ... | [
0.04630253091454506,
-0.01983264833688736,
-0.024050084874033928,
-0.06299614161252975,
0.038690339773893356,
-0.05661565437912941,
0.039264559745788574,
0.056437958031892776,
0.04678124189376831,
-0.057131703943014145,
0.0011702210176736116,
0.021329106763005257,
0.05690109357237816,
-0.0... |
2882aa48-5e0b-5eeb-80e0-f7cf6ee11804 | /course/essentials/day-3/hybrid-search-demo/ | Demo: Implementing a Hybrid Search System | The Hybrid Search Challenge | course | false | Demo: Implementing a Hybrid Search System > The Hybrid Search Challenge
Working with both semantic (dense) and lexical (sparse) search presents interesting challenges:
- **Different scoring systems**: Dense typically uses cosine similarity (\[-1, 1\]), sparse uses BM25 (unbounded)
- **Different result sets**: The same... | [
0.04651293158531189,
-0.03337903320789337,
-0.0751318484544754,
-0.030031545087695122,
0.07545215636491776,
-0.03397178277373314,
0.06377017498016357,
0.014209091663360596,
0.07912120968103409,
-0.024713732302188873,
0.03558577224612236,
0.025987401604652405,
0.06863688677549362,
-0.004752... |
288c63fb-1481-5f10-bd32-c43ddb3530e2 | /documentation/private-cloud/api-reference/ | API Reference | qdrant.io/v1 > Resource Types > QdrantClusterSnapshot | documentation | false | API Reference > qdrant.io/v1 > Resource Types > QdrantClusterSnapshot
QdrantClusterSnapshot is the Schema for the qdrantclustersnapshots API
_Appears in:_
- [QdrantClusterSnapshotList](#qdrantclustersnapshotlist)
| Field | Description | Default | Validation |
| --- | --- | --- | --- |
| `apiVersion` _string_ | `qdra... | [
0.028480330482125282,
-0.03785251826047897,
-0.028010079637169838,
-0.05601274222135544,
0.049251049757003784,
-0.03623638674616814,
0.06866753101348877,
0.06419277936220169,
0.047905925661325455,
-0.06149322912096977,
0.0654243603348732,
0.007138300221413374,
0.06053077429533005,
-0.00136... |
28b315c8-5581-5048-a52c-7fbc15585e89 | /articles/relevance-feedback/ | Relevance Feedback in Qdrant | Vector Search Optimization > Tools of a Vector Search Engine > ...Should Be Cheap | article | false | Relevance Feedback in Qdrant > Vector Search Optimization > Tools of a Vector Search Engine > ...Should Be Cheap
Since additional retrieval iterations already add to the search latency, we can't afford to spend too much time or money on gathering feedback. That means:
**Very few documents for feedback**
The set of ... | [
0.03554912656545639,
-0.04602872207760811,
-0.05928463861346245,
-0.05810822919011116,
0.045775577425956726,
-0.030263535678386688,
0.027929337695240974,
0.009899070486426353,
0.08671974390745163,
-0.00046691313036717474,
0.03639327734708786,
-0.005057863425463438,
0.08021526783704758,
-0.... |
28b6712f-2b2b-5cbb-8eb1-37dbe06e76f0 | /documentation/tutorials-search-engineering/sparse-embeddings-ecommerce-part-2/ | Fine-Tuning Sparse Embeddings for E-Commerce Search | Part 2: Training SPLADE on Modal | documentation | false | Fine-Tuning Sparse Embeddings for E-Commerce Search | Part 2: Training SPLADE on Modal
*This is Part 2 of a 5-part tutorial on fine-tuning sparse embeddings for e-commerce search. In [Part 1](/documentation/tutorials-search-engineering/sparse-embeddings-ecommerce-part-1/), we covered why sparse embeddings beat BM25 fo... | [
0.06667932122945786,
-0.043518535792827606,
-0.04397111386060715,
-0.044350989162921906,
0.04151448607444763,
-0.04098246991634369,
0.07059632986783981,
0.01966003328561783,
0.04870091751217842,
-0.006206348072737455,
0.022588666528463364,
0.019979091361165047,
0.05928408354520798,
-0.0384... | |
28b6e90c-32a6-5558-8385-12f4b6bc8c7c | /course/beginners/module-5/course-summary/ | Course Summary | course | false | Course Summary
Module 5
This module completes the Qdrant Beginners course. Here's what was covered:
| Module | Theme | Key concepts covered |
|--------|-------|----------------------|
| Module 1 | Let's Understand Search | Why keyword search fails; how embeddings and semantic search work; the shift from words to mea... | [
0.0642927885055542,
-0.03132309019565582,
-0.03947625681757927,
-0.02634846232831478,
0.06201335787773132,
-0.038373515009880066,
0.06885883957147598,
0.06640686839818954,
0.07976458966732025,
-0.015718985348939896,
0.027944626286625862,
0.028669312596321106,
0.03966836631298065,
-0.031521... | |
28b929d9-2d9c-56b7-a596-02b1b20b7b4f | /documentation/overview/ | Overview | Data Structure | documentation | false | Overview > Data Structure
 a cluster, and set up an [API key](/documentation/cloud/authentication/), you can access your cluster through the integrated Cluster UI, the REST API and the GRPC API.
## Cluster UI
You can access your [Cluster UI](/documentation/web-ui/) via th... | [
0.02784798853099346,
-0.0260617695748806,
-0.048889994621276855,
-0.011194129474461079,
0.054279718548059464,
-0.06407194584608078,
0.05709371715784073,
0.09415862709283829,
0.0441567562520504,
-0.04790451377630234,
0.040214892476797104,
0.03555341064929962,
0.057938676327466965,
-0.049089... | |
29915333-0d5d-5e6b-9c08-6145b50f2b56 | /documentation/overview/ | Overview | Retrieval Process | documentation | false | Overview > Retrieval Process
Vector search is a transformative information retrieval technique that goes beyond keyword matching to find data based on semantic meaning. It begins with **embedding models**, which convert unstructured data (text, images, audio) into **dense vector embeddings**, fixed-length lists of num... | [
0.048160187900066376,
-0.04512951895594597,
-0.04317725822329521,
-0.06297850608825684,
0.054093748331069946,
-0.033730823546648026,
0.07661140710115433,
0.03286457061767578,
0.06877370923757553,
-0.021027030423283577,
0.01717495545744896,
0.03035874292254448,
0.035842571407556534,
-0.0145... |
29a882e5-6973-5d46-822b-bae701014855 | /documentation/tutorials-build-essentials/semantic-cache/ | Semantic Caching for RAG | Create the Cache Collection | documentation | true | Semantic Caching for RAG > Create the Cache Collection
The cache implies a second collection. Each cached question is one point in this collection: the question's embedding is the vector, and the payload holds:
- `question`: the raw question that produced the answer.
- `answer`: the answer to serve.
- `source_doc_ids... | [
0.032383691519498825,
-0.01586192660033703,
-0.042875535786151886,
-0.0322534553706646,
0.059745632112026215,
-0.0403653047978878,
0.03917699307203293,
0.02906361036002636,
0.0665798932313919,
-0.059165503829717636,
0.0020320469047874212,
0.026684913784265518,
0.02750622294843197,
-0.03307... |
29af4bc4-5b80-5449-8c92-8aba55eaf804 | /documentation/frameworks/mastra/ | Mastra | Methods > describeIndex() | documentation | true | Mastra > Methods > describeIndex()
| Name | Type | Description |
|-------------|----------|----------------------------------|
| `indexName` | `string` | Name of the index to describe |
#### Returns
```typescript
interface IndexStats {
dimension: number;
count: number;
metric... | [
0.04760492220520973,
-0.0353650227189064,
-0.035953063517808914,
-0.02762533165514469,
0.03381574898958206,
-0.020224854350090027,
0.03676716238260269,
0.05218735337257385,
0.042360514402389526,
-0.004103527404367924,
0.040909405797719955,
0.03756946325302124,
0.05222710222005844,
-0.02549... |
29bc2673-3bc1-5f2b-8f12-044a624633d1 | /articles/scalar-quantization/ | Scalar Quantization for Vector Search | Benchmarks > Arxiv-titles-384-angular-no-filters | article | false | Scalar Quantization for Vector Search > Benchmarks > Arxiv-titles-384-angular-no-filters
| | Upload and indexing time | Mean search precision, ef = 128 | Mean search time, ef = 128 | Mean search precision, ef = 256 | Mean search time, ef = 256 | Mean search precision, ef = 512 | Mean search time, ef = 512 |
|---|---|-... | [
0.003336523659527302,
-0.019172420725226402,
-0.06089847907423973,
-0.005774129182100296,
0.04487051069736481,
-0.03231889009475708,
0.06659271568059921,
0.05094122514128685,
0.0979432761669159,
-0.05922967568039894,
0.037939220666885376,
0.018437489867210388,
0.03585843741893768,
-0.05517... |
29c28bec-94e3-5149-9cda-46384d0a6407 | /articles/turboquant-quantization/ | TurboQuant in Qdrant | The Quantization Ladder in Qdrant | article | true | TurboQuant in Qdrant > The Quantization Ladder in Qdrant
Before TurboQuant, Qdrant offered two primary production-grade quantization paths:
* **[Scalar Quantization (SQ)](https://qdrant.tech/articles/scalar-quantization/)** — int8 per coordinate. 4x compression. Recall is essentially indistinguishable from float32 on... | [
0.035483699291944504,
-0.03753610700368881,
-0.029794875532388687,
-0.04424789920449257,
0.033813200891017914,
-0.011004289612174034,
0.04017438367009163,
0.07273709774017334,
0.09725254774093628,
-0.03685420751571655,
0.0455939918756485,
0.00966072827577591,
0.03456767275929451,
-0.042037... |
29d77d78-29f7-5c17-9274-56ccb3eabe0c | /documentation/cloud/create-cluster/ | Create a Cluster | Next Steps | documentation | false | Create a Cluster > Next Steps
You will need to connect to your new Qdrant Cloud cluster. Follow [**Authentication**](/documentation/cloud/authentication/) to create one or more API keys.
You can also scale your cluster both horizontally and vertically. Read more in [**Cluster Scaling**](/documentation/cloud/cluster-s... | [
0.020365770906209946,
-0.01680569536983967,
-0.0275731198489666,
-0.03321574628353119,
0.08027742803096771,
-0.05148657411336899,
0.050804268568754196,
0.0486849769949913,
0.06975224614143372,
-0.03936850279569626,
0.03701705485582352,
0.008572245948016644,
0.06811056286096573,
-0.04212100... |
29e100a8-c2a3-571e-84c2-ccf698fe92b4 | /documentation/private-cloud/ | Private Cloud | documentation | false | Private Cloud
Qdrant Private Cloud allows you to manage Qdrant database clusters in any Kubernetes cluster on any infrastructure. It uses the same Qdrant Operator that powers Qdrant Managed Cloud and Qdrant Hybrid Cloud, but without any connection to the Qdrant Cloud Management Console.
On top of the open source Qdra... | [
0.006340205669403076,
-0.0093876076862216,
-0.06647994369268417,
-0.04255251958966255,
0.0396895706653595,
-0.026360195130109787,
0.08872078359127045,
0.03670421242713928,
0.0511602982878685,
-0.048314183950424194,
0.024526765570044518,
0.017111893743276596,
0.04981784150004387,
-0.0345439... | |
29f96f59-11c2-5d0b-8888-3aeca08eadc6 | /documentation/cloud/authentication/ | Authentication | Test Cluster Access | documentation | true | Authentication > Test Cluster Access
After creation, you will receive a code snippet to access your cluster. Your generated request should look very similar to this one:
```bash
curl \
-X GET 'https://xyz-example.cloud-region.cloud-provider.cloud.qdrant.io:6333' \
--header 'api-key: <paste-your-api-key-here>'
```... | [
0.008802446536719799,
-0.026423713192343712,
-0.04417964443564415,
-0.0379013828933239,
0.07450040429830551,
-0.051572658121585846,
0.029531747102737427,
0.05267892777919769,
0.04299754276871681,
-0.029910778626799583,
0.06109639257192612,
0.0449729785323143,
0.06363116949796677,
-0.055525... |
2a137de8-64e5-5042-ba43-c03fc0b3af9c | /documentation/embeddings/qdrant-cloud-inference/ | Qdrant Cloud Inference | Sparse and Multivector Models | documentation | false | Qdrant Cloud Inference > Sparse and Multivector Models
Sparse and multivector models work the same way, with a `Document` in a sparse or multivector field of the collection. The `qdrant/bm25` model runs inside the cluster itself. See [BM25](/documentation/inference/inference-bm25/), and combine sparse and dense result... | [
0.058403100818395615,
-0.030451813712716103,
-0.040295522660017014,
-0.049552448093891144,
0.08006367087364197,
-0.021431250497698784,
0.11973744630813599,
0.03159623220562935,
0.06623619049787521,
-0.023240171372890472,
0.03133339062333107,
0.03521723672747612,
0.053190577775239944,
-0.00... |
2a1b9af0-f5c7-5713-8c72-d329ba7a9fdc | /documentation/production-operations/bulk-data-import/ | Bulk Uploading Data to Qdrant | Option 2: Create Payload Indexes (Before Uploading) | documentation | true | Bulk Uploading Data to Qdrant > Option 2: Create Payload Indexes (Before Uploading)
_Dense vectors_
Create payload indexes before uploading points when you already know which fields will be used for filtering. Qdrant builds extra HNSW links for each payload index to keep filtered vector search accurate and fast.
If ... | [
0.03158870339393616,
-0.02071218006312847,
-0.03559223189949989,
-0.030058562755584717,
0.0745532214641571,
-0.0480850525200367,
0.0476413257420063,
0.026292309165000916,
0.07227244228124619,
-0.04865826293826103,
0.04000859335064888,
0.05388714373111725,
0.03651649132370949,
-0.0214987155... |
2a41ca7a-2eb3-5b68-a412-2434d2392fdc | /documentation/inference/cloud-inference/ | Cloud Inference | Local Inference Compatibility | documentation | true | Cloud Inference > Local Inference Compatibility
The Python SDK offers a unique capability: it supports both [local](/documentation/fastembed/fastembed-semantic-search/) and cloud inference through an identical interface.
You can easily switch between local and cloud inference by setting the `cloud_inference` flag whe... | [
0.04653826728463173,
-0.028287025168538094,
0.0034700436517596245,
-0.059424176812171936,
0.04036843031644821,
-0.005554831586778164,
0.10925577580928802,
0.03645777702331543,
0.06643018871545792,
-0.005630276165902615,
0.014357002452015877,
0.03586443513631821,
0.0341644249856472,
-0.0216... |
2a42b3a1-0241-5ffb-aaa1-e375fd88d0d9 | /articles/scalar-quantization/ | Scalar Quantization for Vector Search | Where Scalar Quantization Fits Today | article | false | Scalar Quantization for Vector Search > Where Scalar Quantization Fits Today
Qdrant offers more quantization methods today than it did in 2023. Scalar quantization is still what the documentation recommends at 4x compression. Since v1.18, 4-bit [TurboQuant](/articles/turboquant-quantization/) gives comparable recall a... | [
0.050752948969602585,
-0.03377511724829674,
-0.04139535874128342,
-0.0234504584223032,
0.04501122236251831,
-0.004281687084585428,
0.07324199378490448,
0.05872350186109543,
0.10717765986919403,
-0.05733766779303551,
0.030292127281427383,
0.00858370028436184,
0.018468467518687248,
-0.036604... |
2a62fa01-09c2-52e5-aaa2-896e19a46f35 | /articles/rag-is-dead/ | Is RAG Dead? Why Long Context Windows Don't Replace RAG | Cost | article | false | Is RAG Dead? Why Long Context Windows Don't Replace RAG > Cost
LLM input is billed per token, so the cost of a query grows with the prompt. This example uses Anthropic's published price for Claude Sonnet 5.5 in October 2026: $2 per million input tokens. A cache read costs 0.05 times that price, and the full 1M-token w... | [
0.016546187922358513,
-0.049989696592092514,
-0.03356251120567322,
-0.00876942090690136,
0.057069502770900726,
-0.018485387787222862,
0.00038174004293978214,
0.058872681111097336,
0.07516055554151535,
-0.032754119485616684,
0.016563639044761658,
0.005008357111364603,
0.013136166147887707,
... |
2a72675e-df0a-51ee-afaf-dfdc60694381 | /documentation/examples/graphrag-qdrant-neo4j/ | GraphRAG with Qdrant and Neo4j | Implementation > Retrieval & Generation > Setting up the Graph Context | documentation | true | GraphRAG with Qdrant and Neo4j > Implementation > Retrieval & Generation > Setting up the Graph Context
The second part of the implementation involves preparing a graph context. We’ll fetch relevant subgraph data from a Neo4j database and format it for the model. Let’s break it down.
```python
def format_graph_contex... | [
0.010421953164041042,
-0.06286264210939407,
-0.03048689104616642,
-0.022640909999608994,
0.05250183120369911,
-0.0025116961915045977,
0.057819243520498276,
0.05266382545232773,
0.034633390605449677,
-0.004955375101417303,
0.030502695590257645,
0.02452443726360798,
0.06605742871761322,
-0.0... |
2a8c8161-f9bc-5070-a615-0fab11d21c9b | /documentation/cloud-security/ | Cloud Security | Security Considerations > Private Cloud | documentation | false | Cloud Security > Security Considerations > Private Cloud
In Qdrant Private Cloud, Qdrant clusters run completely isolated and air-gapped within your infrastructure without any connection to the Qdrant Cloud Console.
Since there is no connection or communication with Qdrant, you are fully responsible for the security ... | [
0.0324392169713974,
0.0029010563157498837,
-0.0654977411031723,
-0.054222892969846725,
0.032902855426073074,
-0.05492129549384117,
0.05455845966935158,
0.05445592850446701,
0.058667976409196854,
-0.04580697789788246,
0.01793099008500576,
0.020776111632585526,
0.044921379536390305,
-0.03342... |
2a9f17f7-814d-5a22-99b7-3025f72783f5 | /articles/triplet-loss/ | Advanced Introduction to Triplet Loss | Batch-all strategy for online triplet mining | article | true | Advanced Introduction to Triplet Loss > Batch-all strategy for online triplet mining
Now we are ready for actually implementing Triplet Loss itself.
Triplet Loss involves several strategies to form or select triplets, and the simplest one is
to use all valid triplets that can be formed from samples in a batch.
This ca... | [
0.04122511297464371,
-0.050393152981996536,
-0.018048515543341637,
-0.0103408582508564,
0.05927085876464844,
0.006876848638057709,
0.012460720725357533,
0.04050176590681076,
0.05436573922634125,
-0.03846805915236473,
0.04799821227788925,
-0.011034800671041012,
0.022116705775260925,
-0.0325... |
2abbb636-8719-5ebf-a369-c3ca7134f160 | /documentation/embeddings/qdrant-cloud-inference/ | Qdrant Cloud Inference | Next Steps | documentation | false | Qdrant Cloud Inference > Next Steps
- [Inference in Qdrant Managed Cloud](/documentation/cloud/inference/): enabling inference on a cluster, inference regions, and billing.
- [Inference API](/documentation/inference/inference-api/): the `Document`, `Image`, and `Object` inference objects.
- [Matryoshka Models](/docume... | [
0.049736928194761276,
-0.03266505151987076,
-0.0340268574655056,
-0.059553321450948715,
0.05648600682616234,
-0.02096194587647915,
0.10995472222566605,
0.03297554701566696,
0.07198343425989151,
-0.028452781960368156,
0.016896957531571388,
0.01654282957315445,
0.06348823755979538,
-0.020702... |
2adb5431-2f15-566e-83ce-4df44b435763 | /course/essentials/day-7/unstructured/ | Integrating with Unstructured.io | Unstructured Enterprise Data Processing > End-to-End Workflow | course | false | Integrating with Unstructured.io > Unstructured Enterprise Data Processing > End-to-End Workflow
The complete workflow demonstrates seamless integration between AWS S3, Unstructured, and Qdrant:
1. **Data Ingestion**:
- Raw content stored in AWS S3 buckets
- Unstructured's S3 connector pulls documents into the ... | [
0.014749955385923386,
-0.017968827858567238,
-0.07892300933599472,
-0.03385656327009201,
0.045356422662734985,
-0.02028358355164528,
0.061819471418857574,
0.024130413308739662,
0.06179816648364067,
-0.02548234723508358,
-0.005345928482711315,
0.006608166266232729,
0.039182279258966446,
-0.... |
2ae4a818-1e4b-5bb6-a2d5-b1897948e5c0 | /course/essentials/day-1/chunking-strategies/ | Text Chunking Strategies | Chunking Strategies: The Shape Matters > 5. Recursive Chunking | course | false | Text Chunking Strategies > Chunking Strategies: The Shape Matters > 5. Recursive Chunking
**The Approach:** Use a fallback hierarchy of separators when data doesn't follow predictable structure.
Recursive splitting uses a fallback hierarchy of separators. You try to split on large blocks first - like headings or para... | [
0.02887059934437275,
-0.02178933098912239,
-0.034647535532712936,
-0.008646558970212936,
0.0683281421661377,
-0.05045488476753235,
0.054674528539180756,
0.03833245858550072,
0.02947409264743328,
-0.007290284615010023,
0.003203377826139331,
0.017444174736738205,
0.05845349654555321,
-0.0186... |
2aef3d17-915a-5027-9121-1d8d14c39974 | /documentation/migrate-to-qdrant/ | Migration Tool | After Migration | documentation | false | Migration Tool > After Migration
Once your data is in Qdrant, verify that everything arrived correctly:
- **[Migration Verification Guide](/documentation/migration-guidance/)** — a structured framework covering data integrity checks and search quality validation.
- **[Keeping Postgres in Sync](/documentation/data-syn... | [
0.04638071730732918,
-0.05741462484002113,
-0.012206755578517914,
-0.04964418709278107,
0.037830859422683716,
-0.01830739714205265,
0.0638495460152626,
0.012384853325784206,
0.060198038816452026,
-0.015589728951454163,
0.03181380033493042,
0.021727828308939934,
0.06550882756710052,
-0.0121... |
2b01b6c1-c9a0-51e5-9ec7-e1d1879955b8 | /articles/io_uring/ | Faster Disk I/O for Vector Search Using io_uring | Discussion | article | false | Faster Disk I/O for Vector Search Using io_uring > Discussion
Looking back, disk IO used to be very serialized; re-positioning read-write
heads on moving platter was a slow and messy business. So the system overhead
didn't matter as much, but nowadays with SSDs that can often even parallelize
operations while offering... | [
0.01897047646343708,
-0.06410384923219681,
-0.03247729316353798,
-0.031673405319452286,
0.09411142766475677,
-0.01955571398139,
0.02007538266479969,
0.003997469320893288,
0.08763989806175232,
0.0023404047824442387,
0.0012985478388145566,
0.003224005224183202,
0.046413179486989975,
-0.02872... |
2b184c74-4484-5d3e-9ff7-3c1b18234bac | /documentation/data-synchronization/with-postgres/ | Keeping Postgres and Qdrant in Sync | Tier 3: Change Data Capture with Debezium + Redpanda > How It Works | documentation | false | Keeping Postgres and Qdrant in Sync > Tier 3: Change Data Capture with Debezium + Redpanda > How It Works
Postgres's Write-Ahead Log (WAL) is a sequential log of every change to the database — it exists for crash recovery and replication. With `wal_level = logical`, external consumers can read this log in a structured... | [
0.01691698469221592,
-0.05552034452557564,
-0.015446046367287636,
-0.02941555343568325,
0.015933120623230934,
-0.03977097570896149,
0.05971219763159752,
0.06780174374580383,
0.036578696221113205,
-0.02283240482211113,
0.0388159453868866,
0.04734382778406143,
0.03911537677049637,
-0.0234718... |
2b23c653-45ac-5293-aa68-6a7381705656 | /documentation/ | Documentation | Ecosystem | documentation | false | Documentation > Ecosystem
- [Frameworks](/documentation/frameworks/index.md) — Integrations with 40+ AI agent and RAG frameworks: LangChain, LlamaIndex, Haystack, CrewAI, AutoGen, Spring AI, and more.
- [Embedding Providers](/documentation/embeddings/index.md): Find an embedding provider by input type, vector type, an... | [
0.0338587686419487,
-0.01367553323507309,
-0.05719631165266037,
-0.03672705590724945,
0.09912048280239105,
-0.04931141436100006,
0.03973381966352463,
0.019302599132061005,
0.050821732729673386,
-0.05485162138938904,
-0.00039159844163805246,
-0.006113589275628328,
-0.005901955999433994,
-0.... |
2b3014de-1b3a-5f26-98dd-01b69e09ef06 | /documentation/production-checklist/ | Production Checklist | 5. Query Optimization | documentation | false | Production Checklist > 5. Query Optimization
Ensure your search is fast, accurate, and efficient under production load.
- **Create [payload indexes](/documentation/manage-data/indexing/#payload-index) on fields used for filtering.**
Payload indexes speed up filtering and reduce load on the system. Identify which fiel... | [
0.02244485355913639,
-0.043666839599609375,
-0.07188713550567627,
-0.04366214945912361,
0.0655745342373848,
-0.02910272777080536,
0.0403641052544117,
0.027289997786283493,
0.08830589801073074,
-0.024141725152730942,
0.04030349850654602,
-0.005282204132527113,
0.04656340554356575,
-0.041964... |
2b355ef1-cf96-53df-b5b6-776ba6d1ab3e | /documentation/tutorials-and-examples/ | Tutorials & Examples | documentation | false | Tutorials & Examples
## Cloud Tutorials & Examples
| Tutorial | Objective | Stack | Time | Level |
| :--- | :--- | :--- | :--- | :--- |
| [Qdrant Cloud Prometheus Monitoring](/documentation/ops-monitoring/managed-cloud-prometheus/) | Observability with Prometheus and Grafana. | Prometheus | 30m | Intermediate |
| [Mo... | [
-0.001250553410500288,
-0.008318564854562283,
-0.04658736661076546,
-0.0461578294634819,
0.08899446576833725,
0.002303489949554205,
0.06889853626489639,
0.03157569840550423,
0.05457788333296776,
-0.024175751954317093,
0.02043779008090496,
-0.02338532544672489,
0.06384556740522385,
-0.02136... | |
2b40746f-62de-56e4-89d9-14fcae7c96c7 | /documentation/data-synchronization/with-postgres/ | Keeping Postgres and Qdrant in Sync | Tier 3: Change Data Capture with Debezium + Redpanda > Architecture | documentation | false | Keeping Postgres and Qdrant in Sync > Tier 3: Change Data Capture with Debezium + Redpanda > Architecture
CDC is architecturally different from the previous two approaches in a fundamental way: **the application code has no awareness of Qdrant**. The FastAPI routes are pure Postgres CRUD — they don't import the Qdrant... | [
0.020144522190093994,
-0.04825454205274582,
-0.011890665628015995,
-0.01455978024750948,
0.02309301495552063,
-0.024535275995731354,
0.06667506694793701,
0.02773885987699032,
0.06056579574942589,
0.019435349851846695,
0.027432400733232498,
0.023944197222590446,
0.05526692420244217,
-0.0283... |
2b4590e6-03d4-556e-b3d0-5f72ae8ac5f2 | /documentation/frameworks/neo4j-graphrag/ | Neo4j GraphRAG | documentation | true | Neo4j GraphRAG
[Neo4j GraphRAG](https://neo4j.com/docs/neo4j-graphrag-python/current/) is a Python package to build graph retrieval augmented generation (GraphRAG) applications using Neo4j and Python. As a first-party library, it offers a robust, feature-rich, and high-performance solution, with the added assurance of... | [
0.03566756099462509,
-0.041285984218120575,
-0.04007776826620102,
-0.05274878814816475,
0.08641249686479568,
-0.04901188984513283,
0.0605364628136158,
0.06597855687141418,
0.004276179242879152,
-0.0466749481856823,
0.03917594254016876,
0.04910118877887726,
0.05532284080982208,
-0.028623569... | |
2b4d7bc8-b84b-5239-b8ab-fd412e5e256c | /documentation/search/text-search/sparse-vector-search/ | Sparse Vector Search | miniCOIL | documentation | false | Sparse Vector Search > miniCOIL
[miniCOIL](/articles/minicoil/) strikes a balance between the flexibility of BM25 and the performance of SPLADE++. miniCOIL is a transformer-based model that generates sparse vectors for text. Unlike SPLADE++, it doesn't use a vocabulary expansion mechanism. To capture the context and m... | [
0.03717266395688057,
-0.027390016242861748,
-0.016023537144064903,
-0.03903982415795326,
0.03837297856807709,
-0.050005022436380386,
0.12218266725540161,
0.03969980403780937,
0.027658378705382347,
-0.012161288410425186,
0.015695273876190186,
0.027691127732396126,
0.0399235337972641,
0.0338... |
2b62b5ae-3028-5e02-8a83-8b390e44522d | /documentation/ops-monitoring/hybrid-cloud-datadog/ | Monitoring with Datadog | Step 1: Install the Datadog Operator | documentation | true | Monitoring with Datadog > Step 1: Install the Datadog Operator
If you haven't installed Datadog in your cluster yet, you can use the [Datadog Operator](https://docs.datadoghq.com/containers/datadog_operator/) to deploy and manage the Datadog Agent. The Operator runs the Agent as a DaemonSet, with one pod on every node... | [
0.02512345463037491,
-0.03794431686401367,
-0.0483965165913105,
0.007178926840424538,
0.03975915536284447,
-0.06115168333053589,
0.049783799797296524,
0.029700225219130516,
-0.01415240578353405,
-0.01771569810807705,
0.01846027746796608,
0.06453139334917068,
0.058603737503290176,
-0.026972... |
2b653b8e-64c6-51d4-9880-667a25892bb4 | /course/multi-vector-search/module-2/how-colpali-works/ | How ColPali Models Work | Implementing Multi-Modal Search with FastEmbed and Qdrant > Loading ColPali with FastEmbed | course | true | How ColPali Models Work > Implementing Multi-Modal Search with FastEmbed and Qdrant > Loading ColPali with FastEmbed
Let's start by loading a ColPali model. FastEmbed supports several ColPali variants, but in this lesson we'll use ColPali v1.3:
```python
from fastembed import LateInteractionMultimodalEmbedding
# Loa... | [
0.033958904445171356,
-0.023001335561275482,
-0.022375566884875298,
-0.05652724951505661,
0.08086536824703217,
-0.03767555207014084,
0.08621976524591446,
0.05410631000995636,
0.0017801150679588318,
-0.012917610816657543,
0.03694291412830353,
0.015409925021231174,
0.04221507906913757,
-0.02... |
2b94dbf5-54a1-5de5-9539-965a2565339f | /documentation/production-operations/predictable-search-latency/ | Configure Qdrant's Optimizer for Predictable Search Latency | Adjacent Work | documentation | false | Configure Qdrant's Optimizer for Predictable Search Latency > Adjacent Work
- Qdrant's [optimizer docs](/documentation/ops-optimization/optimizer/) describe how the indexing, merge, and vacuum optimizers work and how to configure them.
- The [read-write contention guide](/documentation/ops-optimization/read-write-cont... | [
0.009540306404232979,
-0.06291581690311432,
-0.06508123129606247,
-0.02483905293047428,
0.06464821100234985,
-0.022506939247250557,
0.06625505536794662,
0.007751939818263054,
0.07232558727264404,
-0.03815774992108345,
0.045268747955560684,
-0.011243026703596115,
0.06990194320678711,
-0.043... |
2bc43193-d300-5ee4-b61f-fe6b366913f8 | /documentation/security/ | Security | Authentication > Admin API Key > Rotate an Admin API Key | documentation | true | Security > Authentication > Admin API Key > Rotate an Admin API Key
*Available as of v1.17.0*
In a distributed deployment, you can rotate an admin API key without downtime. Use the `alt_api_key` setting to temporarily configure a second API key that acts identically to the primary `api_key`, allowing both the old and... | [
0.028971174731850624,
-0.033933546394109726,
-0.029854875057935715,
-0.028638381510972977,
0.05740666761994362,
-0.051683228462934494,
-0.003911423031240702,
0.023234447464346886,
0.0781896784901619,
-0.0029695373959839344,
0.028599221259355545,
0.014813419431447983,
0.03409793972969055,
-... |
2bccd5c8-e025-56a5-b383-c37edfbc284f | /documentation/tutorials-search-engineering/pdf-retrieval-at-scale/ | Multivector Document Retrieval | Create Qdrant Collections | documentation | true | Multivector Document Retrieval > Create Qdrant Collections
We can now create a collection in Qdrant to store the multivector representations of PDF pages generated by **ColPali** or **ColQwen**.
Collection will include **mean pooled** by rows and columns representations of a PDF page, as well as the **original** mult... | [
0.025044066831469536,
-0.02342606894671917,
-0.017418382689356804,
-0.05863307788968086,
0.0748651847243309,
-0.0267954058945179,
0.056420452892780304,
0.07226748764514923,
0.044817518442869186,
-0.02650747261941433,
0.037216585129499435,
0.014828728511929512,
0.09362103790044785,
-0.02479... |
2bcd5adb-c7dc-58ee-acc8-3eda641c3abe | /documentation/search-precision/reranking-semantic-search/ | Reranking for Better Search | Implementing Vector Search with Reranking > Overview > Ingestion Stage | documentation | false | Reranking for Better Search > Implementing Vector Search with Reranking > Overview > Ingestion Stage
- **Documents:** This is where it all starts. The system takes in raw data or documents that need to be prepped for search—this is your initial input.
- **Embeddings:** Next, these documents are transformed into sparse... | [
0.033406686037778854,
-0.09378072619438171,
-0.05138238146901131,
-0.026425912976264954,
0.08329104632139206,
-0.031950466334819794,
0.047021716833114624,
0.026106910780072212,
0.09615863114595413,
-0.027988595888018608,
0.0195021852850914,
0.016644174233078957,
0.0512821301817894,
-0.0345... |
2bdc07f2-ece3-5e3a-99ea-38308581b854 | /documentation/tutorials-build-essentials/question-answering-langchain/ | Question Answering with LangChain and Qdrant | Why do we need two different models? | documentation | false | Question Answering with LangChain and Qdrant > Why do we need two different models?
Both solve some different tasks. The first model performs feature extraction, by converting the text into vectors, while
the second one helps in text generation or summarization. Disclaimer: this is not the only way to solve that task ... | [
0.018329109996557236,
-0.042569201439619064,
-0.04689372703433037,
-0.029981572180986404,
0.06620977818965912,
-0.029100460931658745,
0.06708373874425888,
-0.01890004612505436,
0.06285180896520615,
0.004751523490995169,
0.013262253254652023,
0.004729933105409145,
0.08081773668527603,
-0.02... |
2bfdfb47-7c5f-5fdf-ace4-e586b3992213 | /documentation/faq/qdrant-fundamentals/ | Qdrant Fundamentals | Compatibility > Is Qdrant compatible with CPUs or GPUs for vector computation? | documentation | false | Qdrant Fundamentals > Compatibility > Is Qdrant compatible with CPUs or GPUs for vector computation?
Qdrant primarily relies on CPU acceleration for scalability and efficiency. However, we also support GPU-accelerated indexing on all major vendors. | [
-0.0007974442560225725,
-0.04468831792473793,
-0.051684051752090454,
-0.05791458860039711,
0.036836061626672745,
-0.01698659546673298,
0.13101662695407867,
0.03263108804821968,
0.10220959037542343,
-0.022932017222046852,
0.03129241615533829,
-0.0039578997530043125,
0.05505751818418503,
-0.... |
2c46a011-bd87-52c9-bf98-7a698e0e0599 | /documentation/migration-guidance/search-quality/ | Search Quality | Three-Tiered Search Quality Checks | documentation | false | Search Quality > Three-Tiered Search Quality Checks
| Tier | Effort | What It Catches | When to Use |
| ----- | ----- | ----- | ----- |
| **Tier 1: Spot-Check** | 15 min | Gross failures: wrong metric, broken filters, obviously wrong results | Every migration |
| **Tier 2: Statistical Sampling** | 1-2 hours | Systemat... | [
0.02780468389391899,
-0.03220023214817047,
-0.06568088382482529,
-0.052177637815475464,
0.045721471309661865,
-0.009193872101604939,
0.03542899340391159,
0.03553527593612671,
0.10117120295763016,
-0.04127993434667587,
0.016826892271637917,
0.019793422892689705,
0.052997004240751266,
-0.047... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.