krzysztof/landing_page
1
0
Fork 0
mirror of https://github.com/qdrant/landing_page.git synced 2026-09-29 07:58:31 +02:00
Code Issues Packages Projects Releases Wiki Activity

Merge branch 'master' into v1.3.0-article

Browse Source
This commit is contained in:
David Sertic
2023-06-26 08:53:34 +02:00
committed by GitHub
parent eeb82e70d2 1316db4bac
commit 8566e72c7c
6 changed files with 156 additions and 53 deletions
Download Patch File Download Diff File Expand all files Collapse all files
qdrant-landing/content/documentation/guides/quantization.md
+7 -42
View File
@@ -191,7 +191,8 @@ POST /collections/{collection_name}/points/search
"params": {
"quantization": {
"ignore": false,
"rescore": true
"rescore": true,
"oversampling": 2.0
}
},
"vector": [0.2, 0.1, 0.9, 0.7],
@@ -212,6 +213,7 @@ client.search(
quantization=models.QuantizationSearchParams(
ignore=False,
rescore=True,
oversampling=2.0,
)
)
)
@@ -224,48 +226,11 @@ This can improve the search quality, but may slightly decrease the search speed,
It is recommended to disable rescore only if the original vectors are stored on a slow storage (e.g. HDD or network storage).
By default, rescore is enabled.
### Oversampling
**Available as of v1.3.0**
*Available as of v1.3.0*
Oversampling is another way to tune your query accuracy. Keeping your index the same, you can decide how many points to retrieve using the quantized vectors. This method allows for fast pre-selection and allows for parallel I/O, thus speeding up the retrieval od vectors.
```http
POST /collections/{collection_name}/points/search
{
"params": {
"quantization": {
"ignore": false,
"rescore": true,
"oversampling": 2.4
}
},
"vector": [0.2, 0.1, 0.9, 0.7],
"limit": 100
}
```
```python
from qdrant_client import QdrantClient
from qdrant_client.http import models
client = QdrantClient("localhost", port=6333)
client.search(
collection_name="{collection_name}",
query_vector=[0.2, 0.1, 0.9, 0.7],
search_params=models.SearchParams(
quantization=models.QuantizationSearchParams(
ignore=False,
rescore=True,
oversampling=2.4
)
)
)
```
`oversampling` - The default factor is 1.0. For example, if `oversampling` is 2.4 and `limit` is 100, then 240 vectors will be pre-selected using quantized index, and then top-100 will be returned after re-scoring.
`oversampling` - Defines how many extra vectors should be pre-selected using quantized index, and then re-scored using original vectors.
For example, if oversampling is 2.4 and limit is 100, then 240 vectors will be pre-selected using quantized index, and then top-100 will be returned after re-scoring.
Oversampling is useful if you want to tune the tradeoff between search speed and search quality in the query time.
## Quantization tips