krzysztof/landing_page
1
0
Fork 0
mirror of https://github.com/qdrant/landing_page.git synced 2026-10-09 21:08:31 +02:00
Code Issues Packages Projects Releases Wiki Activity

Default rescoring docs are outdated (#2484)

Browse Source 
* update default rescoring config for turboquant

* Update FAQ too

* Revert changes to qdrant-landing/package-lock.json

---------

Co-authored-by: Nathan LeRoy <nleroy917@users.noreply.github.com>
Co-authored-by: Abdon Pijpelink <abdon.pijpelink@qdrant.com>
This commit is contained in:
Nathan LeRoy
2026-07-09 09:37:04 +02:00
committed by GitHub
co-authored by Nathan LeRoy Abdon Pijpelink
parent 7c93fc12e5
commit 1c751faf3d
2 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files
qdrant-landing/content/documentation/faq/qdrant-fundamentals.md
+1 -1
View File
@@ -205,7 +205,7 @@ By default, `hnsw_ef` equals `ef_construct` (default: 100). `ef_construct` is a
### What are the default `rescore` values for different quantization methods?
Only binary quantization uses rescoring by default. All other quantization methods do not rescore by default. The default behavior can be overridden with the query-time `rescore` parameter.
By default, rescoring is only enabled for four quantization methods: binary quantization, TurboQuant 1 Bit, TurboQuant 1.5 Bit, and TurboQuant 2 Bit. Other quantization methods do not rescore by default. The default behavior can be overridden with the query-time `rescore` parameter.
### What does `ignore=true` do in quantization search params?
qdrant-landing/content/documentation/manage-data/quantization.md
+2 -2
View File
@@ -321,7 +321,7 @@ However, there are a few options that you can use to control the search process:
`ignore` - Toggle whether to ignore quantized vectors during the search process. By default, Qdrant will use quantized vectors if they are available.
`rescore` - Qdrant can re-evaluate top-k search results using the original vectors. While this can improve search quality, it may decrease search speed, especially if the original vectors are stored on disk. In such cases, it is recommended to disable rescoring. By default, rescoring is only enabled for binary quantization. Other quantization methods do not rescore by default.
`rescore` - Qdrant can re-evaluate top-k search results using the original vectors. While this can improve search quality, it may decrease search speed, especially if the original vectors are stored on disk. In such cases, it is recommended to disable rescoring. By default, rescoring is only enabled for four quantization methods: binary quantization, TurboQuant 1 Bit, TurboQuant 1.5 Bit, and TurboQuant 2 Bit. Other quantization methods do not rescore by default.
**Available as of v1.3.0**
@@ -345,7 +345,7 @@ By setting it to a value lower than 1.0, you can exclude extreme values (outlier
For example, if you set the quantile to 0.99, 1% of the extreme values will be excluded.
By adjusting the quantile, you find an optimal value that will provide the best search quality for your collection.
- **Enable rescoring**: Qdrant can re-evaluate top-k search results using the original vectors. While this can improve search quality, it may decrease search speed, especially if the original vectors are stored on disk. In such cases, it is recommended to disable rescoring. By default, rescoring is only enabled for binary quantization. Other quantization methods do not rescore by default.
- **Enable rescoring**: Qdrant can re-evaluate top-k search results using the original vectors. While this can improve search quality, it may decrease search speed, especially if the original vectors are stored on disk. In such cases, it is recommended to disable rescoring. By default, rescoring is only enabled for four quantization methods: binary quantization, TurboQuant 1 Bit, TurboQuant 1.5 Bit, and TurboQuant 2 Bit. Other quantization methods do not rescore by default.
### Memory and Speed Tuning