krzysztof/landing_page
1
0
Fork 0
mirror of https://github.com/qdrant/landing_page.git synced 2026-09-28 23:48:31 +02:00
Code Issues Packages Projects Releases Wiki Activity

update images + minor fixes

Browse Source
This commit is contained in:
generall
2023-12-08 10:36:41 +01:00
committed by timvisee
parent 6ffab5f940
commit 306117348b
3 changed files with 2 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files
qdrant-landing/content/documentation/concepts/explore.md
+2 -4
View File
@@ -155,15 +155,13 @@ let score = if best_positive_score > best_negative_score {
```
<aside role="alert">
The performance of `best_score` strategy will be linearly impacted by the amount of examples.
The performance of <code>best_score</code> strategy will be linearly impacted by the amount of examples.
</aside>
Since we are computing similarities to every example at each step of the search, the performance of this strategy will be linearly impacted by the amount of examples. This means that the more examples you provide, the slower the search will be. However, this strategy can be very powerful and should be more embedding-agnostic.
<aside role="status">
Accuracy may be impacted with this strategy. To improve it, increasing the `ef` search parameter to something above 32 will already be much better than the default 16, e.g: `"params": { "ef": 64 }`
Accuracy may be impacted with this strategy. To improve it, increasing the <code>ef</code> search parameter to something above 32 will already be much better than the default 16, e.g: <code>"params": { "ef": 64 }</code>
</aside>
To use this algorithm, you need to set `"strategy": "best_score"` in the recommendation request.