krzysztof/landing_page
1
0
Fork 0
mirror of https://github.com/qdrant/landing_page.git synced 2026-10-01 08:58:31 +02:00
Code Issues Packages Projects Releases Wiki Activity

Merge remote-tracking branch 'upstream/v1.16-release-blog' into v1.16-release-blog

Browse Source
This commit is contained in:
Abdon Pijpelink
2025-11-18 14:21:51 +01:00
parent 8a92a2ec30 d0b81c5aac
commit c8019b62ac
1 changed files with 84 additions and 25 deletions
Download Patch File Download Diff File Expand all files Collapse all files
qdrant-landing/content/blog/qdrant-1.16.x.md
+84 -25
View File
@@ -65,7 +65,41 @@ To address these limitations, in version 1.16 we are introducing support for [AC
You can enable ACORN on a per-query basis, via the optional [query-time `acorn` parameter](/documentation/concepts/search/#acorn-search-algorithm). This doesn't require any changes at index time.
<!-- TODO: Benchmarks of ACORN ... --->
### Benchmarks
The setup:
a single segment of 5,000,000 vectors of dimension 96 (a subset of deep-image-96 dataset).
Two payload fields, each with 5 possible values, uniformly distributed.
During the search, we apply two filters on both payload fields, resulting in approximately 4% of vectors passing the filter.
<table>
<thead>
<tr>
<th>Search Parameters
<th>Accuracy
<th>Latency
<tbody>
<tr>
<td>ef=64 + ACORN
<td><b>97.20%</b>
<td><b>13.86 ms</b>
<tr>
<td>ef=64
<td>53.34%
<td>1.25 ms
<tr>
<td>ef=128
<td>61.77%
<td>1.46 ms
<tr>
<td>ef=256
<td>67.58%
<td>2.27 ms
<tr>
<td>ef=512
<td>71.13%
<td>3.89 ms
</table>
### When Should You Use ACORN?
@@ -122,7 +156,39 @@ Note that quantization needs to be enabled for inline storage to work efficientl
Using a smaller data type and quantization reduces the size of each vector significantly, making it possible to store them inline. When combining inline storage with `float16` data types and quantization, evaluating a node in the HNSW graph requires reading only two pages from disk, rather than making 32 random access reads. This represents a significant improvement over the traditional approach, at the cost of additional storage space.
<!-- TODO ... Benchmarks of Inline Storage ... --->
Benchmark setup: 1,000,000 vectors (a subset of LAION 512d CLIP embeddings), 2-bit quantization, float16 data type.
<table>
<thead>
<tr>
<th>Setup
<th>Container Memory Limit
<th>Total Storage Used
<th>QPS
<th>Mean Accuracy
<tbody>
<tr>
<td>Inline Storage, low RAM (new)
<td>430 MiB
<td>4992 MiB
<td><b>211</b>
<td><b>86.92%</b>
<tr>
<td>No Inline Storage, low RAM
<td>430 MiB
<td>1206 MiB
<td>20
<td>53.32%
<tr>
<td>No Inline Storage, index & quantized vectors in RAM
<td>530 MiB
<td>1206 MiB
<td>334
<td>53.32%
</table>
The benchmark shows that the inline storage not just improves search performance by an order of magnitude for low RAM setups, but also provides better accuracy thanks to the implicit rescoring during the search.
The search performance is comparable to medium RAM setup which has enough memory to hold HNSW index and quantized vectors in RAM.
Inline storage can be enabled by [setting a collection's HNSW configuration `inline_storage` option to `true`](/documentation/guides/optimize/#inline-storage-in-hnsw-index). It requires quantization to be enabled.
@@ -162,41 +228,34 @@ The `text_any` condition matches text fields that contain any of the query terms
A good example of using the `text_any` condition is in e-commerce applications, where users often search for products using multiple keywords. By combining a vector query with a series of increasingly lenient full-text filters, you can ensure that users receive relevant results even if their initial search terms are too restrictive.
```json
batch [
```python
QUERY = {
"text": "best smartphone ever",
"model": "sentence-transformers/all-MiniLM-L6-v2"
}
batch_query = [
{
"query": {
"text": "best smartphone ever",
"model": "sentence-transformers/all-MiniLM-L6-v2"
},
"query": QUERY,
"filter": {
"must": {
"key": "description",
"match": {
"text": "5G 5000mAh OLED"
}
"match": { "text": "5G 5000mAh OLED" }
}
}
},
{
"query": {
"text": "best smartphone ever",
"model": "sentence-transformers/all-MiniLM-L6-v2"
},
{ # fallback to at least one token
"query": QUERY,
"filter": {
"must": {
"key": "description",
"match": {
"text_any": "5G 5000mAh OLED"
}
"match": { "text_any": "5G 5000mAh OLED" }
}
}
},
{
"query": {
"text": "best smartphone ever",
"model": "sentence-transformers/all-MiniLM-L6-v2"
}
{ # fallback to just similarity
"query": QUERY,
"filter": null
}
]
```
@@ -273,4 +332,4 @@ We recommend upgrading versions one by one, for example, 1.14->1.15->1.16.
![Engage](/blog/qdrant-1.16.x/section-9.png)
We would love to hear your thoughts on this release. If you have any questions or feedback, join our [Discord](https://discord.gg/qdrant) or create an issue on [GitHub](https://github.com/qdrant/qdrant/issues).
We would love to hear your thoughts on this release. If you have any questions or feedback, join our [Discord](https://discord.gg/qdrant) or create an issue on [GitHub](https://github.com/qdrant/qdrant/issues).