Skip to content

Commit 5c5e684

Browse files
andy-stark-redisandy-stark-redis
committed
DOC-5073 added CLI examples for evaluation
1 parent b417faf commit 5c5e684

File tree

2 files changed

+166
-0
lines changed

2 files changed

+166
-0
lines changed

‎content/develop/data-types/vector-sets/_index.md

Lines changed: 162 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -38,3 +38,165 @@ The following commands are available for vector sets:
3838
- [VREM]({{< relref "/commands/vrem" >}}) - remove an element from a vector set.
3939
- [VSETATTR]({{< relref "/commands/vsetattr" >}}) - set or replace attributes on a vector set element.
4040
- [VSIM]({{< relref "/commands/vsim" >}}) - retrieve elements similar to a given vector or element with optional filtering.
41+
42+
## Examples
43+
44+
The following examples give an overview of how to use vector sets. For clarity,
45+
we will use a set of two-dimensional vectors that represent points in the
46+
Cartesian coordinate plane. However, in real use cases, the vectors will typically
47+
represent *text embeddings* and have hundreds of dimensions. See
48+
[Redis for AI]({{< relref "/develop/ai" >}}) for more information about using text
49+
embeddings.
50+
51+
The points we will use are A: (1.0, 1.0), B: (-1.0, -1.0), C: (-1.0, 1.0), D: (1.0. -1.0), and
52+
E: (1.0, 0), shown in the diagram below.
53+
54+
{{<image filename="images/vecsets/VecSetExamplePoints.drawio.svg" alt="Example points on the coordinate plane." width="400px">}}
55+
56+
### Basic operations
57+
58+
Start by adding the point vectors to a set called `points` using
59+
[`VADD`]({{< relref "/commands/vadd" >}}). This also creates the vector set object.
60+
The [`TYPE`]({{< relref "/commands/type" >}}) command returns a type of `vectorset`
61+
for this object.
62+
63+
```
64+
> VADD points VALUES 2 1.0 1.0 pt:A
65+
(integer) 1
66+
> VADD points VALUES 2 -1.0 -1.0 pt:B
67+
(integer) 1
68+
> VADD points VALUES 2 -1.0 1.0 pt:C
69+
(integer) 1
70+
> VADD points VALUES 2 1.0 -1.0 pt:D
71+
(integer) 1
72+
> VADD points VALUES 2 1.0 0 pt:E
73+
(integer) 1
74+
> TYPE points
75+
vectorset
76+
```
77+
78+
Get the number of elements in the set (also known as the *cardinality* of the set)
79+
using [`VCARD`]({{< relref "/commands/vcard" >}}) and the number of dimensions of
80+
the vectors using [`VDIM`]({{< relref "/commands/vdim" >}}):
81+
82+
```
83+
> VCARD points
84+
(integer) 5
85+
> VDIM points
86+
(integer) 2
87+
```
88+
89+
Get the coordinate values from the elements using [`VEMB`]({{< relref "/commands/vemb" >}}).
90+
Note that the values will not typically be the exact values you supplied when you added
91+
the vector because
92+
[quantization]({{< relref "/develop/data-types/vector-sets/performance#quantization-effects" >}})
93+
is applied to improve performance.
94+
95+
```
96+
> VEMB points pt:A
97+
1) "0.9999999403953552"
98+
2) "0.9999999403953552"
99+
9> VEMB points pt:B
100+
1) "-0.9999999403953552"
101+
2) "-0.9999999403953552"
102+
> VEMB points pt:C
103+
1) "-0.9999999403953552"
104+
2) "0.9999999403953552"
105+
> VEMB points pt:D
106+
1) "0.9999999403953552"
107+
2) "-0.9999999403953552"
108+
> VEMB points pt:E
109+
1) "1"
110+
2) "0"
111+
```
112+
113+
Set and retrieve an element's JSON attribute data using
114+
[`VSETATTR`]({{< relref "/commands/vsetattr" >}})
115+
and [`VGETATTR`]({{< relref "/commands/vgetattr" >}}). You can also pass an empty string
116+
to `VSETATTR` to delete the attribute data:
117+
118+
```
119+
> VSETATTR points pt:A "{\"name\": \"Point A\", \"description\": \"First point added\"}"
120+
(integer) 1
121+
> VGETATTR points pt:A
122+
"{\"name\": \"Point A\", \"description\": \"First point added\"}"
123+
> VSETATTR points pt:A ""
124+
(integer) 1
125+
> VGETATTR points pt:A
126+
(nil)
127+
```
128+
129+
Remove an unwanted element with [`VREM`]({{< relref "/commands/vrem" >}})
130+
131+
```
132+
> VADD points VALUES 2 0 0 pt:F
133+
(integer) 1
134+
127.0.0.1:6379> VCARD points
135+
(integer) 6
136+
127.0.0.1:6379> VREM points pt:F
137+
(integer) 1
138+
127.0.0.1:6379> VCARD points
139+
(integer) 5
140+
```
141+
142+
### Vector similarity search
143+
144+
Use [`VSIM`]({{< relref "/commands/vsim" >}}) to rank the points in order of their vector distance from a sample point:
145+
146+
```
147+
> VSIM points values 2 0.9 0.1
148+
1) "pt:E"
149+
2) "pt:A"
150+
3) "pt:D"
151+
4) "pt:C"
152+
5) "pt:B"
153+
```
154+
155+
Find the four elements that are closest to point A and show their distance "scores":
156+
157+
```
158+
> VSIM points ELE pt:A WITHSCORES COUNT 4
159+
1) "pt:A"
160+
2) "1"
161+
3) "pt:E"
162+
4) "0.8535534143447876"
163+
5) "pt:C"
164+
6) "0.5"
165+
7) "pt:D"
166+
8) "0.5"
167+
```
168+
169+
Add some JSON attributes and use
170+
[filter expressions]({{< relref "/develop/data-types/vector-sets/filtered-search" >}})
171+
to include them in the search:
172+
173+
```
174+
> VSETATTR points pt:A "{\"size\":\"large\",\"price\": 18.99}"
175+
(integer) 1
176+
> VSETATTR points pt:B "{\"size\":\"large\",\"price\": 35.99}"
177+
(integer) 1
178+
> VSETATTR points pt:C "{\"size\":\"large\",\"price\": 25.99}"
179+
(integer) 1
180+
> VSETATTR points pt:D "{\"size\":\"small\",\"price\": 21.00}"
181+
(integer) 1
182+
> VSETATTR points pt:E "{\"size\":\"small\",\"price\": 17.75}"
183+
(integer) 1
184+
185+
# Return elements in order of distance from point A whose
186+
# `size` attribute is `large`.
187+
> VSIM points ELE pt:A FILTER '.size == "large"'
188+
1) "pt:A"
189+
2) "pt:C"
190+
3) "pt:B"
191+
192+
# Return elements in order of distance from point A whose size is
193+
# `large` and whose price is greater than 20.00.
194+
> VSIM points ELE pt:A FILTER '.size == "large" && .price > 20.00'
195+
1) "pt:C"
196+
2) "pt:B"
197+
```
198+
199+
## More information
200+
201+
See the other pages in this section to learn more about the features
202+
and performance parameters of vector sets.

0 commit comments

Comments
 (0)