Timescale Vector

PostgreSQL++ for AI Applications.

• Signup for Timescale Vector: Get 90 days free to try Timescale Vector on the Timescale cloud data platform. There is no self-managed version at this time.
• Documentation: Learn the key features of Timescale Vector and how to use them.
• Getting Started Tutorial: Learn how to use Timescale Vector for semantic search on a real-world dataset.
• Learn more: Learn more about Timescale Vector, how it works and why we built it.

If you prefer to use an LLM development or data framework, see Timescale Vector’s integrations with LangChain and LlamaIndex

Install

To install the main library use:

pip install timescale_vector

We also use dotenv in our examples for passing around secrets and keys. You can install that with:

pip install python-dotenv

If you run into installation errors related to the psycopg2 package, you will need to install some prerequisites. The timescale-vector package explicitly depends on psycopg2 (the non-binary version). This adheres to the advice provided by psycopg2. Building psycopg from source requires a few prerequisites to be installed. Make sure these are installed before trying to pip install timescale_vector.

Basic usage

First, import all the necessary libraries:

from dotenv import load_dotenv, find_dotenv
import os
from timescale_vector import client
import uuid
from datetime import datetime, timedelta

Load up your PostgreSQL credentials. Safest way is with a .env file:

_ = load_dotenv(find_dotenv(), override=True) 
service_url  = os.environ['TIMESCALE_SERVICE_URL']

Next, create the client. In this tutorial, we will use the sync client. But we have an async client as well (with an identical interface that uses async functions).

The client constructor takes three required arguments:

name	description
service_url	Timescale service URL / connection string
table_name	Name of the table to use for storing the embeddings. Think of this as the collection name
num_dimensions	Number of dimensions in the vector

You can also specify the schema name, distance type, primary key type, etc. as optional parameters. Please see the documentation for details.

vec  = client.Sync(service_url, "my_data", 2)

Next, create the tables for the collection:

vec.create_tables()

Next, insert some data. The data record contains:

• A UUID to uniquely identify the embedding
• A JSON blob of metadata about the embedding
• The text the embedding represents
• The embedding itself

Because this data includes UUIDs which become primary keys, we ingest with upserts.

vec.upsert([\
    (uuid.uuid1(), {"animal": "fox"}, "the brown fox", [1.0,1.3]),\
    (uuid.uuid1(), {"animal": "fox", "action":"jump"}, "jumped over the", [1.0,10.8]),\
])

You can now create a vector index to speed up similarity search:

vec.create_embedding_index(client.DiskAnnIndex())

Now, you can query for similar items:

vec.search([1.0, 9.0])

[[UUID('4494c186-4a0d-11ef-94a3-6ee10b77fd09'),
  {'action': 'jump', 'animal': 'fox'},
  'jumped over the',
  array([ 1. , 10.8], dtype=float32),
  0.00016793422934946456],
 [UUID('4494c12c-4a0d-11ef-94a3-6ee10b77fd09'),
  {'animal': 'fox'},
  'the brown fox',
  array([1. , 1.3], dtype=float32),
  0.14489260377438218]]

There are many search options which we will cover below in the Advanced search section.

As one example, we will return one item using a similarity search constrained by a metadata filter.

vec.search([1.0, 9.0], limit=1, filter={"action": "jump"})

[[UUID('4494c186-4a0d-11ef-94a3-6ee10b77fd09'),
  {'action': 'jump', 'animal': 'fox'},
  'jumped over the',
  array([ 1. , 10.8], dtype=float32),
  0.00016793422934946456]]

The returned records contain 5 fields:

name	description
id	The UUID of the record
metadata	The JSON metadata associated with the record
contents	the text content that was embedded
embedding	The vector embedding
distance	The distance between the query embedding and the vector

You can access the fields by simply using the record as a dictionary keyed on the field name:

records = vec.search([1.0, 9.0], limit=1, filter={"action": "jump"})
(records[0]["id"],records[0]["metadata"], records[0]["contents"], records[0]["embedding"], records[0]["distance"])

(UUID('4494c186-4a0d-11ef-94a3-6ee10b77fd09'),
 {'action': 'jump', 'animal': 'fox'},
 'jumped over the',
 array([ 1. , 10.8], dtype=float32),
 0.00016793422934946456)

You can delete by ID:

vec.delete_by_ids([records[0]["id"]])

Or you can delete by metadata filters:

vec.delete_by_metadata({"action": "jump"})

To delete all records use:

vec.delete_all()

Advanced usage

In this section, we will go into more detail about our feature. We will cover:

1. Search filter options - how to narrow your search by additional constraints
2. Indexing - how to speed up your similarity queries
3. Time-based partitioning - how to optimize similarity queries that filter on time
4. Setting different distance types to use in distance calculations

Search options

The search function is very versatile and allows you to search for the right vector in a wide variety of ways. We’ll describe the search option in 3 parts:

1. We’ll cover basic similarity search.
2. Then, we’ll describe how to filter your search based on the associated metadata.
3. Finally, we’ll talk about filtering on time when time-partitioning is enabled.

Let’s use the following data for our example:

vec.upsert([\
    (uuid.uuid1(), {"animal":"fox", "action": "sit", "times":1}, "the brown fox", [1.0,1.3]),\
    (uuid.uuid1(),  {"animal":"fox", "action": "jump", "times":100}, "jumped over the", [1.0,10.8]),\
])

The basic query looks like:

vec.search([1.0, 9.0])

[[UUID('456dbbbc-4a0d-11ef-94a3-6ee10b77fd09'),
  {'times': 100, 'action': 'jump', 'animal': 'fox'},
  'jumped over the',
  array([ 1. , 10.8], dtype=float32),
  0.00016793422934946456],
 [UUID('456dbb6c-4a0d-11ef-94a3-6ee10b77fd09'),
  {'times': 1, 'action': 'sit', 'animal': 'fox'},
  'the brown fox',
  array([1. , 1.3], dtype=float32),
  0.14489260377438218]]

You could provide a limit for the number of items returned:

vec.search([1.0, 9.0], limit=1)

[[UUID('456dbbbc-4a0d-11ef-94a3-6ee10b77fd09'),
  {'times': 100, 'action': 'jump', 'animal': 'fox'},
  'jumped over the',
  array([ 1. , 10.8], dtype=float32),
  0.00016793422934946456]]

Narrowing your search by metadata

We have two main ways to filter results by metadata: - filters for equality matches on metadata. - predicates for complex conditions on metadata.

Filters are more likely to be performant but are more limited in what they can express, so we suggest using those if your use case allows it.

Filters

You could specify a match on the metadata as a dictionary where all keys have to match the provided values (keys not in the filter are unconstrained):

vec.search([1.0, 9.0], limit=1, filter={"action": "sit"})

[[UUID('456dbb6c-4a0d-11ef-94a3-6ee10b77fd09'),
  {'times': 1, 'action': 'sit', 'animal': 'fox'},
  'the brown fox',
  array([1. , 1.3], dtype=float32),
  0.14489260377438218]]

You can also specify a list of filter dictionaries, where an item is returned if it matches any dict:

vec.search([1.0, 9.0], limit=2, filter=[{"action": "jump"}, {"animal": "fox"}])

[[UUID('456dbbbc-4a0d-11ef-94a3-6ee10b77fd09'),
  {'times': 100, 'action': 'jump', 'animal': 'fox'},
  'jumped over the',
  array([ 1. , 10.8], dtype=float32),
  0.00016793422934946456],
 [UUID('456dbb6c-4a0d-11ef-94a3-6ee10b77fd09'),
  {'times': 1, 'action': 'sit', 'animal': 'fox'},
  'the brown fox',
  array([1. , 1.3], dtype=float32),
  0.14489260377438218]]

Predicates

Predicates allow for more complex search conditions. For example, you could use greater than and less than conditions on numeric values.

vec.search([1.0, 9.0], limit=2, predicates=client.Predicates("times", ">", 1))

[[UUID('456dbbbc-4a0d-11ef-94a3-6ee10b77fd09'),
  {'times': 100, 'action': 'jump', 'animal': 'fox'},
  'jumped over the',
  array([ 1. , 10.8], dtype=float32),
  0.00016793422934946456]]

Predicates objects are defined by the name of the metadata key, an operator, and a value.

The supported operators are: ==, !=, <, <=, >, >=

The type of the values determines the type of comparison to perform. For example, passing in "Sam" (a string) will do a string comparison while a 10 (an int) will perform an integer comparison while a 10.0 (float) will do a float comparison. It is important to note that using a value of "10" will do a string comparison as well so it’s important to use the right type. Supported Python types are: str, int, and float. One more example with a string comparison:

vec.search([1.0, 9.0], limit=2, predicates=client.Predicates("action", "==", "jump"))

[[UUID('456dbbbc-4a0d-11ef-94a3-6ee10b77fd09'),
  {'times': 100, 'action': 'jump', 'animal': 'fox'},
  'jumped over the',
  array([ 1. , 10.8], dtype=float32),
  0.00016793422934946456]]

The real power of predicates is that they can also be combined using the & operator (for combining predicates with AND semantics) and |(for combining using OR semantic). So you can do:

vec.search([1.0, 9.0], limit=2, predicates=client.Predicates("action", "==", "jump") & client.Predicates("times", ">", 1))

[[UUID('456dbbbc-4a0d-11ef-94a3-6ee10b77fd09'),
  {'times': 100, 'action': 'jump', 'animal': 'fox'},
  'jumped over the',
  array([ 1. , 10.8], dtype=float32),
  0.00016793422934946456]]

Just for sanity, let’s show a case where no results are returned because or predicates:

vec.search([1.0, 9.0], limit=2, predicates=client.Predicates("action", "==", "jump") & client.Predicates("times", "==", 1))

[]

And one more example where we define the predicates as a variable and use grouping with parenthesis:

my_predicates = client.Predicates("action", "==", "jump") & (client.Predicates("times", "==", 1) | client.Predicates("times", ">", 1))
vec.search([1.0, 9.0], limit=2, predicates=my_predicates)

[[UUID('456dbbbc-4a0d-11ef-94a3-6ee10b77fd09'),
  {'times': 100, 'action': 'jump', 'animal': 'fox'},
  'jumped over the',
  array([ 1. , 10.8], dtype=float32),
  0.00016793422934946456]]

We also have some semantic sugar for combining many predicates with AND semantics. You can pass in multiple 3-tuples to Predicates:

vec.search([1.0, 9.0], limit=2, predicates=client.Predicates(("action", "==", "jump"), ("times", ">", 10)))

[[UUID('456dbbbc-4a0d-11ef-94a3-6ee10b77fd09'),
  {'times': 100, 'action': 'jump', 'animal': 'fox'},
  'jumped over the',
  array([ 1. , 10.8], dtype=float32),
  0.00016793422934946456]]

Filter your search by time

When using time-partitioning(see below). You can very efficiently filter your search by time. Time-partitioning makes a timestamp embedded as part of the UUID-based ID associated with an embedding. Let us first create a collection with time partitioning and insert some data (one item from January 2018 and another in January 2019):

tpvec = client.Sync(service_url, "time_partitioned_table", 2, time_partition_interval=timedelta(hours=6))
tpvec.create_tables()

specific_datetime = datetime(2018, 1, 1, 12, 0, 0)
tpvec.upsert([\
    (client.uuid_from_time(specific_datetime), {"animal":"fox", "action": "sit", "times":1}, "the brown fox", [1.0,1.3]),\
    (client.uuid_from_time(specific_datetime+timedelta(days=365)),  {"animal":"fox", "action": "jump", "times":100}, "jumped over the", [1.0,10.8]),\
])

Then, you can filter using the timestamps by specifing a uuid_time_filter:

tpvec.search([1.0, 9.0], limit=4, uuid_time_filter=client.UUIDTimeRange(specific_datetime, specific_datetime+timedelta(days=1)))

[[UUID('33c52800-ef15-11e7-8a12-ea51d07b6447'),
  {'times': 1, 'action': 'sit', 'animal': 'fox'},
  'the brown fox',
  array([1. , 1.3], dtype=float32),
  0.14489260377438218]]

A UUIDTimeRange can specify a start_date or end_date or both(as in the example above). Specifying only the start_date or end_date leaves the other end unconstrained.

tpvec.search([1.0, 9.0], limit=4, uuid_time_filter=client.UUIDTimeRange(start_date=specific_datetime))

[[UUID('ac8be800-0de6-11e9-a5fd-5a100e653c25'),
  {'times': 100, 'action': 'jump', 'animal': 'fox'},
  'jumped over the',
  array([ 1. , 10.8], dtype=float32),
  0.00016793422934946456],
 [UUID('33c52800-ef15-11e7-8a12-ea51d07b6447'),
  {'times': 1, 'action': 'sit', 'animal': 'fox'},
  'the brown fox',
  array([1. , 1.3], dtype=float32),
  0.14489260377438218]]

You have the option to define the inclusivity of the start and end dates with the start_inclusive and end_inclusive parameters. Setting start_inclusive to true results in comparisons using the >= operator, whereas setting it to false applies the > operator. By default, the start date is inclusive, while the end date is exclusive. One example:

tpvec.search([1.0, 9.0], limit=4, uuid_time_filter=client.UUIDTimeRange(start_date=specific_datetime, start_inclusive=False))

[[UUID('ac8be800-0de6-11e9-a5fd-5a100e653c25'),
  {'times': 100, 'action': 'jump', 'animal': 'fox'},
  'jumped over the',
  array([ 1. , 10.8], dtype=float32),
  0.00016793422934946456]]

Notice how the results are different when we use the start_inclusive=False option because the first row has the exact timestamp specified by start_date.

We’ve also made it easy to integrate time filters using the filter and predicates parameters described above using special reserved key names to make it appear that the timestamps are part of your metadata. We found this useful when integrating with other systems that just want to specify a set of filters (often these are “auto retriever” type systems). The reserved key names are __start_date and __end_date for filters and __uuid_timestamp for predicates. Some examples below:

tpvec.search([1.0, 9.0], limit=4, filter={ "__start_date": specific_datetime, "__end_date": specific_datetime+timedelta(days=1)})

[[UUID('33c52800-ef15-11e7-8a12-ea51d07b6447'),
  {'times': 1, 'action': 'sit', 'animal': 'fox'},
  'the brown fox',
  array([1. , 1.3], dtype=float32),
  0.14489260377438218]]

tpvec.search([1.0, 9.0], limit=4, 
             predicates=client.Predicates("__uuid_timestamp", ">=", specific_datetime) & client.Predicates("__uuid_timestamp", "<", specific_datetime+timedelta(days=1)))

[[UUID('33c52800-ef15-11e7-8a12-ea51d07b6447'),
  {'times': 1, 'action': 'sit', 'animal': 'fox'},
  'the brown fox',
  array([1. , 1.3], dtype=float32),
  0.14489260377438218]]

Indexing

Indexing speeds up queries over your data. By default, we set up indexes to query your data by the UUID and the metadata.

But to speed up similarity search based on the embeddings, you have to create additional indexes.

Note that if performing a query without an index, you will always get an exact result, but the query will be slow (it has to read all of the data you store for every query). With an index, your queries will be order-of-magnitude faster, but the results are approximate (because there are no known indexing techniques that are exact).

Nevertheless, there are excellent approximate algorithms. There are 3 different indexing algorithms available on the Timescale platform: Timescale Vector index, pgvector HNSW, and pgvector ivfflat. Below are the trade-offs between these algorithms:

Algorithm	Build speed	Query speed	Need to rebuild after updates
StreamingDiskANN	Fast	Fastest	No
pgvector hnsw	Slowest	Faster	No
pgvector ivfflat	Fastest	Slowest	Yes

You can see benchmarks on our blog.

We recommend using the Timescale Vector index for most use cases. This can be created with:

vec.create_embedding_index(client.DiskAnnIndex())

Indexes are created for a particular distance metric type. So it is important that the same distance metric is set on the client during index creation as it is during queries. See the distance type section below.

Each of these indexes has a set of build-time options for controlling the speed/accuracy trade-off when creating the index and an additional query-time option for controlling accuracy during a particular query. We have smart defaults for all of these options but will also describe the details below so that you can adjust these options manually.

StreamingDiskANN index

The StreamingDiskANN index from pgvectorscale is a graph-based algorithm that uses the DiskANN algorithm. You can read more about it on our blog announcing its release.

To create this index, run:

vec.create_embedding_index(client.DiskAnnIndex())

The above command will create the index using smart defaults. There are a number of parameters you could tune to adjust the accuracy/speed trade-off.

The parameters you can set at index build time are:

Parameter name	Description	Default value
storage_layout	memory_optimized which uses SBQ to compress vector data or plain which stores data uncompressed	memory_optimized
num_neighbors	Sets the maximum number of neighbors per node. Higher values increase accuracy but make the graph traversal slower.	50
search_list_size	This is the S parameter used in the greedy search algorithm used during construction. Higher values improve graph quality at the cost of slower index builds.	100
max_alpha	Is the alpha parameter in the algorithm. Higher values improve graph quality at the cost of slower index builds.	1.2
num_dimensions	The number of dimensions to index. By default, all dimensions are indexed. But you can also index less dimensions to make use of Matryoshka embeddings	0 (all dimensions)
num_bits_per_dimension	Number of bits used to encode each dimension when using SBQ	2 for less than 900 dimensions, 1 otherwise

To set these parameters, you could run:

vec.create_embedding_index(client.DiskAnnIndex(num_neighbors=50, search_list_size=100, max_alpha=1.0, storage_layout="memory_optimized", num_dimensions=0, num_bits_per_dimension=1))

You can also set a parameter to control the accuracy vs. query speed trade-off at query time. The parameter is set in the search() function using the query_params argment.

Parameter name	Description	Default value
search_list_size	The number of additional candidates considered during the graph search.	100
rescore	The number of elements rescored (0 to disable rescoring)	50

We suggest using the rescore parameter to fine-tune accuracy.

vec.search([1.0, 9.0], limit=4, query_params=client.DiskAnnIndexParams(rescore=400, search_list_size=10))

[[UUID('456dbbbc-4a0d-11ef-94a3-6ee10b77fd09'),
  {'times': 100, 'action': 'jump', 'animal': 'fox'},
  'jumped over the',
  array([ 1. , 10.8], dtype=float32),
  0.00016793422934946456],
 [UUID('456dbb6c-4a0d-11ef-94a3-6ee10b77fd09'),
  {'times': 1, 'action': 'sit', 'animal': 'fox'},
  'the brown fox',
  array([1. , 1.3], dtype=float32),
  0.14489260377438218]]

To drop the index, run:

vec.drop_embedding_index()

pgvector HNSW index

Pgvector provides a graph-based indexing algorithm based on the popular HNSW algorithm.

To create this index, run:

vec.create_embedding_index(client.HNSWIndex())

The above command will create the index using smart defaults. There are a number of parameters you could tune to adjust the accuracy/speed trade-off.

The parameters you can set at index build time are:

Parameter name	Description	Default value
m	Represents the maximum number of connections per layer. Think of these connections as edges created for each node during graph construction. Increasing m increases accuracy but also increases index build time and size.	16
ef_construction	Represents the size of the dynamic candidate list for constructing the graph. It influences the trade-off between index quality and construction speed. Increasing ef_construction enables more accurate search results at the expense of lengthier index build times.	64

To set these parameters, you could run:

vec.create_embedding_index(client.HNSWIndex(m=16, ef_construction=64))

You can also set a parameter to control the accuracy vs. query speed trade-off at query time. The parameter is set in the search() function using the query_params argument. You can set the ef_search(default: 40). This parameter specifies the size of the dynamic candidate list used during search. Higher values improve query accuracy while making the query slower.

You can specify this value during search as follows:

vec.search([1.0, 9.0], limit=4, query_params=client.HNSWIndexParams(ef_search=10))

[[UUID('456dbbbc-4a0d-11ef-94a3-6ee10b77fd09'),
  {'times': 100, 'action': 'jump', 'animal': 'fox'},
  'jumped over the',
  array([ 1. , 10.8], dtype=float32),
  0.00016793422934946456],
 [UUID('456dbb6c-4a0d-11ef-94a3-6ee10b77fd09'),
  {'times': 1, 'action': 'sit', 'animal': 'fox'},
  'the brown fox',
  array([1. , 1.3], dtype=float32),
  0.14489260377438218]]

To drop the index run:

vec.drop_embedding_index()

pgvector ivfflat index

Pgvector provides a clustering-based indexing algorithm. Our blog post describes how it works in detail. It provides the fastest index-build speed but the slowest query speeds of any indexing algorithm.

To create this index, run:

vec.create_embedding_index(client.IvfflatIndex())

Note: ivfflat should never be created on empty tables because it needs to cluster data, and that only happens when an index is first created, not when new rows are inserted or modified. Also, if your table undergoes a lot of modifications, you will need to rebuild this index occasionally to maintain good accuracy. See our blog post for details.

Pgvector ivfflat has a lists index parameter that is automatically set with a smart default based on the number of rows in your table. If you know that you’ll have a different table size, you can specify the number of records to use for calculating the lists parameter as follows:

vec.create_embedding_index(client.IvfflatIndex(num_records=1000000))

You can also set the lists parameter directly:

vec.create_embedding_index(client.IvfflatIndex(num_lists=100))

You can also set a parameter to control the accuracy vs. query speed trade-off at query time. The parameter is set in the search() function using the query_params argument. You can set the probes. This parameter specifies the number of clusters searched during a query. It is recommended to set this parameter to sqrt(lists) where lists is the num_list parameter used above during index creation. Higher values improve query accuracy while making the query slower.

You can specify this value during search as follows:

vec.search([1.0, 9.0], limit=4, query_params=client.IvfflatIndexParams(probes=10))

[[UUID('456dbbbc-4a0d-11ef-94a3-6ee10b77fd09'),
  {'times': 100, 'action': 'jump', 'animal': 'fox'},
  'jumped over the',
  array([ 1. , 10.8], dtype=float32),
  0.00016793422934946456],
 [UUID('456dbb6c-4a0d-11ef-94a3-6ee10b77fd09'),
  {'times': 1, 'action': 'sit', 'animal': 'fox'},
  'the brown fox',
  array([1. , 1.3], dtype=float32),
  0.14489260377438218]]

To drop the index, run:

vec.drop_embedding_index()

Time partitioning

In many use cases where you have many embeddings, time is an important component associated with the embeddings. For example, when embedding news stories, you often search by time as well as similarity (e.g., stories related to Bitcoin in the past week or stories about Clinton in November 2016).

Yet, traditionally, searching by two components “similarity” and “time” is challenging for Approximate Nearest Neighbor (ANN) indexes and makes the similarity-search index less effective.

One approach to solving this is partitioning the data by time and creating ANN indexes on each partition individually. Then, during search, you can:

• Step 1: filter our partitions that don’t match the time predicate.
• Step 2: perform the similarity search on all matching partitions.
• Step 3: combine all the results from each partition in step 2, rerank, and filter out results by time.

Step 1 makes the search a lot more efficient by filtering out whole swaths of data in one go.

Timescale-vector supports time partitioning using TimescaleDB’s hypertables. To use this feature, simply indicate the length of time for each partition when creating the client:

from datetime import timedelta
from datetime import datetime

vec = client.Async(service_url, "my_data_with_time_partition", 2, time_partition_interval=timedelta(hours=6))
await vec.create_tables()

Then, insert data where the IDs use UUIDs v1 and the time component of the UUID specifies the time of the embedding. For example, to create an embedding for the current time, simply do:

id = uuid.uuid1()
await vec.upsert([(id, {"key": "val"}, "the brown fox", [1.0, 1.2])])

To insert data for a specific time in the past, create the UUID using our uuid_from_time function

specific_datetime = datetime(2018, 8, 10, 15, 30, 0)
await vec.upsert([(client.uuid_from_time(specific_datetime), {"key": "val"}, "the brown fox", [1.0, 1.2])])

You can then query the data by specifying a uuid_time_filter in the search call:

rec = await vec.search([1.0, 2.0], limit=4, uuid_time_filter=client.UUIDTimeRange(specific_datetime-timedelta(days=7), specific_datetime+timedelta(days=7)))

Distance metrics

By default, we use cosine distance to measure how similarly an embedding is to a given query. In addition to cosine distance, we also support Euclidean/L2 distance. The distance type is set when creating the client using the distance_type parameter. For example, to use the Euclidean distance metric, you can create the client with:

vec  = client.Sync(service_url, "my_data", 2, distance_type="euclidean")

Valid values for distance_type are cosine and euclidean.

It is important to note that you should use consistent distance types on clients that create indexes and perform queries. That is because an index is only valid for one particular type of distance measure.

Please note the Timescale Vector index only supports cosine distance at this time.

LangChain integration

LangChain is a popular framework for development applications powered by LLMs. Timescale Vector has a native LangChain integration, enabling you to use Timescale Vector as a vectorstore and leverage all its capabilities in your applications built with LangChain.

Here are resources about using Timescale Vector with LangChain:

• Getting started with LangChain and Timescale Vector: You’ll learn how to use Timescale Vector for (1) semantic search, (2) time-based vector search, (3) self-querying, and (4) how to create indexes to speed up queries.
• PostgreSQL Self Querying: Learn how to use Timescale Vector with self-querying in LangChain.
• LangChain template: RAG with conversational retrieval: This template is used for conversational retrieval, which is one of the most popular LLM use-cases. It passes both a conversation history and retrieved documents into an LLM for synthesis.
• LangChain template: RAG with time-based search and self-query retrieval:This template shows how to use timescale-vector with the self-query retriver to perform hybrid search on similarity and time. This is useful any time your data has a strong time-based component.
• Learn more about Timescale Vector and LangChain

LlamaIndex integration

[LlamaIndex] is a popular data framework for connecting custom data sources to large language models (LLMs). Timescale Vector has a native LlamaIndex integration, enabling you to use Timescale Vector as a vectorstore and leverage all its capabilities in your applications built with LlamaIndex.

Here are resources about using Timescale Vector with LlamaIndex:

• Getting started with LlamaIndex and Timescale Vector: You’ll learn how to use Timescale Vector for (1) similarity search, (2) time-based vector search, (3) faster search with indexes, and (4) retrieval and query engine.
• Time-based retrieval: Learn how to power RAG applications with time-based retrieval.
• Llama Pack: Auto Retrieval with time-based search: This pack demonstrates performing auto-retrieval for hybrid search based on both similarity and time, using the timescale-vector (PostgreSQL) vectorstore.
• Learn more about Timescale Vector and LlamaIndex

PgVectorize

PgVectorize enables you to create vector embeddings from any data that you already have stored in PostgreSQL. You can get more background information in our blog post announcing this feature, as well as a “how we built in” post going into the details of the design.

To create vector embeddings, simply attach PgVectorize to any PostgreSQL table, and it will automatically sync that table’s data with a set of embeddings stored in Timescale Vector. For example, let’s say you have a blog table defined in the following way:

import psycopg2
from langchain.docstore.document import Document
from langchain.text_splitter import CharacterTextSplitter
from timescale_vector import client, pgvectorizer
from langchain_openai import OpenAIEmbeddings
from langchain_community.vectorstores.timescalevector import TimescaleVector
from datetime import timedelta

with psycopg2.connect(service_url) as conn:
    with conn.cursor() as cursor:
        cursor.execute('''
        CREATE TABLE IF NOT EXISTS blog (
            id              SERIAL PRIMARY KEY NOT NULL,
            title           TEXT NOT NULL,
            author          TEXT NOT NULL,
            contents        TEXT NOT NULL,
            category        TEXT NOT NULL,
            published_time  TIMESTAMPTZ NULL --NULL if not yet published
        );
        ''')

You can insert some data as follows:

with psycopg2.connect(service_url) as conn:
    with conn.cursor() as cursor:
        cursor.execute('''
            INSERT INTO blog (title, author, contents, category, published_time) VALUES ('First Post', 'Matvey Arye', 'some super interesting content about cats.', 'AI', '2021-01-01');
        ''')

Now, say you want to embed these blogs in Timescale Vector. First, you need to define an embed_and_write function that takes a set of blog posts, creates the embeddings, and writes them into TimescaleVector. For example, if using LangChain, it could look something like the following.

def get_document(blog):
    text_splitter = CharacterTextSplitter(
        chunk_size=1000,
        chunk_overlap=200,
    )
    docs = []
    for chunk in text_splitter.split_text(blog['contents']):
        content = f"Author {blog['author']}, title: {blog['title']}, contents:{chunk}"
        metadata = {
            "id": str(client.uuid_from_time(blog['published_time'])),
            "blog_id": blog['id'], 
            "author": blog['author'], 
            "category": blog['category'],
            "published_time": blog['published_time'].isoformat(),
        }
        docs.append(Document(page_content=content, metadata=metadata))
    return docs

def embed_and_write(blog_instances, vectorizer):
    embedding = OpenAIEmbeddings()
    vector_store = TimescaleVector(
        collection_name="blog_embedding",
        service_url=service_url,
        embedding=embedding,
        time_partition_interval=timedelta(days=30),
    )

    # delete old embeddings for all ids in the work queue. locked_id is a special column that is set to the primary key of the table being
    # embedded. For items that are deleted, it is the only key that is set.
    metadata_for_delete = [{"blog_id": blog['locked_id']} for blog in blog_instances]
    vector_store.delete_by_metadata(metadata_for_delete)

    documents = []
    for blog in blog_instances:
        # skip blogs that are not published yet, or are deleted (in which case it will be NULL)
        if blog['published_time'] != None:
            documents.extend(get_document(blog))

    if len(documents) == 0:
        return
    
    texts = [d.page_content for d in documents]
    metadatas = [d.metadata for d in documents]
    ids = [d.metadata["id"] for d in documents]
    vector_store.add_texts(texts, metadatas, ids)

Then, all you have to do is run the following code in a scheduled job (cron job, Lambda job, etc):

# this job should be run on a schedule
vectorizer = pgvectorizer.Vectorize(service_url, 'blog')
while vectorizer.process(embed_and_write) > 0:
    pass

Every time that job runs, it will sync the table with your embeddings. It will sync all inserts, updates, and deletes to an embeddings table called blog_embedding.

Now, you can simply search the embeddings as follows (again, using LangChain in the example):

embedding = OpenAIEmbeddings()
vector_store = TimescaleVector(
    collection_name="blog_embedding",
    service_url=service_url,
    embedding=embedding,
    time_partition_interval=timedelta(days=30),
)

res = vector_store.similarity_search_with_score("Blogs about cats")
res

[(Document(metadata={'id': '334e4800-4bee-11eb-a52a-57b3c4a96ccb', 'author': 'Matvey Arye', 'blog_id': 1, 'category': 'AI', 'published_time': '2021-01-01T00:00:00-05:00'}, page_content='Author Matvey Arye, title: First Post, contents:some super interesting content about cats.'),
  0.12680577303752072)]

Development

This project is developed with nbdev. Please see that website for the development process.