pgvector-php

pgvector support for PHP

Getting Started

Follow the instructions for your database library:

• Laravel
• PHP

Or check out some examples:

• Embeddings with OpenAI
• Binary embeddings with Cohere
• Morgan fingerprints with RDKit
• Recommendations with Disco
• Bulk loading with COPY

Laravel

Install the package

composer require pgvector/pgvector

Enable the extension

php artisan vendor:publish --tag="pgvector-migrations"
php artisan migrate

You can now use the vector type in future migrations

Schema::create('items', function (Blueprint $table) {
    $table->vector('embedding', 3);
});

Update your model

use Pgvector\Laravel\Vector;

class Item extends Model
{
    use HasNeighbors;

    protected $casts = ['embedding' => Vector::class];
}

Insert a vector

$item = new Item();
$item->embedding = [1, 2, 3];
$item->save();

Get the nearest neighbors to a record

use Pgvector\Laravel\Distance;

$neighbors = $item->nearestNeighbors('embedding', Distance::L2)->take(5)->get();

Also supports InnerProduct, Cosine, L1, Hamming, and Jaccard distance

Get the nearest neighbors to a vector

$neighbors = Item::query()->nearestNeighbors('embedding', [1, 2, 3], Distance::L2)->take(5)->get();

Get the distances

$neighbors->pluck('neighbor_distance');

Add an approximate index in a migration

public function up()
{
    DB::statement('CREATE INDEX my_index ON items USING hnsw (embedding vector_l2_ops)');
    // or
    DB::statement('CREATE INDEX my_index ON items USING ivfflat (embedding vector_l2_ops) WITH (lists = 100)');
}

public function down()
{
    DB::statement('DROP INDEX my_index');
}

Use vector_ip_ops for inner product and vector_cosine_ops for cosine distance

PHP

Enable the extension

pg_query($db, 'CREATE EXTENSION IF NOT EXISTS vector');

Create a table

pg_query($db, 'CREATE TABLE items (embedding vector(3))');

Insert a vector

use Pgvector\Vector;

$embedding = new Vector([1, 2, 3]);
pg_query_params($db, 'INSERT INTO items (embedding) VALUES ($1)', [$embedding]);

Get the nearest neighbors to a vector

$embedding = new Vector([1, 2, 3]);
$result = pg_query_params($db, 'SELECT * FROM items ORDER BY embedding <-> $1 LIMIT 5', [$embedding]);

Add an approximate index

pg_query($db, 'CREATE INDEX ON items USING hnsw (embedding vector_l2_ops)');
// or
pg_query($db, 'CREATE INDEX ON items USING ivfflat (embedding vector_l2_ops) WITH (lists = 100)');

See a full example

Upgrading

0.1.4

The package name was changed from ankane/pgvector to pgvector/pgvector. Update it in composer.json to remove the message.

History

View the changelog

Contributing

Everyone is encouraged to help improve this project. Here are a few ways you can help:

• Report bugs
• Fix bugs and submit pull requests
• Write, clarify, or fix documentation
• Suggest or add new features

To get started with development:

git clone https://github.com/pgvector/pgvector-php.git
cd pgvector-php
composer install
createdb pgvector_php_test
composer test

To run an example:

cd examples/loading
composer install
createdb pgvector_example
php example.php