Download the PHP package allanpichardo/mysql-vector without Composer
On this page you can find all versions of the php package allanpichardo/mysql-vector. It is possible to download/install these versions without Composer. Possible dependencies are resolved automatically.
Download allanpichardo/mysql-vector
More information about allanpichardo/mysql-vector
Files in allanpichardo/mysql-vector
Package mysql-vector
Short Description Perform vector operations natively on MySQL
License MIT
FAQ
Example:
In general, it is recommended to use always a project to download your libraries. In an application normally there is more than one library needed.
- Some hosting areas are not accessible by a terminal or SSH. Then it is not possible to use Composer.
- To use Composer is sometimes complicated. Especially for beginners.
- Composer needs much resources. Sometimes they are not available on a simple webspace.
- If you are using private repositories you don't need to share your credentials. You can set up everything on our site and then you provide a simple download link to your team member.
- Simplify your Composer build process. Use our own command line tool to download the vendor folder as binary. This makes your build process faster and you don't need to expose your credentials for private repositories.
Informations about the package mysql-vector
A Library for MySQL Vector Operations and Text Embeddings
Overview
The VectorTable class is a PHP implementation designed to facilitate the storage, retrieval, and comparison of high-dimensional vectors in a MySQL database. This class utilizes MySQL JSON data types and a custom cosine similarity function (COSIM) to perform vector comparisons efficiently.
Search Performance
Vectors are binary quantized upon insertion into the database to optimize search speed and reranked to improve accuracy. However, this library is only suitable for small datasets (less than 1,000,000 vectors). For large datasets, it is recommended that you use a dedicated vector database such as Qdrant.
| Search Benchmarks (384-dimensional vectors): Vectors | Time (seconds) |
|---|---|
| 100 | 0.02 |
| 1000 | 0.02 |
| 10000 | 0.03 |
| 100000 | 0.06 |
| 1000000 | 0.48 |
Features
- Store vectors in a MySQL database using JSON data types.
- Calculate cosine similarity between vectors using a custom MySQL function.
- Normalize vectors and handle vector operations such as insertion, deletion, and searching.
- Support for vector quantization for optimized search operations.
- Native PHP support for generating for text embeddings using the BGE embedding model.
Requirements
- PHP 8.0 or higher.
- MySQL 5.7 or higher with support for JSON data types and stored functions.
- A MySQLi extension for PHP.
Installation
- Ensure that PHP and MySQL are installed and properly configured on your system.
- Install the library using Composer.
Usage
Initializing the Vector Table
Import the VectorTable class and create a new instance using the MySQLi connection, table name, and vector dimension.
Setting Up the Vector Table in MySQL
The initialize method will create the vector table in MySQL if it does not already exist. This method will also create the COSIM function in MySQL if it does not already exist.
Inserting and Managing Vectors
Calculating Cosine Similarity
Searching for Similar Vectors
Perform a search for vectors similar to a given vector using the cosine similarity criteria. The topN parameter specifies the maximum number of similar vectors to return.
Text Embeddings
The Embedder class calculates 384-dimensional text embeddings using the BGE embedding model. The first time you instanciate the Embedder class, the ONNX runtime will be installed automatically.
The maximum length of the input text is 512 characters. The Embedder class will automatically truncate the input text to 512 characters if it is longer than 512 characters.
Contributions
Contributions to this project are welcome. Please ensure that your code adheres to the existing coding standards and includes appropriate tests.
Development
This project uses DDEV, a Docker-based development environment. To get started, install DDEV and run the following commands:
To run the tests, use the following command:
License
MIT License
All versions of mysql-vector with dependencies
ext-mysqli Version *
bdelespierre/php-kmeans Version ^2.2
ext-mbstring Version *
ext-intl Version *
symfony/polyfill-intl-normalizer Version ^1.28
symfony/polyfill-mbstring Version ^1.28
ext-ctype Version *
symfony/polyfill-ctype Version ^1.28
ext-iconv Version *
symfony/polyfill-iconv Version ^1.28
ankane/onnxruntime Version ^0.2.1