Embedin - Empower AI with persistent memory

Embedin is an open-source vector database and efficient library that seamlessly converts popular databases like MySQL, PostgreSQL, and MS SQL Server into vector databases with zero effort.

Embedin is an ideal solution for AI applications like natural language processing, image recognition, and recommendation systems, offering fast indexing and retrieval. Its simple API and query language ensure ease of use and seamless integration.

Minimum Requirements

Python 3.7 or higher.

Installation

pip install embedin

Quick Start

Store embeddings in memory

from embedin import Embedin

client = Embedin(collection_name="test_collection", texts=["This is a test", "Hello world!"])
result = client.query("These are tests", top_k=1)  # Query the most similar text from the collection
print(result)

Store embeddings in SQLite

from embedin import Embedin

url = 'sqlite:///test.db'
client = Embedin(collection_name="test_collection", texts=["This is a test", "Hello world!"], url=url)
result = client.query("These are tests", top_k=1)

Store embeddings in PostgreSQL

Start PostgreSQL Docker container

cd docker
docker-compose up embedin-postgres

example

import os

from embedin import Embedin

url = os.getenv('EMBEDIN_POSGRES_URL', "postgresql+psycopg2://embedin:embedin@localhost/embedin_db")
client = Embedin(collection_name="test_collection", texts=["This is a test", "Hello world!"], url=url)
result = client.query("These are tests", top_k=1)

Store embeddings in MySQL

Start MySQL Docker container

cd docker
docker-compose up embedin-mysql

example

import os

from embedin import Embedin

url = os.getenv('EMBEDIN_MYSQL_URL', "mysql+pymysql://embedin:embedin@localhost/embedin_db")
client = Embedin(collection_name="test_collection", texts=["This is a test", "Hello world!"], url=url)
result = client.query("These are tests", top_k=1)

Store embeddings in SQL Server

Start SQL Server Docker container

cd docker
docker-compose up embedin-mssql

example

import os

from embedin import Embedin

url = os.getenv('EMBEDIN_MSSQL_URL', "mssql+pymssql://sa:StrongPassword123@localhost/tempdb")
client = Embedin(collection_name="test_collection", url=url)
client.add_data(texts=["This is a test"], meta_data=[{"source": "abc4"}])
result = client.query("These are tests", top_k=1)

More References

• API Reference

Contribute

Please refer Contributors Guide

TODO

• unit test fail on faiss index (segmentation fault)