ChromaDB

In this section, we present how to connect ChromaDB to MindsDB.

ChromaDB is the open-source embedding database. Chroma makes it easy to build LLM apps by making knowledge, facts, and skills pluggable for LLMs.

​

Prerequisites

Before proceeding, ensure the following prerequisites are met:

1. Install MindsDB locally via Docker or Docker Desktop.
2. To connect ChromaDB to MindsDB, install the required dependencies following this instruction.
3. Install or ensure access to ChromaDB.

​

Connection

This handler is implemented using the chromadb Python library.

The required arguments to establish a connection are:

• host: the host name or IP address of the ChromaDB instance.
• port: the TCP/IP port of the ChromaDB instance.

OR

• persist_directory: the directory to use for persisting data.

To connect to a remote ChromaDB instance, the following CREATE DATABASE can be used:

CREATE DATABASE chromadb_datasource
WITH ENGINE = 'chromadb'
PARAMETERS = {
    "host": "YOUR_HOST",
    "port": YOUR_PORT
}

Alternateively, to connect to an in-memory ChromaDB instance, the following CREATE DATABASE can be used:

CREATE DATABASE chromadb_datasource
WITH ENGINE = "chromadb",
PARAMETERS = {
    "persist_directory": "YOUR_PERSIST_DIRECTORY"
}

​

Usage

Now, you can use the established connection to create a collection (or table in the context of MindsDB) in ChromaDB and insert data into it:

CREATE TABLE chromadb_datasource.test_embeddings (
    SELECT embeddings,'{"source": "fda"}' as metadata
    FROM mysql_datasource.test_embeddings
);

You can query your collection (table) as shown below:

SELECT * 
FROM chromadb_datasource.test_embeddings;

To filter the data in your collection (table) by metadata, you can use the following query:

SELECT * 
FROM chromadb_datasource.test_embeddings
WHERE `metadata.source` = "fda";

To conduct a similarity search, the following query can be used:

SELECT *
FROM chromadb_datasource.test_embeddings
WHERE search_vector = (
    SELECT embeddings
    FROM mysql_datasource.test_embeddings
    LIMIT 1
);