Chroma

Chroma: The AI-native, open-source embedding database. We offer a fully managed, scalable deployment on AWS and Azure, complete with a Vector Management UI.

Vector Management.

Self host Prerequisite

1. Download & Install Docker and Git

2. Clone Chroma's repository with your terminal

git clone https://github.com/chroma-core/chroma.git

3. Change directory path to your cloned Chroma

cd chroma

4. Run docker compose to build up Chroma image and container

docker-compose up -d --build

5. If success, you will be able to see the docker images spun up:

Setup

Input	Description	Default
Document	Can be connected with nodes from Document Loader
Embeddings	Can be connected with nodes from Embeddings
Collection Name	Chroma collection name. Refer to here for naming convention
Chroma URL	Specify the URL of your chroma instance	http://localhost:8000

• Chroma Upsert Document - Upsert document chunks with embeddings into Chroma

• Chroma Load Existing Index - load existing collection from Chroma

Additional

1. On the Chroma URL, specify http://host.docker.internal:8000

Resources

• LangChain JS Chroma

• Chroma Getting Started