ChromaDB

The AI-native embedding database

Exposing Publicly

This ChromaDB is by default exposed under the Private Network. To use it under the public network you'll need to expose it under the public network and then replace --host :: with --host 0.0.0.0 in the service's Start Command.

Documentation

All the documentation is covered here!

Examples

Javascript

// install the JS client and the (optional) default embedding function
npm install --save chromadb chromadb-default-embed
// set "type": "module" in your package.json

import { ChromaClient } from 'chromadb'

async function chromaExample() {
  const client = new ChromaClient();
  const collection = await client.createCollection({name: "sample_collection"});

  await collection.add({
    documents: ["This is a document", "This is another document"], // we embed for you, or bring your own
    metadatas: [{ source: "my_source" }, { source: "my_source" }], // filter on arbitrary metadata!
    ids: ["id1", "id2"] // must be unique for each doc 
  });

  const results = await collection.query({
    queryTexts: ["This is a query document"],
    nResults: 2,
    // where: {"metadata_field": "is_equal_to_this"}, // optional filter
    // whereDocument: {"$contains":"search_string"} // optional filter
  });
}

chromaExample();

Python

# python can also run in-memory with no server running: chromadb.PersistentClient()

import chromadb
client = chromadb.HttpClient()
collection = client.create_collection("sample_collection")

# Add docs to the collection. Can also update and delete. Row-based API coming soon!
collection.add(
    documents=["This is document1", "This is document2"], # we embed for you, or bring your own
    metadatas=[{"source": "notion"}, {"source": "google-docs"}], # filter on arbitrary metadata!
    ids=["doc1", "doc2"], # must be unique for each doc 
)

results = collection.query(
    query_texts=["This is a query document"],
    n_results=2,
    # where={"metadata_field": "is_equal_to_this"}, # optional filter
    # where_document={"$contains":"search_string"}  # optional filter
)