Skip to main content

🧬 Embeddings

Select a language

Embeddings are the A.I-native way to represent any kind of data, making them the perfect fit for working with all kinds of A.I-powered tools and algorithms. They can represent text, images, and soon audio and video. There are many options for creating embeddings, whether locally using an installed library, or by calling an API.

Chroma provides lightweight wrappers around popular embedding providers, making it easy to use them in your apps. You can set an embedding function when you create a Chroma collection, which will be used automatically, or you can call them directly yourself.

OpenAI​

Chroma provides a convenient wrapper around OpenAI's embedding API. This embedding function runs remotely on OpenAI's servers, and requires an API key. You can get an API key by signing up for an account at OpenAI.

const {OpenAIEmbeddingFunction} = require('chromadb');
const embedder = new OpenAIEmbeddingFunction({openai_api_key: "apiKey"})

// use directly 
const embeddings = embedder.generate(["document1","document2"])

// pass documents to query for .add and .query
const collection = await client.createCollection({name: "name", embeddingFunction: embedder})
const collection = await client.getCollection({name: "name", embeddingFunction: embedder})

You can pass in an optional model_name argument, which lets you choose which OpenAI embeddings model to use. By default, Chroma uses text-embedding-ada-002. You can see a list of all available models here.

Cohere​

Chroma also provides a convenient wrapper around Cohere's embedding API. This embedding function runs remotely on Cohere’s servers, and requires an API key. You can get an API key by signing up for an account at Cohere.

const {CohereEmbeddingFunction} = require('chromadb');
const embedder = new CohereEmbeddingFunction("apiKey")

// use directly 
const embeddings = embedder.generate(["document1","document2"])

// pass documents to query for .add and .query
const collection = await client.createCollection({name: "name", embeddingFunction: embedder})
const collectionGet = await client.getCollection({name:"name", embeddingFunction: embedder})

You can pass in an optional model_name argument, which lets you choose which Cohere embeddings model to use. By default, Chroma uses large model. You can see the available models under Get embeddings section here.

Multilingual model example​

const {CohereEmbeddingFunction} = require('chromadb');
const embedder = new CohereEmbeddingFunction("apiKey")

multilingual_texts  = [ 'Hello from Cohere!', 'مرحبًا من كوهير!', 
        'Hallo von Cohere!', 'Bonjour de Cohere!', 
        '¡Hola desde Cohere!', 'Olá do Cohere!', 
        'Ciao da Cohere!', '您好，来自 Cohere！',
        'कोहेरे से नमस्ते!'  ]

const embeddings = embedder.generate(multilingual_texts)

For more information on multilingual model you can read here.

Instructor models​

The instructor-embeddings library is another option, especially when running on a machine with a cuda-capable GPU. They are a good local alternative to OpenAI (see the Massive Text Embedding Benchmark rankings). The embedding function requires the InstructorEmbedding package. To install it, run pip install InstructorEmbedding.

There are three models available. The default is hkunlp/instructor-base, and for better performance you can use hkunlp/instructor-large or hkunlp/instructor-xl. You can also specify whether to use cpu (default) or cuda. For example:

#uses base model and cpu
ef = embedding_functions.InstructorEmbeddingFunction()

or

ef = embedding_functions.InstructorEmbeddingFunction(
model_name="hkunlp/instructor-xl", device="cuda")

Keep in mind that the large and xl models are 1.5GB and 5GB respectively, and are best suited to running on a GPU.

Google PaLM API models​

Google PaLM APIs are currently in private preview, but if you are part of this preview, you can use them with Chroma via the GooglePalmEmbeddingFunction.

To use the PaLM embedding API, you must have google.generativeai Python package installed and have the API key. To use:

palm_embedding = embedding_functions.GooglePalmEmbeddingFunction(
    api_key=api_key, model=model_name)

HuggingFace​

Chroma also provides a convenient wrapper around HuggingFace's embedding API. This embedding function runs remotely on HuggingFace's servers, and requires an API key. You can get an API key by signing up for an account at HuggingFace.

Custom Embedding Functions​

You can create your own embedding function to use with Chroma, it just needs to implement the EmbeddingFunction protocol. The .generate method in a class is strictly all you need.

class MyEmbeddingFunction {
  private api_key: string;

  constructor(api_key: string) {
    this.api_key = api_key;
  }

  public async generate(texts: string[]): Promise<number[][]> {
    // do things to turn texts into embeddings with an api_key perhaps
    return embeddings;
  }
}

We welcome pull requests to add new Embedding Functions to the community.