Together AI Embedding Function for Chroma

This package provides a Together AI embedding provider for Chroma.

Installation

npm install @chroma-core/together-ai

Usage

import { ChromaClient } from 'chromadb';
import { TogetherAIEmbeddingFunction } from '@chroma-core/together-ai';

// Initialize the embedder
const embedder = new TogetherAIEmbeddingFunction({
  apiKey: 'your-api-key', // Or set TOGETHER_API_KEY env var
  modelName: 'togethercomputer/m2-bert-80M-8k-retrieval',
});

// Create a new ChromaClient
const client = new ChromaClient({
  path: 'http://localhost:8000',
});

// Create a collection with the embedder
const collection = await client.createCollection({
  name: 'my-collection',
  embeddingFunction: embedder,
});

// Add documents
await collection.add({
  ids: ["1", "2", "3"],
  documents: ["Document 1", "Document 2", "Document 3"],
});

// Query documents
const results = await collection.query({
  queryTexts: ["Sample query"],
  nResults: 2,
});

Configuration

Set your Together AI API key as an environment variable:

export TOGETHER_API_KEY=your-api-key

Get your API key from Together AI.

Configuration Options

• apiKey: Your Together AI API key (or set via environment variable)
• apiKeyEnvVar: Environment variable name for API key (default: TOGETHER_API_KEY)
• modelName: Model to use for embeddings (required)

Supported Models

Popular embedding models available through Together AI:

• togethercomputer/m2-bert-80M-8k-retrieval
• togethercomputer/m2-bert-80M-32k-retrieval
• WhereIsAI/UAE-Large-V1
• BAAI/bge-large-en-v1.5
• BAAI/bge-base-en-v1.5

Check the Together AI documentation for the complete list of available embedding models and their specifications.

Features

• High Performance: Optimized inference infrastructure for fast embeddings
• Multiple Models: Access to various state-of-the-art embedding models
• Scalable: Built for production workloads with reliable uptime