Jina Embedding Function for Chroma

This package provides a Jina AI embedding provider for Chroma.

Installation

npm install @chroma-core/jina

Usage

import { ChromaClient } from 'chromadb';
import { JinaEmbeddingFunction } from '@chroma-core/jina';

// Initialize the embedder
const embedder = new JinaEmbeddingFunction({
  apiKey: 'your-api-key', // Or set JINA_API_KEY env var
  modelName: 'jina-embeddings-v2-base-en',
  // Optional configuration
  task: 'retrieval.passage',
  dimensions: 768,
  lateChunking: false,
  truncate: true,
  normalized: true,
  embeddingType: 'float'
});

// 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 Jina API key as an environment variable:

export JINA_API_KEY=your-api-key

Get your API key from Jina AI.

Configuration Options

• apiKey: Your Jina API key (or set via environment variable)
• apiKeyEnvVar: Environment variable name for API key (default: JINA_API_KEY)
• modelName: Model to use for embeddings
• task: Task type (e.g., retrieval.passage, retrieval.query)
• dimensions: Output embedding dimensions
• lateChunking: Enable late chunking for better long document handling
• truncate: Whether to truncate input text that exceeds model limits
• normalized: Whether to return normalized embeddings
• embeddingType: Output type (float, base64, binary)

Supported Models

• jina-embeddings-v2-base-en
• jina-embeddings-v2-small-en
• jina-clip-v1

Check the Jina AI documentation for the complete list of available models and their capabilities.