WeaviateStore

Weaviate is an open source vector database that stores both objects and vectors, allowing for combining vector search with structured filtering. LangChain connects to Weaviate via the weaviate-client package, the official Typescript client for Weaviate.

This guide provides a quick overview for getting started with Weaviate vector stores. For detailed documentation of all WeaviateStore features and configurations head to the API reference.

Overview

Integration details

Class	Package	PY support	Package latest
WeaviateStore	@langchain/weaviate	✅

Setup

To use Weaviate vector stores, you’ll need to set up a Weaviate instance and install the @langchain/weaviate integration package. You should also install the weaviate-client package to initialize a client to connect to your instance with, and the uuid package if you want to assign indexed documents ids.

This guide will also use OpenAI embeddings, which require you to install the @langchain/openai integration package. You can also use other supported embeddings models if you wish.

tip

See this section for general instructions on installing integration packages.

npm

npm i @langchain/weaviate @langchain/core weaviate-client uuid @langchain/openai

yarn

yarn add @langchain/weaviate @langchain/core weaviate-client uuid @langchain/openai

pnpm

pnpm add @langchain/weaviate @langchain/core weaviate-client uuid @langchain/openai

You’ll need to run Weaviate either locally or on a server. See the Weaviate documentation for more information.

Credentials

Once you’ve set up your instance, set the following environment variables:

// If running locally, include port e.g. "localhost:8080"
process.env.WEAVIATE_URL = "YOUR_WEAVIATE_URL";
// Optional, for cloud deployments
process.env.WEAVIATE_API_KEY = "YOUR_API_KEY";

If you are using OpenAI embeddings for this guide, you’ll need to set your OpenAI key as well:

process.env.OPENAI_API_KEY = "YOUR_API_KEY";

If you want to get automated tracing of your model calls you can also set your LangSmith API key by uncommenting below:

// process.env.LANGSMITH_TRACING="true"
// process.env.LANGSMITH_API_KEY="your-api-key"

Instantiation

Connect a weaviate client

In most cases, you should use one of the connection helper functions to connect to your Weaviate instance:

• connectToWeaviateCloud
• connectToLocal
• connectToCustom

import { WeaviateStore } from "@langchain/weaviate";
import { OpenAIEmbeddings } from "@langchain/openai";
import weaviate from "weaviate-client";

const embeddings = new OpenAIEmbeddings({
  model: "text-embedding-3-small",
});

const weaviateClient = weaviate.connectToWeaviateCloud({
  clusterURL: process.env.WEAVIATE_URL!,
  options: {
    authCredentials: new weaviate.ApiKey(process.env.WEAVIATE_API_KEY || ""),
    headers: {
      "X-OpenAI-Api-Key": process.env.OPENAI_API_KEY || "",
      "X-Cohere-Api-Key": process.env.COHERE_API_KEY || "",
    },
  },
});

Initiate the vectorStore

To create a collection, specify at least the collection name. If you don’t specify any properties, auto-schema creates them.

const vectorStore = new WeaviateStore(embeddings, {
  client: weaviateClient,
  // Must start with a capital letter
  indexName: "Langchainjs_test",
});

To use Weaviate’s named vectors, vectorizers, reranker, generative-models etc., use the schema property when enabling the vector store. The collection name and other properties in schema will take precedence when creating the vector store.

const vectorStore = new WeaviateStore(embeddings, {
  client: weaviateClient,
  schema: {
    name: "Langchainjs_test",
    description: "A simple dataset",
    properties: [
      {
        name: "title",
        dataType: dataType.TEXT,
      },
      {
        name: "foo",
        dataType: dataType.TEXT,
      },
    ],
    vectorizers: [
      vectorizer.text2VecOpenAI({
        name: "title",
        sourceProperties: ["title"], // (Optional) Set the source property(ies)
        // vectorIndexConfig: configure.vectorIndex.hnsw()   // (Optional) Set the vector index configuration
      }),
    ],
    generative: weaviate.configure.generative.openAI(),
    reranker: weaviate.configure.reranker.cohere(),
  },
});

Manage vector store

Add items to vector store

Note: If you want to associate ids with your indexed documents, they must be UUIDs.

import type { Document } from "@langchain/core/documents";
import { v4 as uuidv4 } from "uuid";

const document1: Document = {
  pageContent: "The powerhouse of the cell is the mitochondria",
  metadata: { source: "https://example.com" },
};

const document2: Document = {
  pageContent: "Buildings are made out of brick",
  metadata: { source: "https://example.com" },
};

const document3: Document = {
  pageContent: "Mitochondria are made out of lipids",
  metadata: { source: "https://example.com" },
};

const document4: Document = {
  pageContent: "The 2024 Olympics are in Paris",
  metadata: { source: "https://example.com" },
};

const documents = [document1, document2, document3, document4];
const uuids = [uuidv4(), uuidv4(), uuidv4(), uuidv4()];

await vectorStore.addDocuments(documents, { ids: uuids });

[
  '610f9b92-9bee-473f-a4db-8f2ca6e3442d',
  '995160fa-441e-41a0-b476-cf3785518a0d',
  '0cdbe6d4-0df8-4f99-9b67-184009fee9a2',
  '18a8211c-0649-467b-a7c5-50ebb4b9ca9d'
]

Delete items from vector store

You can delete by id as by passing a filter param:

await vectorStore.delete({ ids: [uuids[3]] });

Query vector store

Once your vector store has been created and the relevant documents have been added you will most likely wish to query it during the running of your chain or agent. In weaviate’s v3, the client interacts with collections as the primary way to work with objects in the database. The collection object can be re-used throughout the codebase ### Query directly

Performing a simple similarity search can be done as follows. The Filter helper class makes it easier to use filters with conditions. The v3 client streamlines how you use Filter so your code is cleaner and more concise.

See this page for more on Weaviate filter syntax.

import { Filters } from "weaviate-client";

const collection = client.collections.use("Langchainjs_test");

const filter = Filters.and(
  collection.filter.byProperty("source").equal("https://example.com")
);

const similaritySearchResults = await vectorStore.similaritySearch(
  "biology",
  2,
  filter
);

for (const doc of similaritySearchResults) {
  console.log(`* ${doc.pageContent} [${JSON.stringify(doc.metadata, null)}]`);
}

* The powerhouse of the cell is the mitochondria [{"source":"https://example.com"}]
* Mitochondria are made out of lipids [{"source":"https://example.com"}]

If you want to execute a similarity search and receive the corresponding scores you can run:

const similaritySearchWithScoreResults =
  await vectorStore.similaritySearchWithScore("biology", 2, filter);

for (const [doc, score] of similaritySearchWithScoreResults) {
  console.log(
    `* [SIM=${score.toFixed(3)}] ${doc.pageContent} [${JSON.stringify(
      doc.metadata
    )}]`
  );
}

* [SIM=0.835] The powerhouse of the cell is the mitochondria [{"source":"https://example.com"}]
* [SIM=0.852] Mitochondria are made out of lipids [{"source":"https://example.com"}]

Hybrid Search

In Weaviate, Hybrid search combines the results of a vector search and a keyword (BM25F) search by fusing the two result sets. To change the relative weights of the keyword and vector components, set the alpha value in your query.

Check docs for the full list of hybrid search options.

const results = await vectorStore.hybridSearch("biology", {
  limit: 1,
  alpha: 0.25,
  targetVector: ["title"],
  rerank: {
    property: "title",
    query: "greeting",
  },
});

Retrieval Augmented Generation (RAG)

Retrieval Augmented Generation (RAG) combines information retrieval with generative AI models.

In Weaviate, a RAG query consists of two parts: a search query, and a prompt for the model. Weaviate first performs the search, then passes both the search results and your prompt to a generative AI model before returning the generated response. * @param query The query to search for. * @param options available options for performing the hybrid search * @param generate available options for the generation. Check docs for complete list

const results = await vectorStore.generate(
  "hello world",
  {
    singlePrompt: {
      prompt: "Translate this into German: {title}",
    },
    config: generativeParameters.openAI({
      model: "gpt-3.5-turbo",
    }),
  },
  {
    limit: 2,
    targetVector: ["title"],
  }
);

Query by turning into retriever

You can also transform the vector store into a retriever for easier usage in your chains.

const retriever = vectorStore.asRetriever({
  // Optional filter
  filter: filter,
  k: 2,
});
await retriever.invoke("biology");

[
  Document {
    pageContent: 'The powerhouse of the cell is the mitochondria',
    metadata: { source: 'https://example.com' },
    id: undefined
  },
  Document {
    pageContent: 'Mitochondria are made out of lipids',
    metadata: { source: 'https://example.com' },
    id: undefined
  }
]

Usage for retrieval-augmented generation

For guides on how to use this vector store for retrieval-augmented generation (RAG), see the following sections:

• Tutorials: working with external knowledge.
• How-to: Question and answer with RAG
• Retrieval conceptual docs

API reference

For detailed documentation of all WeaviateStore features and configurations head to the API reference.

Related

• Vector store conceptual guide
• Vector store how-to guides