Introduction

The embedding function in VectorStores is responsible for creating numerical representations of text. This can be configured through Settings.embedFunc.

Using Settings.embedFunc

The embedFunc is a simple async function that takes text input and returns embedding vectors. You can configure it globally via Settings:

import { Settings, TextEmbedFunc } from "@vectorstores/core";

// Define your embedding function
const myEmbedFunc: TextEmbedFunc = async (input) => {
  // Your embedding logic here
  // Returns Promise<number[][]> - an array of embeddings
};

// Set it globally
Settings.embedFunc = myEmbedFunc;

OpenAI Embeddings

For OpenAI embeddings, you can use the OpenAI instructions.

Local Embedding

For local embeddings, you can use the HuggingFace instructions.