simple-embeddings

Simple, zero-config embeddings generation for Next.js applications using Hugging Face Transformers. Get semantic search capabilities in your app with just one function call.

Installation

npm install simple-embeddings

Quick Start

import { generateEmbedding } from 'simple-embeddings'

const embedding = await generateEmbedding('Your text here')
console.log(embedding.length) // 384

Usage

Next.js API Route

// app/api/embeddings/route.ts
import { NextRequest, NextResponse } from 'next/server'
import { generateEmbedding } from 'simple-embeddings'

export async function POST(request: NextRequest) {
  try {
    const { text } = await request.json()
    const embedding = await generateEmbedding(text)

    return NextResponse.json({
      embedding,
      dimensions: embedding.length, // 384
    })
  } catch (error) {
    return NextResponse.json(
      { error: 'Failed to generate embedding' },
      { status: 500 }
    )
  }
}

API Reference

generateEmbedding(text: string): Promise<number[]>

Generates a 384-dimensional embedding vector for the given text.

Parameters:

• text (string): The input text to generate an embedding for

Returns:

• Promise<number[]>: A promise that resolves to a 384-dimensional array of numbers

Notes:

• First call downloads and loads the model (~90MB), subsequent calls are fast
• Works in Node.js server environments (API routes, server components)
• Returns normalized vectors suitable for cosine similarity

Technical Details

• Model: Xenova/all-MiniLM-L6-v2
• Vector dimensions: 384
• Framework: Hugging Face Transformers.js

License

MIT