Transformers.js

Overview

Transformers.js is a JavaScript library for running 🤗 Transformers directly in your browser with no server required. Designed to be functionally equivalent to the original Python library with a similar API.

Key Features

• No Server Required: Run models entirely in the browser or Node.js
• ONNX Runtime: Uses ONNX Runtime for efficient model execution
• Python API Parity: Similar API to Python Transformers library
• Multimodal Support: Text, image, and audio tasks

Capabilities

• Text Embeddings: Generate vector embeddings for semantic search
• Zero-Shot Classification: Classify text, images, and audio without training
• Object Detection: Detect objects in images
• Text Generation: Generate text with various models
• Question Answering: Extract answers from context

Common Use Cases

Vector Embeddings in Browser: Popular model all-MiniLM-L6-v2 produces 384-dimension vectors from text, running entirely client-side

Related Content: Build semantic related posts or document similarity without backend infrastructure

Privacy-Focused Applications: All processing happens locally, ensuring user data privacy

Installation

import { pipeline } from "@huggingface/transformers";

Example Applications

• Text embeddings in Bun, Deno, and Node.js
• Browser-based semantic search
• Client-side content recommendations
• Offline-capable AI features

Platform Support

Works in:

• Modern browsers
• Node.js (ESM and CJS)
• Bun
• Deno

Pricing

Free and open-source.