Features

🤗 HuggingFace Integration - Use any HuggingFace tokenizer model
🔌 Optional Plugin - Install only when you need real tokenization
📦 Zero Config - Works automatically with @chonkiejs/core
⚡ Progressive Enhancement - Core works without it, better with it

Installation

Install with npm:

npm i @chonkiejs/token @chonkiejs/core

Install with pnpm:

pnpm add @chonkiejs/token @chonkiejs/core

Install with yarn:

yarn add @chonkiejs/token @chonkiejs/core

Install with bun:

bun add @chonkiejs/token @chonkiejs/core

Quick Start

Simply install this package alongside @chonkiejs/core, then use tokenizer names:

import { RecursiveChunker, TokenChunker } from '@chonkiejs/core';

// Use GPT-2 tokenization (automatically uses @chonkiejs/token)
const chunker = await RecursiveChunker.create({
  tokenizer: 'Xenova/gpt2',
  chunkSize: 512
});

const chunks = await chunker.chunk('Your text here...');

Supported Models

Any HuggingFace model from transformers.js:

• Xenova/gpt2
• Xenova/gpt-4
• bert-base-uncased
• google-bert/bert-base-multilingual-cased
• And many more!

See: https://huggingface.co/models?library=transformers.js

Usage Examples

With RecursiveChunker

import { RecursiveChunker } from '@chonkiejs/core';

const chunker = await RecursiveChunker.create({
  tokenizer: 'Xenova/gpt2',
  chunkSize: 512
});

const chunks = await chunker.chunk('Your document...');

With TokenChunker

import { TokenChunker } from '@chonkiejs/core';

const chunker = await TokenChunker.create({
  tokenizer: 'bert-base-uncased',
  chunkSize: 256,
  chunkOverlap: 50
});

const chunks = await chunker.chunk('Your text...');

Direct Tokenizer Usage

import { HuggingFaceTokenizer } from '@chonkiejs/token';

const tokenizer = await HuggingFaceTokenizer.create('Xenova/gpt2');

const count = tokenizer.countTokens('Hello world!');
const tokens = tokenizer.encode('Hello world!');
const text = tokenizer.decode(tokens);

console.log(`Token count: ${count}`);

How It Works

When you call Tokenizer.create('gpt2') in @chonkiejs/core:

1. Core tries to dynamically import @chonkiejs/token
2. If installed: Uses HuggingFaceTokenizer
3. If not installed: Shows helpful error message

This keeps core lightweight while allowing advanced tokenization when needed!

Contributing

Want to help grow Chonkie? Check out CONTRIBUTING.md to get started! Whether you're fixing bugs, adding features, improving docs, or simply leaving a ⭐️ on the repo, every contribution helps make Chonkie a better CHONK for everyone.

Remember: No contribution is too small for this tiny hippo!