Package Exports

zerolabel
zerolabel/dist/index.js

This package does not declare an exports field, so the exports above have been automatically detected and optimized by JSPM instead. If any package subpath is missing, it is recommended to post an issue to the original package (zerolabel) to support the "exports" field. If that is not possible, create a JSPM override to customize the exports field for this package.

Readme

zerolabel

Zero-shot multimodal classification made simple.

What is zerolabel?

Zero-shot classification means you can classify text and images without training data. Just provide your labels and get instant predictions.

Instead of spending weeks training models, you write one line of code.

The Problem

Traditional classification requires:

❌ Collecting thousands of labeled examples
❌ Training models for hours/days
❌ Managing ML infrastructure
❌ Retraining when you want new labels

The Solution

import { classify } from 'zerolabel';

// Classify anything, instantly
const results = await classify({
  texts: ['I love this product!'],
  labels: ['positive', 'negative', 'neutral'],
  inference_api_key: process.env.INFERENCE_API_KEY
});

console.log(results[0].predicted_label); // 'positive'

That's it. No training, no infrastructure, no complexity.

⚡ Powered by inference.net

High-performance AI inference infrastructure

zerolabel uses Google Gemma 3-27B through inference.net - production-ready AI infrastructure that just works.

✨ What You Can Classify

📝 Text

await classify({
  texts: ['This movie was incredible!'],
  labels: ['positive review', 'negative review'],
  inference_api_key: process.env.INFERENCE_API_KEY
});

🖼️ Images

await classify({
  images: [imageBase64],
  labels: ['cat', 'dog', 'bird'],
  inference_api_key: process.env.INFERENCE_API_KEY
});

🔀 Both Together

await classify({
  texts: ['Check out my new pet!'],
  images: [petPhotoBase64],
  labels: ['cute cat', 'cute dog', 'not cute'],
  inference_api_key: process.env.INFERENCE_API_KEY
});

🚀 Installation

npm install zerolabel

Get your API key: inference.net

💡 Real Examples

Email Classification

const emails = [
  'Meeting moved to 3pm',
  'Your order has shipped!', 
  'URGENT: Account security alert'
];

const results = await classify({
  texts: emails,
  labels: ['meeting', 'order_update', 'security', 'spam'],
  inference_api_key: process.env.INFERENCE_API_KEY
});

Content Moderation

const result = await classify({
  texts: ['User comment here...'],
  labels: ['appropriate', 'inappropriate', 'needs_review'],
  inference_api_key: process.env.INFERENCE_API_KEY
});

Image Recognition

const result = await classify({
  images: [productImageBase64],
  labels: ['clothing', 'electronics', 'books', 'home_goods'],
  inference_api_key: process.env.INFERENCE_API_KEY
});

🎯 Why zerolabel?

Traditional ML	zerolabel
❌ Weeks of training	✅ Instant results
❌ Thousands of examples needed	✅ Just provide labels
❌ Expensive GPU infrastructure	✅ Pay per prediction
❌ Complex model management	✅ One simple function
❌ Fixed categories	✅ Change labels anytime

📊 Response Format

{
  text: "I love this product!",
  predicted_label: "positive",
  confidence: 85.3,
  probabilities: {
    "positive": 0.853,
    "negative": 0.097,
    "neutral": 0.050
  }
}

🛠️ Advanced Usage

Custom Instructions

await classify({
  texts: ['The API response time was 50ms'],
  labels: ['technical', 'non-technical'],
  criteria: 'Focus on technical complexity and jargon',
  inference_api_key: process.env.INFERENCE_API_KEY
});

Batch Processing

await classify({
  texts: ['Comment 1', 'Comment 2', 'Comment 3'],
  labels: ['positive', 'negative', 'neutral'],
  inference_api_key: process.env.INFERENCE_API_KEY
});

🔐 Environment Setup

# .env
INFERENCE_API_KEY=your_key_here

import { classify } from 'zerolabel';

const results = await classify({
  texts: ['Your text here'],
  labels: ['label1', 'label2'],
  inference_api_key: process.env.INFERENCE_API_KEY
});

⚙️ Configuration

import { ZeroLabelClient } from 'zerolabel';

const client = new ZeroLabelClient({
  inference_api_key: 'your-key',
  maxRetries: 3  // Optional: retry failed requests
});

const results = await client.classify({
  texts: ['Hello world'],
  labels: ['greeting', 'question']
});

🧠 How It Works

Send your text/images → zerolabel
zerolabel formats prompt → Google Gemma 3-27B via inference.net
AI analyzes content → Returns predictions
Get structured results → JSON with labels and confidence

No training required. The AI already understands your labels.

📝 TypeScript Support

Full type safety out of the box:

import type { ClassificationResult } from 'zerolabel';

const results: ClassificationResult[] = await classify({
  texts: ['TypeScript is great!'],
  labels: ['positive', 'negative'],
  inference_api_key: process.env.INFERENCE_API_KEY
});

🤝 Support & Community

🐛 Issues: GitHub Issues
📧 Email: contact@zerolabel.dev
🌐 Docs: zerolabel.dev

📄 License

MIT License - Build amazing things! 🚀

Made with ❤️ for developers who want AI that just works

Powered by inference.net