README.md

---
title: CPU Only Zero Shot Text Classification
emoji: 🏃
colorFrom: gray
colorTo: purple
sdk: gradio
sdk_version: 5.31.0
app_file: app.py
pinned: false
license: apache-2.0
short_description: ' CPU-only Zero-Shot Text Classification'
---
# 🏷️ Zero-Shot Text Classification  
[![Hugging Face Space](https://img.shields.io/badge/HuggingFace-Spaces-blue?logo=huggingface)](https://huggingface.co/spaces/your-username/zero-shot-classifier)  
[![Gradio UI](https://img.shields.io/badge/Gradio-5.31.0-brightgreen?logo=gradio)]  
[![Model](https://img.shields.io/badge/Model-BART--MNLI-orange)](https://huggingface.co/facebook/bart-large-mnli)  
[![License](https://img.shields.io/badge/License-MIT-lightgrey)](LICENSE)

---

## 🚀 Overview  
Unlock **zero-shot classification** for any text—no fine-tuning required.  
Define your own label set on the fly and see how well each label matches your input, powered by **BART-MNLI** on CPU.

> **AI buzzwords:**  
> Natural Language Inference • Zero‐shot Learning • Transformer-based NLP • Real-time Inference • Edge Deployment • Cloud-native Demo

---

## ✨ Features

| 🔑 Feature                     | 🔍 Description                                                  |
|--------------------------------|----------------------------------------------------------------|
| **🚀 Zero-Shot**                | Classify into arbitrary categories without task-specific data  |
| **⚡ CPU-Only Inference**        | Runs on free Hugging Face Spaces (2 vCPU / 16 GB RAM)         |
| **🎛️ Single vs. Multi-Label**   | Toggle between exclusive or overlapping labels                |
| **🎨 Interactive UI**           | Gradio Blocks with text input, label list, mode toggle, table |
| **🔧 No Training Needed**       | Leverages pre-trained BART-MNLI via HF Transformers           |
| **☁️ Instant Deploy**           | Commit three files—Spaces auto-builds & hosts your demo       |

---

## 🏗️ How It Works

1. **User Input** – Paste any sentence or paragraph.  
2. **Label Definition** – Enter comma-separated candidate labels (e.g. “Positive, Negative, Question”).  
3. **Model Inference** – Pipeline computes entailment scores for each label.  
4. **Result Table** – Display each label with its confidence score.

All done **locally on the Space**, ensuring **privacy**, **zero API cost**, and **lightning speed**.

---

## 🛠️ Local Development

```bash
git clone https://github.com/your-username/zero-shot-classifier.git
cd zero-shot-classifier
python3 -m venv venv && source venv/bin/activate
pip install -r requirements.txt
python app.py

## Latest Update

- Updated BART-MNLI model for better accuracy. - May 29, 2025 📝

**Website**: https://ghostainews.com/
**Discord**: https://discord.gg/BfA23aYz