README.md

metadata

title: Zero Shot Classifier
emoji: 🏃
colorFrom: purple
colorTo: yellow
sdk: gradio
sdk_version: 5.35.0
app_file: app.py
pinned: false
license: apache-2.0
short_description: Zero-shot text label predictor.

🧠 Zero-Shot Text Classifier (Hugging Face Version)

A smart and lightweight web app built with Gradio and Transformers that classifies your input text into the most likely label — using Zero-Shot Learning.

🧠 About the Model

• Pipeline: zero-shot-classification
• Model: facebook/bart-large-mnli
• Framework: Hugging Face Transformers
• Task: Predict a relevant label even if the model wasn't trained on it

💡 Features

• Accepts custom comma-separated labels
• Returns top predictions with confidence scores
• Works in real-time — hosted via Hugging Face Spaces

⚙️ Instructions for Users

This app uses zero-shot classification to find the most relevant label based on your input and label list.

👉 How to use:

1. Enter a sentence or paragraph
2. Enter comma-separated labels like: Technology, Sports, Food
3. The app will return top labels with confidence scores

⚠️ Note:

• Avoid overlapping or vague labels. It may reduce prediction accuracy.
• For example, a sentence about economy and healthcare might score both Finance and Health.
• The answer may reflect multiple topics if the sentence spans more than one area — this is expected behavior in such cases.

✅ Example 1:

• Text: Roger Federer won another grand slam title, cementing his legacy in tennis.
• Labels: ['Politics', 'Fashion', 'Sports']
• Prediction: Sports — 99.2%

✅ Example 2:

• Text: The chef used fresh ingredients and spices to prepare a delicious Indian curry.
• Labels: ['Food', 'Health', 'Travel']
• Prediction: Food — 88.9%

✅ Example 3:

• Text: Climate change is leading to rising sea levels and more frequent extreme weather events.
• Labels: ['Environment', 'Fashion', 'Technology']
• Prediction: Environment — 88.5%

---

🚀 How to Run Locally

Install the required packages:

pip install -r requirements.txt

Then run the app:

python app.py

Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference