metadata

title: ClickBERT Detector
emoji: 🐨
colorFrom: green
colorTo: indigo
sdk: gradio
sdk_version: 5.12.0
app_file: app.py
pinned: false
short_description: Fine-tuned BERT-uncased for headline clickbait detection

🚀 ClickBERT Detector

A robust, fine-tuned BERT-based model for clickbait detection.

Overview

ClickBERT Detector leverages state-of-the-art NLP techniques to identify clickbait headlines with high accuracy. Designed for both researchers and developers, this tool aims to streamline the detection of misleading headlines in online content.

Features

Pretrained BERT Backbone: Fine-tuned specifically for clickbait classification.
Interactive Web Interface: Seamlessly test the model through an intuitive web application.
Easy Integration: API-ready for integration into larger systems.
Lightweight Model: Optimized with safetensors for efficient storage and faster loading.

File Structure

app.py: Main script for launching the web interface.
model.safetensors: The fine-tuned BERT model for clickbait detection.
requirements.txt: Python dependencies required for the project.
config.json: Configuration settings for the model and tokenizer.
training_args.bin: Training parameters used for fine-tuning the BERT model.

Usage

Web Interface: Test the model by providing headlines through a user-friendly interface.
API Integration: Send POST requests with headline text to retrieve predictions.
Batch Processing: Use the provided scripts to process datasets efficiently.

Model Details

ClickBERT is fine-tuned on a curated dataset of clickbait and non-clickbait headlines, achieving exceptional performance metrics:

Accuracy: 95%
F1 Score: 0.94