title	emoji	colorFrom	colorTo	sdk	app_file	app_port	pinned	license	python_version	env
ClaimCracker	🔍	blue	red	docker	web/main.py	10000	false	mit	3.11.11	PYTHONUNBUFFERED=1 MAX_THREADS=2 PYTORCH_NO_CUDA=1 OMP_NUM_THREADS=1 MKL_NUM_THREADS=1 ENVIRONMENT=production LOG_LEVEL=INFO HF_HOME=/tmp/.cache/huggingface LOG_DIR=/tmp/logs

ClaimCracker v2

Fake news detection system combining ML with a modern web API.

Features

• ML Pipeline

  • DistilBERT-based classification (96.03% accuracy)
  • Fast inference (~13.94ms/text)
  • Efficient text preprocessing
  • Production-optimized model loading

• Web API

  • FastAPI with async support
  • Request caching and rate limiting
  • Prometheus monitoring
  • Comprehensive error handling
  • OpenAPI documentation

Environment Setup

Requirements

• Python 3.11.11
• Conda (recommended) or virtualenv

Using Conda

# Create conda environment
conda create -n claimcracker2 python=3.11.11

# Activate environment
conda activate claimcracker2

# Install dependencies
pip install -r requirements.txt

Using virtualenv

# Create virtual environment
python -m venv venv

# Activate environment (Windows)
.\venv\Scripts\activate
# OR (Unix/macOS)
source venv/bin/activate

# Install dependencies
pip install -r requirements.txt

Quick Start

1. Environment Setup

python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate
pip install -r requirements.txt

2. Run API

cd web
uvicorn main:app --host 0.0.0.0 --port 10000

API will be available at:

• API: http://localhost:10000
• Docs: http://localhost:10000/docs
• Metrics: http://localhost:10000/metrics

API Endpoints

Core Endpoints

• GET / - Welcome and status
• GET /health - Health check
• POST /predict - Fake news detection

Management

• GET /metrics - Prometheus metrics
• GET /cache/stats - Cache statistics
• POST /cache/clear - Clear prediction cache

Rate Limits

• /predict: 30 requests/minute
• /health: 60 requests/minute
• /cache/*: 10 requests/minute

Performance

Model Metrics

• Validation Accuracy: 96.03%
• F1 Score: 0.9603
• Inference Time: 13.94ms/text
• Model Size: <500MB

API Performance

• Response Time: <1s
• Cache Hit Ratio: ~70%
• Memory Usage: <512MB

Development

Prerequisites

• Python 3.11
• FastAPI
• PyTorch
• Transformers

Testing

pytest tests/

Project Structure

ClaimCracker-v2/
├── Dataset/           # Training data
├── src/              # ML pipeline
│   └── ml/
│       ├── config/   # Configuration
│       ├── data/     # Data processing
│       ├── models/   # Model architecture
│       └── training/ # Training logic
├── web/              # FastAPI application
├── tests/            # Test suite
└── models/           # Model artifacts

Documentation

• API Reference: docs/api_reference.md
• Model Details: models/final_model/README.md