polygraf-ai/ai-text-detector-dnn-based-supplementary

AI Text Detection Model

A PyTorch DNN model that detects whether text is human-written or AI-generated.

Install

git clone https://huggingface.co/polygraf-ai/ai-text-detector-dnn-based-supplementary
cd ai-text-detector-dnn-based-supplementary
pip install torch numpy nltk

Usage

from inference import predict_text

text = "Your text here to analyze"
result = predict_text(text, model_path="model_artifacts")
print(result)

# Output format:
{
    'label': 'Human-written',  # or 'AI-generated'
    'confidence': 0.85,
    'probabilities': {
        'Human-written': 0.85,
        'AI-generated': 0.15
    }
}

Model

• Task: Binary classification (Human vs AI)
• Architecture: DNN with embedding layer
• Framework: PyTorch
• Performance metrics available in training logs