app.py

import torch
import joblib
from flask import Flask, request, jsonify
from transformers import BertTokenizer, BertForSequenceClassification
import torch.nn.functional as F

# Initialize Flask application
app = Flask(__name__)

# Load model and tokenizer
model = joblib.load('model.joblib')
tokenizer = BertTokenizer.from_pretrained('bert-base-uncased')
model.eval()

# Set device to CUDA if available
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model.to(device)

# Inference function
def classify_text(text):
    encoding = tokenizer(str(text), truncation=True, padding=True, max_length=128, return_tensors='pt')
    input_ids = encoding['input_ids'].to(device)
    attention_mask = encoding['attention_mask'].to(device)

    with torch.no_grad():
        outputs = model(input_ids, attention_mask=attention_mask)
        logits = outputs.logits
        probabilities = F.softmax(logits, dim=-1)
        confidence, predicted_class = torch.max(probabilities, dim=-1)
    
    class_names = ["JAILBREAK", "INJECTION", "PHISHING", "SAFE"]
    predicted_label = class_names[predicted_class.item()]
    confidence_score = confidence.item()

    return predicted_label, confidence_score

# Define the inference route
@app.route('/inference', methods=['POST'])
def inference():
    data = request.json
    if 'text' not in data:
        return jsonify({"error": "No text provided"}), 400

    text = data['text']
    label, confidence = classify_text(text)

    return jsonify({
        'text': text,
        'classification': label,
        'confidence': confidence
    })

# Start the Flask server
if __name__ == '__main__':
    app.run(host='0.0.0.0', port=8080)