app.py

# Import dependencies
import gradio as gr
from transformers import AutoTokenizer, AutoModelForSequenceClassification, T5Tokenizer, T5ForConditionalGeneration
import torch
import nltk
from nltk.corpus import wordnet
import subprocess

# Download NLTK data (if not already downloaded)
nltk.download('punkt')
nltk.download('stopwords')
nltk.download('wordnet')  # Download WordNet

# Check for GPU and set the device accordingly
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")

# Load AI Detector model and tokenizer from Hugging Face (DistilBERT)
tokenizer = AutoTokenizer.from_pretrained("distilbert-base-uncased-finetuned-sst-2-english")
model = AutoModelForSequenceClassification.from_pretrained("distilbert-base-uncased-finetuned-sst-2-english").to(device)

# Load Parrot Paraphraser model and tokenizer for humanizing text
paraphrase_tokenizer = T5Tokenizer.from_pretrained("prithivida/parrot_paraphraser_on_T5")
paraphrase_model = T5ForConditionalGeneration.from_pretrained("prithivida/parrot_paraphraser_on_T5").to(device)

# AI detection function using DistilBERT
def detect_ai_generated(text):
    inputs = tokenizer(text, return_tensors="pt", truncation=True, max_length=512).to(device)
    with torch.no_grad():
        outputs = model(**inputs)
        probabilities = torch.softmax(outputs.logits, dim=1)
    ai_probability = probabilities[0][1].item()  # Probability of being AI-generated
    return f"AI-Generated Content Probability: {ai_probability:.2f}%"

# Humanize the AI-detected text using the Parrot Paraphraser model
def humanize_text(AI_text):
    inputs = paraphrase_tokenizer(AI_text, return_tensors="pt", max_length=512, truncation=True).to(device)
    with torch.no_grad():  # Avoid gradient calculations for faster inference
        paraphrased_ids = paraphrase_model.generate(
            inputs['input_ids'],
            max_length=inputs['input_ids'].shape[-1] + 20,  # Slightly more than the original input length
            num_beams=4,
            early_stopping=True,
            length_penalty=1.0,
            no_repeat_ngram_size=3,
        )
    paraphrased_text = paraphrase_tokenizer.decode(paraphrased_ids[0], skip_special_tokens=True)
    return f"Humanized Text:\n{paraphrased_text}"

# Gradio interface definition
ai_detection_interface = gr.Interface(
    fn=detect_ai_generated,
    inputs="textbox",
    outputs="text",
    title="AI Text Detection",
    description="Enter text to determine the probability of it being AI-generated."
)

humanization_interface = gr.Interface(
    fn=humanize_text,
    inputs="textbox",
    outputs="text",
    title="Text Humanizer",
    description="Enter text to get a human-written version, paraphrased for natural output."
)

# Combine both interfaces into a single Gradio app with tabs
interface = gr.TabbedInterface(
    [ai_detection_interface, humanization_interface],
    ["AI Detection", "Humanization"]
)

# Launch the Gradio app
interface.launch(debug=False)