app.py

import torch
from transformers import DistilBertTokenizer, DistilBertForSequenceClassification
import gradio as gr

# Load tokenizer and model
tokenizer = DistilBertTokenizer.from_pretrained('distilbert-base-uncased')
model = DistilBertForSequenceClassification.from_pretrained('distilbert-base-uncased', num_labels=2)
model.load_state_dict(torch.load('best_model (4).pth', map_location=torch.device('cpu')))
model.eval()

# Prediction function
def classify_news(text):
    inputs = tokenizer(text, return_tensors='pt', truncation=True, padding=True, max_length=512)
    with torch.no_grad():
        outputs = model(**inputs)
        probs = torch.nn.functional.softmax(outputs.logits, dim=1)
        predicted_class = torch.argmax(probs, dim=1).item()
        labels = ["Fake", "True"]
        return {labels[0]: float(probs[0][0]), labels[1]: float(probs[0][1])}

# Gradio interface
iface = gr.Interface(
    fn=classify_news,
    inputs=gr.Textbox(lines=10, placeholder="Paste a news article here..."),
    outputs=gr.Label(num_top_classes=2),
    title="Fake News Detector",
    description="Detect whether a news article is real or fake using a fine-tuned DistilBERT model."
)

# Launch the app
if __name__ == "__main__":
    iface.launch()