ShivamBharti-29/sentiment-analysis-rnn-streamlit ? reverse-engineered prompt

Build me a simple web app that can take a movie review and tell me if it sounds positive or negative.

I want the app to clean up the text first, like making it lowercase and removing punctuation, numbers, and URLs, then turn the review into features with TF IDF and run it through an RNN sentiment model built with PyTorch. The app should have a nice Streamlit interface where I can paste a review, click a button, and see the prediction right away.

Please keep the code organized so the preprocessing, model loading, and prediction logic are easy to follow. If you need to check current docs or best practices for Streamlit or PyTorch, go ahead and look them up online. Also make sure it uses the saved model and vectorizer files so it works without retraining every time.