Movie Review Sentiment Classifier
The problem
A practical end-to-end deep-learning build: take raw text reviews, train a recurrent network on them, and deploy a model that classifies new reviews in real time.
What I built
A three-stage pipeline on the IMDB dataset. Preprocessing covers HTML stripping, lemmatization, stop-word removal, and a train/test split. Feature extraction and training use TF-IDF at 5,000 features feeding a PyTorch LSTM with cross-entropy loss and Adam over 20 epochs. Real-time inference reloads the saved model state and vectorizer for arbitrary user input.
The outcome
A working binary classifier with full model checkpointing and a reusable inference path, documented step by step so other beginners can replicate the pipeline.
Interested in this kind of work? Get in touch