Transformer-based Sentiment Analysis for Movie Reviews

This project, completed in the Future AWS AI Scientist program, focused on natural language processing using transformer models.

Highlights

• Built a sentiment classifier that labels movie reviews as positive or negative.
• Used a pre-trained transformer model and fine-tuned it on a labeled dataset.
• Evaluated performance with metrics like accuracy and F1 score, and inspected example predictions.

What I learned

• Practical steps for fine-tuning transformers for text classification.
• Tokenization, batching, and handling variable sequence lengths.
• How model performance changes with training time, learning rate, and dataset size.

Together with the dog-breed classifier, this project demonstrates my ability to work with both vision and text AI workloads on AWS.