About This Project
Overview
A PyTorch LSTM-based question-answering system that processes natural language queries and returns relevant answers, built as a hands-on exploration of sequence models before transformer-based QA.
Motivation
I wanted to understand sequence models from first principles by building an LSTM-based QA system before reaching for transformer-based approaches.
Lessons Learned
LSTMs struggle to retain context over longer passages — watching that limitation firsthand made the motivation behind attention mechanisms and transformers click in a way reading about them never did.