Back to projects
Machine Learning · Shipped
TextGen
A natural-language generation app — Bi-LSTM/LSTM networks combined with a pre-trained LLM, tuned with attention mechanisms and prompt engineering for coherent, contextually relevant text.
Overview
A natural-language generation app — Bi-LSTM/LSTM networks combined with a pre-trained LLM, tuned with attention mechanisms and prompt engineering for coherent, contextually relevant text.
What I built
Bi-LSTM + LSTM architecture combined with a pre-trained LLM, trained and served with PyTorch
Attention mechanisms + prompt engineering for contextual coherence
BLEU/ROUGE evaluation against reference text, plus an interactive Streamlit UI for live generation
Built with
PythonPyTorchLSTMStreamlit