README.md

---
title: BitNet Text Summarizer
emoji: 📝
colorFrom: purple
colorTo: indigo
sdk: docker
app_port: 8501
pinned: false
short_description: Streamlit summarizer using microsoft/bitnet-b1.58-2B-4T
---


# 📝 BitNet Text Summarizer (Streamlit)


An open‑source text summarizer running on **microsoft/bitnet-b1.58-2B-4T** with a map‑reduce strategy for long inputs. Deployed on **Hugging Face Spaces** using **Docker + Streamlit**.


## Features
- **BitNet (local)** inference inside the Space (GPU recommended)
- **HF Inference API fallback** (optional) if you provide `HF_TOKEN`
- **Map‑Reduce** summarization for long documents
- Adjustable generation parameters (temperature, top_p, token budgets)


## Quickstart
1. Create/Duplicate a Space with **SDK: Docker**.
2. Add files from this portfolio: `Dockerfile`, `requirements.txt`, `README.md`, and `src/` folder.
3. In **Settings → Hardware**, pick a **GPU** for faster startup (CPU works but is slower).
4. (Optional) Add a repo secret `HF_TOKEN` to enable the fallback engine.
5. Open the app, paste text, and click **Summarize**.


## Local Dev (optional)
```bash
python -m venv .venv && source .venv/bin/activate # Windows: .venv\Scripts\activate
pip install -r requirements.txt
streamlit run src/streamlit_app.py