README.md

---
title: BitNet CPU Assistant
emoji: 🧠
colorFrom: blue
colorTo: indigo
sdk: streamlit
sdk_version: 1.53.1
app_file: app.py
pinned: false
---

# 🧠 BitNet CPU Assistant

Run Microsoft's 1-bit LLMs at lightning speeds on standard CPUs.

## Features
- **1-bit Inference**: Powered by `bitnet.cpp` for maximum efficiency.
- **CPU Optimized**: No GPU required, perfect for free-tier hosting.
- **Low Memory**: 3B models run in under 4GB of RAM.

## Deployment Instructions
1. Create a new Streamlit Space on Hugging Face.
2. Upload all files from this directory.
3. The Space will automatically install dependencies from `packages.txt` and `requirements.txt`.
4. The first run will compile the BitNet kernels (approx 3-5 mins).

## Local Development
```bash
pip install -r requirements.txt
streamlit run app.py
```