deepseek-r1-local 1.0.0

Offline AI Web UI with Council Deliberation System

🚀 DeepSeek R1 Local Web UI

An offline, privacy-focused AI chat application with TinyLlama model, Web Search Integration, and Council Deliberation System optimized for CPU performance.

Features

• 🤖 Offline AI Chat (TinyLlama-1.1B, CPU-optimized)
• 🔍 Web Search Integration (DuckDuckGo)
• 🏛️ Council Deliberation System (5 AI personas, voting)
• ⚡ Performance Optimizations (response caching, fast inference)
• 🔒 100% Privacy (all processing local)

Council Members

• 🧠 Dr. Logic (Analytical Rationalist)
• 📚 Professor Sage (Historical Scholar)
• 💡 Innovator Nova (Creative Visionary)
• ❤️ Advocate Heart (Empathetic Humanist)
• 🎯 Pragmatist Ray (Practical Realist)

Installation

1. Install via pip (when published)

pip install deepseek-r1-local

2. Download Model

deepseek-r1-local download-model

3. Start Server

deepseek-r1-local start

4. Open Browser

Go to: http://localhost:5000

Usage

• Use toggles for Web Search and Council Mode in the UI
• Ask questions, get AI and council responses
• Council mode: 5 personas deliberate, vote, and select a winning proposal

Python API Example

from deepseek_r1_local import ModelManager, Council, WebSearcher

model = ModelManager()
model.load_model()
response = model.generate_response("Hello!", max_length=50)

council = Council()
results = council.deliberate("Should I learn Rust or Go?", model)
print(council.format_results(results))

searcher = WebSearcher()
results = searcher.search("Python tutorials", max_results=5)
print(searcher.format_search_results(results))

File Structure

deepseek_r1-local/
├── deepseek_r1_local/
│   ├── __init__.py
│   ├── app.py
│   ├── cli.py
│   └── templates/
│       └── index.html
├── models/           # Model files (downloaded)
├── setup.py
├── pyproject.toml
├── requirements.txt
├── README.md
├── LICENSE
├── MANIFEST.in
├── CHANGELOG.md
├── INSTALL.md
├── QUICKREF.md
├── PACKAGE_GUIDE.md
├── build.sh
├── test_package.py

Troubleshooting

• If model fails to load, re-run download-model
• Use deepseek-r1-local start --port 8080 for a different port
• For help: deepseek-r1-local --help

License

MIT License

Links

• PyPI: https://pypi.org/project/deepseek-r1-local/
• GitHub: https://github.com/yourusername/deepseek-r1-local