Local Mini DeepSeek with Qwen3 & Ollama

This project leverages Qwen3:4B and Ollama inference to create a 100% local ChatGPT-like app with a hybrid thinking UI built on Streamlit.

---

🚀 Installation and Setup

1. Setup Ollama

# Install Ollama on Linux
curl -fsSL https://ollama.com/install.sh | sh
# Pull the Qwen3:4B model
ollama pull qwen3:4b

2. Install Python Dependencies

pip install streamlit ollama

---

▶️ Running the App

streamlit run app.py

---

✨ Features

• Local inference using Ollama and Qwen3
• Hybrid Thinking UI: enable /think mode to reveal chain-of-thought steps
• Toggle reasoning on/off directly in the chat input

---

🗂 Project Structure

.
├── assets/
│   ├── logo_qwen3.png
│   └── ollama.jpg
├── app.py
└── README.md

---

📝 Usage

1. Launch the app with streamlit run app.py.
2. Toggle Enable step-by-step reasoning 🧠 at the bottom.
3. Type your question in the input box and press Enter.
4. View answer with or without chain-of-thought.