🛠️ Setting up Ollama with DeepSeek-R1 on Windows

This guide walks you through setting up Ollama on Windows to run the DeepSeek-R1:7B model with a local FastAPI interface.

---

📌 Prerequisites

• Windows with Docker Desktop installed
• NVIDIA GPU (for GPU acceleration)
• WSL 2 Enabled (if running Docker on Windows)
• Python 3.12 (for the FastAPI interface)

---

🚀 1. Start Ollama & Install DeepSeek

(a) Ensure Docker is Running

Run the following command to verify that Docker is active:

docker ps

(b) Start the Ollama Container

Start Ollama in a Dockerized environment:

docker-compose up -d

(c) Download & Load the DeepSeek-R1 Model

Pull the DeepSeek-R1:7B model into Ollama:

docker exec -it ollama ollama pull deepseek-r1:7b

Run the model inside the Ollama container:

docker exec -it ollama ollama run deepseek-r1:7b

(d) Test DeepSeek Using cURL

You can manually test DeepSeek via Ollama's API:

curl -X POST http://localhost:11434/api/generate -d '{
  "model": "deepseek-r1:7b",
  "prompt": "What is DeepSeek?",
  "stream": false
}'

---

🐍 2. Set Up Python Environment

(a) Install & Create a Conda Environment

If you don’t have Conda, install Miniconda.

Then, create a new environment for FastAPI:

conda create -n ds-ollama python=3.12

Activate the environment:

conda activate ds-ollama

(b) Install FastAPI & Dependencies

Inside the activated environment, install the required Python packages:

pip install fastapi uvicorn requests jinja2

---

💡 3. Start the FastAPI Interface

To launch the web interface for DeepSeek:

1. Ensure Ollama is running:

   docker-compose up -d

2. Run FastAPI locally:

   uvicorn main:app --host 0.0.0.0 --port 8000 --reload

3. Open the chat interface in your browser:

   • 📌 http://localhost:8000

---

🔄 4. Stop & Restart Services

(a) Stopping Ollama

To stop Ollama, run:

docker-compose down

(b) Restarting the FastAPI Server

If you need to restart the FastAPI server:

uvicorn main:app --host 0.0.0.0 --port 8000 --reload

---

🎯 Done!

Now you have Ollama running DeepSeek-R1 7B inside Docker and a FastAPI-powered ChatGPT-style interface on Windows! 🚀

Need Help? 🤔
If you run into any issues, check:

• Docker logs: docker logs ollama
• FastAPI logs: Look at the terminal where FastAPI is running

---

Made with ❤️.