A Python-based interface for interacting with Ollama's API, specifically designed to work with the DeepSeek R1 8B model.

This project provides several Python scripts to interact with the Ollama API in different ways:

• 📝 Simple text generation
• 💬 Chat interface with streaming responses
• 🧵 Chat with conversation context/history

The project also includes a Dockerfile to set up Ollama with the DeepSeek-R1 8B model pre-loaded.

• 🐳 Docker (for container-based setup)
• 🐍 Python 3.6+
• 📦 requests library

pip install -r requirements.txt

Build and run the Docker container with the DeepSeek model:

# Build the Docker image
docker build -t ollama-deepseek .

# Run the container
docker run -d --name ollama-deepseek -p 11434:11434 ollama-deepseek

Ensure the model is running properly:

# Check available models
docker exec -it ollama-deepseek ollama list

# Test with a simple curl request
curl -X POST http://localhost:11434/api/generate -d '{
  "model": "deepseek-r1:8b",
  "prompt": "Explain what is debt in simple terms."
}' -H "Content-Type: application/json"

For basic text generation:

python interactive.py

You'll be prompted to enter your query and optionally specify a different model.

For a chat-style interface (single query):

python chat.py

This allows you to specify a user message, an optional system prompt, and choose the model.

For an interactive chat that maintains conversation history:

python chat-context.py

This script maintains the conversation context, allowing for more coherent multi-turn interactions. Type 'exit' to end the conversation.

Simple text generation using the /api/generate endpoint. Streams tokens as they're generated.

Uses the /api/chat endpoint for a more chat-like interaction. Supports system prompts for setting the behavior of the model.

Extends the chat functionality by maintaining conversation history, allowing for contextual multi-turn conversations.

# View logs
docker logs ollama-deepseek

# Stop the container
docker stop ollama-deepseek

# Remove the container
docker rm ollama-deepseek

You can easily modify these scripts to work with other Ollama models by changing the default model parameter.

• If you encounter connection issues, ensure the Ollama server is running and accessible at localhost:11434
• Check Docker logs for any errors: docker logs ollama-deepseek
• Verify that the model was downloaded successfully by running ollama list inside the container