This repository provides instructions for setting up Ollama with the DeepSeek-R1 model for local LLM interactions. This setup serves as a foundation for Python script integration and n8n workflow automation.
Ollama enables local execution of large language models with:
- GPU acceleration support
- Simple CLI interface
- REST API for application integration
- No data transmission to third-party services
Download and install Ollama from the official website
To run the DeepSeek-R1 model:
- Start the Ollama Server:
ollama serve- Install and run the model:
ollama run deepseek-r1:7bThe setup includes several services working together:
- Ollama Server: LLM inference engine (port 11434)
- OpenWeb UI: Web interface for Ollama (port 3000)
- n8n: Workflow automation platform (port 5678)
- Qdrant: Vector database for embeddings (ports 6333, 6334)
- PostgreSQL: Database for n8n (port 5432)
- Start all containers:
docker compose up -dThe DeepSeek-R1 7B model will be automatically downloaded and loaded when the container starts.
- Access the services:
- OpenWeb UI: http://localhost:3000
- n8n Dashboard: http://localhost:5678
- Qdrant Dashboard: http://localhost:6333/dashboard
To stop the containers:
docker compose downData persistence:
- Model data:
ollama-model-cachevolume - n8n data:
n8n-datavolume - OpenWeb UI data:
open-webui-datavolume - PostgreSQL data:
postgres-datavolume - Qdrant data:
qdrant-datavolume
To enable GPU support:
- Uncomment the GPU-related sections in
docker-compose.yml - Change OpenWeb UI image to
ghcr.io/open-webui/open-webui:cuda
Browse additional models in the Ollama DeepSeek Library.