Ollama Web UI Setup Instructions

Prerequisites

• Docker Desktop installed and running
• Ollama installed on your host system
• NVIDIA GPU with updated drivers (if using GPU acceleration)

Setup Steps

1. Start Ollama on your system

   ollama serve

2. Verify Docker Desktop

   • Ensure Docker Desktop is running
   • Check WSL integration is enabled in Docker Desktop settings
   • Make sure virtualization is enabled in BIOS

3. Start the containers

   docker-compose up -d

Accessing the Web UIs

• Ollama Web UI: http://localhost:3001
• CUDA Web UI: http://localhost:3002

Available Models

To download and use models:

1. Open terminal and run:

   ollama pull <model_name>

2. Example models:
   • llama2
   • mistral
   • codellama
   • neural-chat

Troubleshooting

1. Connection Issues

   • Ensure Ollama is running on port 11434
   • Check Docker Desktop status
   • Run docker ps to verify containers are running

2. GPU Issues

   • Verify NVIDIA drivers are up to date
   • Check nvidia-smi output
   • Ensure GPU is recognized in Docker Desktop

3. Container Logs

   docker logs open-webui-ollama
   docker logs open-webui-cuda

Stopping the Services

docker-compose down

Data Persistence

• UI settings and chat history are stored in the open-webui volume
• Models downloaded through Ollama are stored on your host system