Tech Undocs

Appearance

ollama

Running Ollama with a User-Friendly Web Interface ​

This guide offers instructions for setting up a web-based interface for Ollama using Docker. It simplifies the management of local language models by providing a user-friendly front-end called Open WebUI.

Open WebUI ​

Open WebUI is a user-friendly web interface for Ollama that simplifies the interaction with LLMs. It offers several key features:

  • Intuitive Interface: Provides a clean and easy-to-navigate interface for managing and interacting with LLMs.
  • Model Management: Allows you to easily pull, list, and manage Ollama models.
  • Chat Interface: Offers a chat-like interface for interacting with the loaded LLM.
  • RAG (Retrieval Augmented Generation): Supports uploading documents and querying them using the LLM, enabling powerful knowledge-based interactions.
  • Customizable Settings: Allows you to configure various settings, such as system prompts and model parameters.
  • Cross-Platform Compatibility: Accessible from any device with a web browser.

Docker Compose ​

Using Docker Compose simplifies the deployment and management of Ollama and Open WebUI. The following compose.yaml file defines the necessary services:

yaml
services:
  ollama:
    image: docker.io/ollama/ollama
    volumes:
      - ollama:/root/.ollama

  open-webui:
    image: ghcr.io/open-webui/open-webui
    environment:
      OLLAMA_BASE_URL: http://ollama:11434
      WEBUI_AUTH: "False"  # Disable authentication for simplicity in this example
    depends_on:
      - ollama  # Ensure Ollama is running before starting Open WebUI
    ports:
      - 8080:8080  # Expose port 8080 for accessing the Web UI
    volumes:
      - open-webui:/app/backend/data

  cloudflared:
    image: docker.io/cloudflare/cloudflared
    command: tunnel --no-autoupdate --url http://open-webui:8080

volumes:
  ollama:
  open-webui:

Environment variables ​

Open WebUI uses environment variables for configuration. Here are the descriptions for the ones used in the compose.yaml file:

  • WEBUI_AUTH
    • Default: True
    • Description: Enables or disables authentication for the Web UI. Setting it to False disables authentication, making it easier to access in a local development environment. Use with caution in production environments.
  • OLLAMA_BASE_URL
    • Default: http://localhost:11434
    • Description: Specifies the URL of the Ollama server. In this Docker setup, it points to the ollama service within the Docker network.

For more environment variables and advanced configurations, refer to the official Open WebUI documentation: https://docs.openwebui.com/getting-started/advanced-topics/env-configuration

Start containers ​

To start the containers, navigate to the directory containing the compose.yaml file and run:

sh
docker compose up -d

The -d flag runs the containers in detached mode (in the background).

Pull models ​

After the containers are running, you can pull models using the following command:

sh
docker compose exec ollama ollama pull llama3.2:1b

This command executes ollama pull llama3.2:1b inside the ollama container, downloading the llama3.2:1b model. You can replace it with any other available Ollama model.

Once the model is downloaded, you can access the Open WebUI by navigating to http://localhost:8080 in your web browser.