Skip to content

Latest commit

 

History

History
271 lines (179 loc) · 8.65 KB

README.md

File metadata and controls

271 lines (179 loc) · 8.65 KB
Raw
title date description tags
A Complete Guide to Installing Ollama and OpenWebUI Locally
2024-11-26
A complete guide to installing Ollama and OpenWebUI locally on your device.
generative-ai
llm
ollama
openwebui

A Complete Guide to Installing Ollama and OpenWebUI Locally

Welcome to this step-by-step guide, where you'll learn how to set up Ollama and OpenWebUI locally on your device. Whether you're a curious learner, a tech enthusiast, or someone looking to experiment with large language models (LLMs), this guide will help you navigate through the process. By the end of this project, you'll have a fully functioning environment to run AI language models locally, giving you control, privacy, and flexibility.

What is OpenWebUI?

OpenWebUI is a powerful open-source frontend for interacting with language models (LLMs). It provides an intuitive chat-based interface where users can ask questions, generate text, and even interact with the models using voice or images. OpenWebUI supports features like:

  • Highlighting code.
  • Formatting with Markdown and LaTeX.
  • Managing model parameters.
  • Retrieving external information using Retrieval Augmented Generation (RAG).

This platform is designed for ease of use, making it accessible on both computers and mobile devices.

What is Ollama?

Ollama is a lightweight, open-source backend tool that manages and runs large language models locally on your device. Written in Go, it allows you to deploy and interact with models like Llama2, Mistral, and others. Ollama runs in the background, acting as the engine behind the scenes for OpenWebUI or other frontend interfaces.

Key benefits of Ollama include:

  1. Local execution of models, ensuring privacy.
  2. No reliance on cloud-based services, reducing costs.
  3. Compatibility with robust GPUs for faster processing.

What is a Model?

A model is the brain of an AI system. In this context, it refers to a pre-trained machine learning model, like Llama2 or Llama 3.2, which generates human-like text based on input prompts. These models vary in complexity:

  • Smaller models (e.g., 1 billion parameters): Require less memory and are faster but may lack depth in understanding.
  • Larger models (e.g., 70 billion parameters): Offer better accuracy and nuanced responses but demand more computational resources.

Why Run Models Locally Instead of in the Cloud?

Running models locally has several advantages:

  • Privacy: Your data stays on your device, ensuring complete control and security.
  • Cost-Effectiveness: Avoid subscription fees for cloud services.
  • Customizability: Fine-tune models and integrate them with your own workflows.
  • Offline Access: Use models without internet connectivity.

However, local setups may require a decent computer with sufficient RAM, CPU, and GPU for optimal performance.

Project Overview: Installing Ollama and OpenWebUI Locally

This project is divided into the following steps:

  1. Install Docker (and WSL for Windows users).
  2. Install Ollama.
  3. Install Llama 3.2: 1B Model.
  4. Run and configure OpenWebUI.
  5. Test and interact with your setup.

Step 1: Install Docker

Docker is a containerization platform that allows you to run applications in isolated environments. OpenWebUI uses Docker to simplify its setup.

For Windows Users

Docker Desktop requires Windows Subsystem for Linux (WSL). Follow these steps:

  • Install WSL:

    • Open PowerShell as Administrator and run: 
      wsl --install
    • Restart your computer if prompted.
    • Confirm installation by running: 
      wsl --list --online

  • Install Docker Desktop:

    • Download Docker Desktop from the official website and install it.
    • During installation, ensure the "Enable WSL 2 integration" option is selected.
    • After installation, start Docker Desktop and verify it is running.

  • Test Docker:

    • Open a terminal (Command Prompt, PowerShell, or your WSL distribution) and run: 
      docker --version
    • You should see the Docker version information.

For macOS and Linux Users

  • Visit the Docker installation page and download the version tailored to your OS.
  • Install Docker and follow the on-screen instructions.
  • Test Docker by running: 
    docker --version

Step 2: Install Ollama

Ollama serves as the backend for running models. Follow these steps:

For macOS Users

  • Open your terminal and install Ollama via Homebrew:

    brew install ollama

  • Verify installation:

    ollama --version

For Windows Users

  • Download the Windows installer from Ollama's official website.

  • Run the installer and follow the prompts.

  • Add the following environment variable to allow Ollama to listen on all interfaces:

    • Search for "Environment Variables" in the Start menu.
    • Add a new User Variable:
      • Variable Name: OLLAMA_HOST
      • Variable Value: 0.0.0.0:8080
    • Save changes and restart your computer.

  • Verify installation:

    • Open a terminal and run: 
      ollama --version

Step 3: Install Llama 3.2: 1B Model

Llama 3.2 is one of the most advanced large language models, offering improved efficiency and multimodal capabilities (text and image processing). Follow these steps to install the 1B parameter version of Llama 3.2:

  • Open a terminal and ensure Ollama is running:

    ollama serve

  • Pull the Llama 3.2: 1B model:

    ollama pull llama3.2:1b
    • This command downloads the lightweight 1B parameter version of Llama 3.2 to your local system.
    • Depending on your internet speed, this process may take a few minutes.

  • Verify the model is installed:

    ollama list
    • You should see llama3.2:1b in the list of available models.

  • Test the model by running it directly in the terminal:

    ollama run llama3.2:1b
    • Example prompt: 
      >>> What is the capital of France?
    • You should get a response like: 
      The capital of France is Paris.

Ollama is now ready to serve Llama 3.2: 1B as part of your local AI setup.

Step 4: Run OpenWebUI

OpenWebUI provides the interface to interact with models. It uses Docker to simplify deployment.

  1. Open a terminal and run the following command to start OpenWebUI:

    docker run -d -p 3000:8080 --add-host=host.docker.internal:host-gateway \
-v open-webui:/app/backend/data --name open-webui --restart always \
ghcr.io/open-webui/open-webui:main

  2. Confirm the Docker container is running:

    docker ps

    You should see open-webui listed.

  3. Open your browser and navigate to:

    http://localhost:3000

Step 5: Test and Interact with the Setup

Now that everything is set up, it’s time to explore the interface and test the model.

  1. Log in to OpenWebUI:

    • When prompted, create an account or log in with default credentials.

  2. Connect Llama 3.2: 1B Model:

    • Go to the Settings menu in OpenWebUI.
    • Navigate to Models and select llama3.2:1b from the list.
    • Click Connect to link the model to the interface.

  3. Test the Model:

    • Go back to the main chat window.
    • Enter a prompt and interact with the Llama 3.2 model.

    Example Prompt:

    Write a short story about a cat who becomes a detective.

  4. Experiment with Other Features:

    • Use Markdown or LaTeX for formatting text and mathematical equations.
    • Adjust model parameters to customize responses.

Troubleshooting Tips

  • Docker Issues on Windows: Ensure WSL 2 is installed and properly integrated with Docker Desktop.
  • Ollama Not Responding: Check the environment variable OLLAMA_HOST and ensure it’s set to 0.0.0.0:8080.
  • Model Not Downloading: Verify your internet connection and ensure the Docker container for OpenWebUI is running.

Conclusion

Congratulations! 🎉 You’ve successfully set up Ollama and OpenWebUI locally on your device with the Llama 3.2: 1B model. This setup empowers you to explore the capabilities of large language models in a private, customizable, and cost-effective way. Use this opportunity to experiment with AI, generate creative content, or integrate the system into your projects.

Feel free to share your experiences or ask questions as you continue exploring this exciting technology. Happy experimenting!