Correctly Install and Run Large Language Models and Ollama by Using Windows Subsystem for Linux

In this tutorial, we explain how to correctly install Ollama and Large Language Models (LLMs) by using Windows Subsystem for Linux (WSL). For those of you who are not familiar with WSL, WSL enables you to run a Linux Ubuntu distribution on the Windows Operating System. By using WSL, you can run a Linux Ubuntu distribution in a Windows terminal. This enables you to experiment with Linux Ubuntu application and development environment without actually installing Linux Ubuntu. This is especially important for people interested in machine learning and large language modeling since it is easier to install and run machine learning applications on Linux Ubuntu. This is mainly because most of the machine learning developers use Linux Ubuntu for developing machine learning applications. The YouTube tutorial explaining all the steps is given below.

In this tutorial, we will install Ollama and several AI models on the WSL. First, we need to install WSL. For that purpose open a command prompt in the administrator mode, and type the following command

wsl --install

This command will install WSL as well as the default version of Linux Ubuntu which is 24.04. After this command is executed you will need to restart the computer in order to start Linux Ubuntu. After that you will need to setup your username and password in order to be able to use Linux Ubuntu in WSL.

The next step is to install Ollama. To do that, make sure that Ollama is not running on Windows (if you have previously installed Ollama on your Windows computer). To stop Ollama, click on the icon in the bottom right corner of your screen, and click on “Quit Ollama”

You would need to quit Ollama before installing Ollama on WSL. This is mainly because only a single instance of Ollama can run at the same time. To make sure that Ollama is not running, on your web browser, enter the following address:

http://localhost:11434/

If Ollama is disabled, then the following error screen will appear

The next step is to install Ollama on WSL. To do that, start a WSL Ubuntu Session by clicking on Start, and searching for Ubuntu and clicking on the Ubuntu icon

Then, you need to install curl. To install Curl, type

sudo apt update && sudo apt upgrade 
sudo apt install curl 
curl --version

Then, you need to allow inbound connections to the appropriate Ollama internal port:

sudo apt install ufw
sudo ufw allow 11434/tcp

Then, we can install Ollama by typing

curl -fsSL https://ollama.com/install.sh | sh

To test that Ollama is working, open a web browser and go to the address

http://localhost:11434/

The output should look like this.

After installing Ollama, let us test Ollama by downloading and running the model. Go to the Ollama website, and search for a model. For example, search for Llama3.2

Then to download and run this model, simply type in the Linux Ubuntu terminal:

ollama run llama3.2:3b

This should start the model. The exit the model, simply type CTRL+D

Then, you will notice that after you exit WSL, Ollama will still run. To fix this issue, go back to the Ubuntu WSL, and stop Ollama by typing these commands

systemctl stop ollama.service

To disable Ollama, type this

systemctl disable ollama.service

To confirm its status, type this

systemctl status ollama.service

To check the status, you can also try this

ps aux | grep ollama
pgrep ollama

Finally, to check the GPU processes, you need to type this

sudo apt install gpustat
gpustat --show-cmd --show-pid

To start Ollama once again, type this

sudo systemctl daemon-reload
sudo systemctl enable ollama
sudo systemctl start ollama
ollama serve
sudo systemctl status ollama