hpi-swa-teaching / ai4se-2024-setup

Resources to set up and manage LLM environments for the "AI in Software Engineering" Seminar

Environment Setup

Preparation

1. Clone this repository
2. Open run.sh with your editor of choice
3. Set AI_GROUP to your group number (1 - 8)
4. Set AI_LOGIN to a secure password to access your notebook
5. Save the file
6. Make sure there is a directory workspace directly in your user home. Otherwise, mkdir ~/workspace.

Creating a Container

Run sudo ./run.sh. This will create your container. A container for group x will spawn a Jupyter Lab server at port 8000 + x and use GPU x % 4.

Follow the URL printed by the script to access your Jupyter Lab in the browser. Now, enter the secure password chosen before (you might want to change it here).

Note: From now on, you cannot re-run the script unless you delete your container using sudo docker rm ai4se-<GROUP> with <GROUP> being your group number. MAKE SURE YOU ONLY DELETE YOUR OWN CONTAINER!

Managing a Container

You can start and stop your container using sudo docker start|stop ai4se-<GROUP>

Data created within Jupyter Lab (/workspace directory within the container) is synchronized to your ~/workspace directory in your user home. Everything inside this directory survives when the container is removed and re-created. Data outside this directory does not!