JupyterHub is a platform that brings the power of Jupyter Notebooks to groups of users. It gives users access to computational environments and resources without burdening the users with installation and maintenance tasks.

This project utilises and customises the JupyterHub platform to support some of the online experiments of the AI Curriculum.

You can access to the JupyterHub platform directly via:

• Direct links to a specific experiment in “Master Course”, or
• Shortcut links in “E-Platform Portal”, or
• https://cuhkjc-aiforfuture.hk/jupyterhub

For more information about JupyterHub, you can also refer to https://jupyter.org/hub

It takes approximately 10 minutes to load all the experiments into the user's home directory for any first-time access, please wait patiently and refresh the page after 10 minutes.

To retrieve your own API key, please go to the user profile page, you may find your API key accordingly.

1. Access Guidelines
i) Each Project School will be entitled for no more than 232 user-hours (user-hours = number of users*hours used by each user) per week on the GPU resources.

ii) To better utilize the GPU resources, and to allow students to play around with more different AI experiments by reducing the number of concurrent users, Project Schools are recommended to divide students into groups to play with each experiment.

iii) Testing and preparation on “Interaction” (IN) or “Empowerment” (EM) Modules by teachers are recommended to be conducted outside normal school hours (i.e. before 0800 and after 1730), to reduce potential collision with classroom usage.

iv) The servers are scheduled to be under maintenance during 2300 to 0200 every day.

v) Any system update, policy changes or special maintenance will be announced in the “Announcement Section” on Moodle.

vi) Limited technical support will be provided during school and public holidays.


2. Usage Policies
i) JupyterHub session will be timed out after no user activity is observed for two hours.

ii) Users should only use the web platform to access the course materials according to the instructions provided. Any other non-designed usages are strictly prohibited.

iii) The Project will provide a Jupyter Notebook in the server so that users can check the current GPU usage in the server.



3. Memory Policies
i) Memory available for each user is 300 MB.

ii) Chapter 3 EM, Chapter 4 IN & EM, Chapter 5 EM, Chapter 6 EM involve uploading user data. Users should pay more attention to those chapters.

iii) Users can check the current memory usage of their accounts in Moodle.

iv) The data stored in the web platform will be reset every academic year.

You may check the GPU status and memory balance of the server by the following steps.

Step 1: Log-in JupyterHub platform.


Step 2: Open “check_server_status” folder.


Step 3 : Click “Check_server_status.ipynb” to start the Jupyter Notebook.


Step 4: In Jupyter Notebook, you can check the memory usage of the server. You can check the usage data of each chapter by selecting the experiment from the drop-down list. The unit of memory usage is Byte.


If you see a “No such file or directory” error, usually because the corresponding experiment does not keep user-created data in the server.


Step 5: You can also use the Jupyter Notebook to check GPU usage data by clicking the “Check GPU Usage” button.


Step 6: You can check the availabilities of the experiment(s) of each chapter based on available GPU resources by clicking the button “Runnable Experiments”. You can also view how many experiments you can start based on available GPU resources.