JupyterLab (Jupyter) is an open-source environment primarily used in fields like data science, machine learning, and scientific computing. It can be accessed via a web browser.
Usage Instructions
- Log in to the frontend server fe1 from your PC/Mac.
$ ssh (account-name)@fe1.scl.kyoto-u.ac.jp
- On fe1, execute a PBS interactive batch job to log in to a compute server.
Example:
$ qsub -I -q APG -l select=1:ncpus=32:mem=512gb:ngpus=1
(Note) If the system is busy and no machine meets the specified resources (number of cores, memory size, GPU count), it will remain in the pending state.
- Once you have logged in to the compute server, check the hostname of the machine you are logged into (this will be needed later for SSH port forwarding).
$ hostname
- Navigate to your working directory, load the necessary module files (Python, PyTorch, NVIDIA-PyTorch, etc.), and start Jupyter.
Example:
$ cd (working-directory)/
$ module load pytorch/2.4.1
$ jupyter lab --ip=0.0.0.0 --no-browser --port=8123
(Note) The port number "8123" can generally be any number between 1024 and 49151. If you receive a message like "port already in use" and the process terminates, please specify a different port number.
- On the terminal of your PC/Mac, run the following SSH command (replace "ch8" with the hostname of the machine you logged into with PBS).
$ ssh -L 8123:ch8:8123 (account-name)@fe1.scl.kyoto-u.ac.jp
- From the web browser on your PC/Mac, open the following link:
http://localhost:8123/
Note: After finishing your work in Jupyter, make sure to exit the PBS interactive batch job and log out from the SSH port forwarding session.
|
