I have a jupyter lab script for AI training and it works well for shared-cpu. The algorithm takes around 11 hours to run in our lab system. I have been trying to run a job on shared-gpu but I am unable to do so. Here is my script header where I specify the shared-gpu and the time required.
When I submit the job using sbatch it launchs the debug-cpu instead of shared-gpu (shown in figure below) and it also shuts down after 15 minutes.
I have been looking around but unable to sort the issue. I would be grateful if anyone could help me out to run this algorithm on shared-gpu.