You can use the AIBEX documentation, which is hosted on the DIUF GitLab and is therefore accessible only within the UniFR intranet.
In that documentation, choosing a GPU is always referred to as using the identifier of the GPU, such as gpu:l40s_48gb, but this is currently not the case and all GPUs are just called gpu, which you need to select by specifying the node, e.g. --nodelist=diufrd203.
Similarly, the shared directory for the HuggingFace is not yet available.
Here is the outline:
- Getting started: https://diuf-gitlab.unifr.ch:8091/slurm/docs/getting-started/
- Connecting: https://diuf-gitlab.unifr.ch:8091/slurm/docs/getting-started/connecting
- Setting up SSH Keys: https://diuf-gitlab.unifr.ch:8091/slurm/docs/getting-started/ssh-keys
- Setting up environment: https://diuf-gitlab.unifr.ch:8091/slurm/docs/getting-started/environment
- Running jobs: https://diuf-gitlab.unifr.ch:8091/slurm/docs/getting-started/running-jobs/
- Interactive jobs with srun: https://diuf-gitlab.unifr.ch:8091/slurm/docs/getting-started/running-jobs/srun
- Background jobs with sbatch: https://diuf-gitlab.unifr.ch:8091/slurm/docs/getting-started/running-jobs/sbatch
- Monitoring jobs: https://diuf-gitlab.unifr.ch:8091/slurm/docs/getting-started/monitoring
- Testing GPU Support with PyTorch: https://diuf-gitlab.unifr.ch:8091/slurm/docs/getting-started/test-gpu
- Hardware: https://diuf-gitlab.unifr.ch:8091/slurm/docs/hardware/
- Available Nodes: https://diuf-gitlab.unifr.ch:8091/slurm/docs/hardware/nodes
- Choosing Resources: https://diuf-gitlab.unifr.ch:8091/slurm/docs/hardware/choosing-resources