Good practices

Note

About these examples: (Click the arrow below to expand)

This section contains some minimal examples of how to run jobs on the Mila cluster. Each example is self-contained and can be run as-is directly on the cluster without error. Each example has the following structure:

• job.sh: SLURM sbatch script. Can be launched with sbatch job.sh.

• main.py: Example python script.

Some examples are displayed as a difference with respect to a “base” example. For instance, the multi-gpu example is shown as a difference with respect to the single-gpu example.

Work-in-progress: This section will contain minimal examples illustrating various good practices that should be observed when using the Mila cluster.

• Checkpointing
• Wandb Setup
• Launch many jobs from same shell script
• Hyperparameter Optimization with Oríon
• Launch many tasks on same GPU
• Launch many jobs using SLURM job arrays