Best Practices for Running Models on HPC

• Request Only What You Need:
  Estimate resources and adjust based on past jobs.

• Organize Jobs & Logs:
  Use descriptive job names and structured output folders.

• Use the Right Storage:
  Place large and temporary files in $VSC_DATA and $VSC_SCRATCH.

• Manage Environments:
  Use modules/conda and document dependencies. Version-control your code.

• Promote Reproducibility:
  Save SLURM scripts, configs, and note job parameters.

• Debug Efficiently:
  Check logs, use interactive/test jobs, and fail fast on errors.

• Document & Share:
  Keep READMEs and notebooks. Organize scripts, configs, and results.

• Clean Up:
  Remove unused data and files regularly.

---