launch-with-slurm

What problem does it solve?

Dispatching work inside a SLURM allocation using srun to efficiently utilize reserved compute on single-node or multi-node clusters.

Core Features & Use Cases

• Context-aware dispatch: prefer srun over raw torchrun or bash when $SLURM_JOB_ID is set to preserve environment and IO behavior across ranks.
• Allocation reading: read AllocTRES and NodeList via scontrol to tailor commands to the allocation.
• Logging and robustness: use -o with --open-mode to capture distributed logs and support resumed runs.
• Use Case: launching a training job or validation step inside an existing SLURM allocation, possibly across multiple nodes.

Quick Start

Invoke srun with the appropriate flags inside the current SLURM allocation to start your job.