sue-exp-naming

What problem does it solve?

Inconsistent, invalid, or colliding experiment names across W&B, Slurm, per-GPU labels, and result ledgers cause tracking errors, failed job submissions, and mismatched results for SUE scale-up ML/HPC experiments.

Core Features & Use Cases

• Deterministic Name Generation: Creates consistent run IDs, W&B entities, Slurm job names, per-GPU labels, and ledger fields that remain identical across retries and reruns for reliable end-to-end tracking.
• Comprehensive Validation: Checks for name collisions, length limit compliance with Slurm/W&B/CSV constraints, and CSV-safe formatting to prevent submission and logging failures.
• Use Case: Before submitting a scale-up experiment, use this skill to generate all required identifiers so your W&B runs, Slurm jobs, and result ledgers are correctly linked and fully traceable.

Quick Start

Invoke the sue-exp-naming skill to generate and validate all required experiment identifiers for your upcoming SUE dryrun or fullrun.