Skills
.Work. You
Rest

sue-job-max-parallel

Community

Get safe max parallel jobs for scale-up ML/HPC runs.

Software Engineering#hpc#job-scheduling#ml-training#sue-pipeline#scale-up-experiments#max-parallel#sandbox-queries
Authordongzhuoyao
Version1.0.0
Installs0

System Documentation

What problem does it solve?

Guessing parallel job counts for ML/HPC scale-up experiments often leads to quota overruns, failed job submissions, or wasted compute resources. This Skill eliminates that guesswork by querying sandbox systems for a dynamic, point-in-time safe concurrency limit.

Core Features & Use Cases

  • Dynamic Capacity Querying: Pulls real-time job limit, queue state, and quota data from Slurm backends (LUMI, Snellius, NM5) and direct-execution platforms (Brev, RunPod, AutoDL).
  • Pipeline Integration: Enforces DeepResearch workspace contracts and integrates with the SUE workflow to block stale or invalid capacity values before dryrun or fullrun launches.
  • Use Case: Before submitting a 200-variant distributed training experiment on LUMI, use this Skill to get the exact number of parallel jobs you can safely run without hitting your account or partition limits.

Quick Start

Use the sue-job-max-parallel skill at the start of your dryrun or fullrun planning to retrieve the safe maximum parallel job count for your current scale-up experiment.

Dependency Matrix

Required Modules

None required

Components

Standard package

💻 Claude Code Installation

Recommended: Let Claude install automatically. Simply copy and paste the text below to Claude Code.

Please help me install this Skill:
Name: sue-job-max-parallel
Download link: https://github.com/dongzhuoyao/deepresearch/archive/main.zip#sue-job-max-parallel

Please download this .zip file, extract it, and install it in the .claude/skills/ directory.
View Source Repository

Agent Skills Search Helper

Install a tiny helper to your Agent, search and equip skill from 536,000+ vetted skills library on demand.