Archive epoch artifacts and reset ML/HPC experiment workspaces.
npx skills add https://github.com/dongzhuoyao/deepresearch --skill sue-reset
Please help me install this Agent Skill. Skill: sue-reset Source: https://github.com/dongzhuoyao/deepresearch/tree/main/.codex/skills/sue-reset Command: npx skills add https://github.com/dongzhuoyao/deepresearch --skill sue-reset
This skill solves the pain point of manually cleaning up completed DeepResearch scale-up experiment workspaces, which risks accidental deletion of critical source code, configuration, or historical experiment data, and often leads to inode bloat from scattered loose output files.
Use the sue-reset skill to archive your current scale-up experiment outputs and initialize a clean workspace for a new experiment epoch.
High-intent search queries and answers about installing and using this skill.
To safely reset an ML experiment workspace, you must archive epoch artifacts to a versioned history directory while preserving source code, Hydra configs, and orchestration state. This ensures workspace hygiene and prevents accidental deletion of critical files during epoch transitions.
The best way to archive HPC scale-up experiment artifacts is moving runtime configs, ledgers, logs, and local W&B cache to a structured versioned history directory. This generates versioned archive manifests ensuring experiment reproducibility across sandbox backends like LUMI and Snellius.
Yes, you can clean up W&B runs and remote sandbox outputs during ML experiment archiving. This requires explicit user confirmations for destructive actions, ensuring safe deletion of W&B groups and remote sandbox output roots after archiving completes.
Workspace reset supports sandbox backends including RunPod, AutoDL, Brev, LUMI, Snellius, and NM5. It handles versioned archiving across these platforms, enforcing inode-safe output contracts and path contract validation for scale-up ML workflows.
Inode bloat happens when scattered loose output files accumulate across experiment epochs. Prevent it by archiving artifacts to a structured versioned history directory and enforcing inode-safe output contracts during workspace reset to maintain hygiene across epoch transitions.
StreamPilotOrgScore growth experiments using the ICE framework and update experiment JSON.
manutejAutomate ML lifecycle management with MLflow experiment tracking and model registry.
JeffallanOrchestrate end-to-end ML pipelines across Kubeflow, Airflow, and cloud-native platforms.
Elad73Audit stale project artifacts and generate a confirmation-gated cleanup plan.
