autolab-managed-experiment Skill
Run one Autolab benchmark experiment safely on Hugging Face Jobs. Use when a planner, reviewer, or experiment worker is preparing, auditing, launching, or reviewing a single train.py hypothesis against the current local promoted master. Published by huggingface in context-course.
What is autolab-managed-experiment Skill?
Run one Autolab benchmark experiment safely on Hugging Face Jobs. Use when a planner, reviewer, or experiment worker is preparing, auditing, launching, or reviewing a single train.py hypothesis against the current local promoted master. Published by huggingface in context-course. This profile combines repository metadata with install, compatibility, and usage signals so developers can quickly decide whether it fits their agent workflow before opening the source repository.
Automated repository signals based on public metadata such as recency, license, installation evidence, and adoption. These are not a security audit or endorsement.
Key capabilities
- Includes SKILL.md support
- Reusable instructions support
- Developers using context-course
Technical details
- Install or run with Copy skill directory
When to use autolab-managed-experiment Skill
- Use it for developers using context-course.
Built with
Editorial notes
Source
- Creator: huggingface
- Repository: huggingface/context-course
- Skill file: projects/pre-training/.agents/skills/autolab-managed-experiment/SKILL.md
What it does
Run one Autolab benchmark experiment safely on Hugging Face Jobs. Use when a planner, reviewer, or experiment worker is preparing, auditing, launching, or reviewing a single train.py hypothesis against the current local promoted master.
Skill instructions
Use this for any single Autolab experiment that should result in exactly one managed benchmark run. Workflow 1. Load the local operator env and refresh from local master: - . ~/.autolab/credentials - uv run scripts/refreshmaster.py --fetch-dag 2. Edit only train.py for the single intended hypothesis. 3. Run preflight before launch: - uv run scripts/hfjob.py preflight 4. Launch exactly one managed experiment: - uv run scripts/hfjob.py launch --mode experiment 5. Stream logs and parse the final metric: - uv run scripts/hfjob.py logs <JOBID --follow --output /tmp/autolab-run.log - uv run scripts/parsemetric.py /tmp/autolab-run.log 6. Record the run locally: - uv run scripts/submitpatch.py --comment "..." 7. Promotion is local and only happens if valbpb beats current master. Guardrails - Treat trainorig.py as the refreshed local-master base. If preflight reports multiple known hypothesis categories, stop and inspect the diff before launching. - Ignore repo git main and origin/main when jud
Explore related resources
Frequently asked questions
What is autolab-managed-experiment?
autolab-managed-experiment is a open-source AI agent skill with Copy skill directory. Run one Autolab benchmark experiment safely on Hugging Face Jobs. Use when a planner, reviewer, or experiment worker is preparing, auditing, launching, or reviewing a single train.py.
Who is autolab-managed-experiment best for?
autolab-managed-experiment is best for reusing agent instructions, scripts, and references.
How do I install autolab-managed-experiment?
Install or run autolab-managed-experiment using Copy skill directory. Check autolab-managed-experiment for the latest setup command.
Is autolab-managed-experiment actively maintained?
autolab-managed-experiment may need a closer maintenance check before production use.
Auto-fetched from GitHub.
