์Šคํ‚ฌ hugging-face-jobs
๐Ÿ“ฆ

hugging-face-jobs

์•ˆ์ „

Execute ML workloads on Hugging Face cloud

Run GPU/TPU workloads without local hardware setup. Submit Python scripts to managed Hugging Face Jobs infrastructure with secure token authentication.

์ง€์›: Claude Codex Code(CC)
๐Ÿฅ‰ 72 ๋ธŒ๋ก ์ฆˆ
1

์Šคํ‚ฌ ZIP ๋‹ค์šด๋กœ๋“œ

2

Claude์—์„œ ์—…๋กœ๋“œ

์„ค์ • โ†’ ๊ธฐ๋Šฅ โ†’ ์Šคํ‚ฌ โ†’ ์Šคํ‚ฌ ์—…๋กœ๋“œ๋กœ ์ด๋™

3

ํ† ๊ธ€์„ ์ผœ๊ณ  ์‚ฌ์šฉ ์‹œ์ž‘

ํ…Œ์ŠคํŠธํ•ด ๋ณด๊ธฐ

"hugging-face-jobs" ์‚ฌ์šฉ ์ค‘์ž…๋‹ˆ๋‹ค. Submit UV script for sentiment analysis

์˜ˆ์ƒ ๊ฒฐ๊ณผ:

Job submitted successfully. Job ID: job_abc123. Monitor at https://huggingface.co/jobs/username/job_abc123. Estimated completion: 5 minutes.

"hugging-face-jobs" ์‚ฌ์šฉ ์ค‘์ž…๋‹ˆ๋‹ค. Check status of running jobs

์˜ˆ์ƒ ๊ฒฐ๊ณผ:

Found 2 running jobs: job_abc123 (RUNNING, 45% complete), job_def456 (QUEUED). Use hf_jobs('logs', {'job_id': 'job_abc123'}) to view logs.

๋ณด์•ˆ ๊ฐ์‚ฌ

์•ˆ์ „
v1 โ€ข 2/25/2026

All 284 static analysis findings are false positives. The 'Ruby/shell backtick execution' detections are actually Python code examples in markdown documentation. 'Hardcoded URLs' are legitimate documentation links to Hugging Face resources. Environment variable access (HF_TOKEN) is documented authentication behavior for Hub operations. No malicious patterns detected.

1
์Šค์บ”๋œ ํŒŒ์ผ
1,039
๋ถ„์„๋œ ์ค„ ์ˆ˜
0
๋ฐœ๊ฒฌ ์‚ฌํ•ญ
1
์ด ๊ฐ์‚ฌ ์ˆ˜
๋ณด์•ˆ ๋ฌธ์ œ๋ฅผ ์ฐพ์ง€ ๋ชปํ–ˆ์Šต๋‹ˆ๋‹ค
๊ฐ์‚ฌ์ž: claude

ํ’ˆ์งˆ ์ ์ˆ˜

38
์•„ํ‚คํ…์ฒ˜
100
์œ ์ง€๋ณด์ˆ˜์„ฑ
87
์ฝ˜ํ…์ธ 
50
์ปค๋ฎค๋‹ˆํ‹ฐ
100
๋ณด์•ˆ
78
์‚ฌ์–‘ ์ค€์ˆ˜

๋งŒ๋“ค ์ˆ˜ ์žˆ๋Š” ๊ฒƒ

ML Engineer running batch inference

Process thousands of samples through a model using cloud GPUs without local hardware investment. Submit UV scripts with vLLM for high-throughput generation.

Data Scientist transforming datasets

Execute data processing pipelines on Hugging Face datasets using Polars or Pandas. Push transformed results back to Hub securely.

Researcher running experiments

Execute reproducible ML experiments with scheduled jobs. Monitor progress via job URLs and retrieve results from Hub repositories.

์ด ํ”„๋กฌํ”„ํŠธ๋ฅผ ์‚ฌ์šฉํ•ด ๋ณด์„ธ์š”

Basic script execution 
Run this Python script on Hugging Face Jobs: [paste script]. Use CPU and 30 minute timeout.
GPU inference with token authentication 
Execute this inference script on an A10G GPU. Push results to my Hub repo username/results. Include HF_TOKEN authentication.
Scheduled data processing 
Create a scheduled job that runs this data transformation script daily at 9 AM. Use cpu-upgrade hardware and persist output to Hub.
Docker-based custom workload 
Run this command using the pytorch/pytorch CUDA image on an A100 GPU. Set timeout to 4 hours and monitor for completion.

๋ชจ๋ฒ” ์‚ฌ๋ก€

  • Always use secrets={'HF_TOKEN': '$HF_TOKEN'} for Hub authentication - never hardcode tokens
  • Set appropriate timeouts with 20-30% buffer for your workload type
  • Persist results to Hub or external storage before job completion - environment is ephemeral

ํ”ผํ•˜๊ธฐ

  • Using local file paths with hf_jobs() MCP tool - scripts must be inline code or URLs
  • Polling job status repeatedly - wait for user to request status checks
  • Using env instead of secrets for tokens - env variables are visible in logs

์ž์ฃผ ๋ฌป๋Š” ์งˆ๋ฌธ

Do I need a paid Hugging Face plan to use Jobs?
Yes. Hugging Face Jobs requires a Pro, Team, or Enterprise plan. Free accounts cannot submit jobs.
How do I authenticate my jobs with the Hub?
Use secrets={'HF_TOKEN': '$HF_TOKEN'} in your job config. The $HF_TOKEN placeholder is automatically replaced with your logged-in token.
What happens to my results after the job finishes?
All files are deleted when the job ends. You must push results to Hugging Face Hub, upload to cloud storage (S3/GCS), or send via API before completion.
Can I use local script files with hf_jobs()?
No. The hf_jobs() MCP tool requires inline script content or a URL. Read local files with Path('script.py').read_text() and pass the content string.
How do I choose the right GPU for my workload?
Start with t4-small for testing. Use a10g-large for 7-13B models. Use a100-large for 13B+ models. See the hardware guide in this skill for detailed recommendations.
Can I schedule jobs to run automatically?
Yes. Use hf_jobs('scheduled uv', {...}) with a schedule parameter like '@daily' or CRON expression '0 9 * * 1' for 9 AM Mondays.

๊ฐœ๋ฐœ์ž ์„ธ๋ถ€ ์ •๋ณด

์ž‘์„ฑ์ž

sickn33

๋ผ์ด์„ ์Šค

Complete terms in LICENSE.txt

๋ฆฌํฌ์ง€ํ† ๋ฆฌ

https://github.com/sickn33/antigravity-awesome-skills/tree/main/skills/hugging-face-jobs

์ฐธ์กฐ

main

ํŒŒ์ผ ๊ตฌ์กฐ

๐Ÿ“„ SKILL.md