hf-jobs

Execute Python workloads on Hugging Face Jobs with configurable hardware.

Updated Mar 12, 2026

Search Tags:#gpu #batch processing #huggingface #jobs #cloud compute #uv scripts

Ready to run

npx skills add https://github.com/3kpro/aiarsenal --skill hf-jobs

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: hf-jobs
Source: https://github.com/3kpro/aiarsenal/tree/main/hugging-face-skills/skills/hugging-face-jobs
Command: npx skills add https://github.com/3kpro/aiarsenal --skill hf-jobs

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill allows you to run any Python workload on Hugging Face's managed cloud infrastructure, eliminating the need for local setup and providing access to powerful GPUs/TPUs.

Core Features & Use Cases

Cloud Execution: Run scripts on CPUs, GPUs, or TPUs without local configuration.
Scalable Workloads: Ideal for data processing, batch inference, model training, and experiments.
Result Persistence: Seamlessly save outputs back to the Hugging Face Hub.
Use Case: You need to run a batch inference job on a large dataset using a GPU. Instead of setting up a local environment, you can use this Skill to submit the job directly to Hugging Face Jobs, specifying the hardware and script.

Quick Start

Use the hf-jobs skill to run a Python script named 'my_script.py' on a CPU basic instance for 30 minutes.

Frequently Asked Questions about hf-jobs

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I run Python batch processing jobs on cloud GPUs without local environment setup?▼

You can run Python batch processing jobs on cloud GPUs by submitting workloads directly to Hugging Face Jobs infrastructure. This approach eliminates local configuration, allowing you to specify hardware flavors, timeouts, and execute UV scripts for scalable compute without environment setup.

Can I use Docker images to run scalable workloads on Hugging Face cloud compute?▼

Yes, you can use Docker images to run scalable workloads on Hugging Face cloud compute. The infrastructure supports executing arbitrary Python workloads via both UV scripts and Docker images, allowing configurable hardware access for data processing and model training tasks.

What is the best way to persist batch inference results back to the Hugging Face Hub?▼

The best way to persist batch inference results is to use Hugging Face Jobs, which provides secure token management for Hub interactions. You can seamlessly save outputs from your cloud execution directly back to the Hugging Face Hub during or after workload processing.

Does Hugging Face Jobs support configurable hardware flavors for model training experiments?▼

Yes, Hugging Face Jobs supports configurable hardware flavors for model training experiments. You can specify CPUs, GPUs, or TPUs and set timeouts for your workloads, ensuring reproducible batch processing and flexible access to powerful cloud compute resources.

How do I manage secure tokens for Hub interactions when running cloud workloads on Hugging Face?▼

To manage secure tokens for Hub interactions when running cloud workloads on Hugging Face, the Jobs infrastructure includes built-in secure token management. This ensures safe authentication and seamless result persistence back to the Hub during your Python execution.

When should I not use Hugging Face Jobs for cloud execution?▼

You should not use Hugging Face Jobs for cloud execution if your workload requires highly customized local environment configurations or real-time interactive processing. It is designed for reproducible batch processing and scalable workloads rather than continuous local development.