Hugging Face Jobs

What This Skill Does

Run workloads on Hugging Face Jobs infrastructure — data processing, inference, experiments, batch jobs, GPU compute, and Python tasks.

Capabilities

• GPU Compute — Run workloads on GPU-enabled infrastructure
• Docker & UV Scripts — Support for Docker-based and UV script jobs
• Cost Estimation — Estimate costs before running jobs
• Result Persistence — Store and retrieve job outputs reliably

How It Works

1. Define your workload as a script or Docker container
2. Select hardware and estimate costs
3. Submit the job to HF Jobs infrastructure
4. Results are persisted and accessible after completion

Example Usage

"Run a batch inference job on GPU"
"Process a large dataset with distributed compute"
"Estimate the cost of a 4-hour training run"