Hugging Face Jobs
Run training and batch ML workloads on Hugging Face Jobs infrastructure. Launch UV scripts and Docker-based jobs, pick the right GPU/CPU hardware, estimate cost, and authenticate with tokens.
This skill runs workloads on Hugging Face Jobs. It packages code as UV scripts or Docker jobs, selects appropriate hardware, estimates cost before launching, wires token authentication, and manages job lifecycle — for fine-tuning runs, evaluations, and one-off batch processing without provisioning your own cluster.
huggingface jobs training gpu mlops
When to use
Use when you want to run a training run, evaluation, or batch ML job on managed Hugging Face infrastructure with defined hardware and cost.
Examples
Launch a training job
Run on managed GPUs
Package my fine-tuning script as a Hugging Face Job on an A100 and estimate the cost before running it
Run a batch evaluation
Docker-based job
Create a Docker-based Hugging Face Job that evaluates my model on a benchmark dataset and writes results to the Hub