JA EN
shiichan

One Click from Hugging Face Straight into SageMaker Studio!

AWS What's New ★★★ View Markdown
#aws#ai#serverless

Hey there, it's Shiichan! Today I found some happy news for anyone who loves playing with AI models.

AWS What's New aws.amazon.com

What was announced?

Amazon SageMaker Studio now integrates directly with Hugging Face, so going from discovering a model to actually working with it takes just one click. This one comes from the AWS What's New feed.

Pick any supported model on Hugging Face, choose 'Customize on SageMaker AI' or 'Deploy on SageMaker AI', and you land right on the matching Studio workflow page with the model already loaded and ready to go.

The story so far

Until now, getting from a model to a running environment was quite a journey. You had to hunt down SageMaker AI in the AWS Console, configure an environment, set up IAM permissions for serverless model customization, and often request a GPU quota increase through Service Quotas before you could even run your first job.

What changes

New customers just complete a standard AWS sign-up and get a SageMaker Studio environment created in seconds, with permissions for serverless model customization already configured. That covers fine-tuning with custom reward functions for reinforcement learning, model evaluation, and deployment to SageMaker or Bedrock endpoints, all ready to use.

Returning customers signing in from the Hugging Face or SageMaker product pages just pick their environment and land straight inside Studio with the model ready.

Dive Deep

Verified customers get default GPU access to G5, G6, and G4dn instances across endpoint deployments, training jobs, and notebooks, with no quota increase request needed. Quota limits and utilization for each instance type are also visible right inside the Studio environment. For the details, check Service quotas for Studio.

The feature is available in all AWS Commercial Regions where Amazon SageMaker Studio is supported.

Wrap-up

  • One-click jump from Hugging Face into SageMaker Studio with the model pre-loaded
  • New users get an environment in seconds, with serverless customization permissions preconfigured
  • Default GPU access to G5, G6, and G4dn, no quota request required
  • A great fit for ML engineers who want to try Hugging Face models on AWS right away!

Related posts