MACHINE LEARNING / huggingface
SageMaker Serving Image Selection
Choose a region-compatible serving container from the AWS catalogue and inspect architecture, runtime and mirroring requirements.
“Pick the right serving container for a SageMaker model deployment and find its current image URI. Use this skill whenever about to deploy a model”
01 / THE REASONING
Why this made the selection.
- It picks a SageMaker serving image and prefers a compatible Hugging Face curated family over a newer generic engine tag.
- TGI is rejected as archived, and SM_VLLM_TRUST_REMOTE_CODE stays false unless a pinned architecture needs repository code.
- Distinguishes encoder rerankers from generative causal-LM rerankers before choosing a serving family.
- Provides a concrete private-ECR mirroring helper and calls out runtime model-code permissions.
02 / THE REVIEW RECORD
What we actually inspected.
Source review has boundaries.
A clear record is more useful than a “safe” badge.
Material inspected
- skills/hf-cloud-serving-image-selection/SKILL.md
- skills/hf-cloud-serving-image-selection/scripts/mirror_image.py
- README.md
- LICENSE
Our findings
- Rule zero says a compatible Hugging Face curated image wins over a generic family even when the generic tag is newer.
- The skill says not to use TGI because it is archived.
- SM_VLLM_TRUST_REMOTE_CODE defaults to false and is set true only for a reviewed architecture that cannot load otherwise.
- A slow first boot is not a failure, and the skill does not tell the operator to tear the endpoint down for that delay.
- mirror_image.py obtains identity/region from the active shell and uses password-stdin for registry login.
- It avoids implicit latest tags but uses tag-existence as its idempotence check and does not verify a source/destination digest match.
Not established by this review
- The resource was not executed and its tools were not called.
- No skill execution, authenticated service operation, or outcome benchmark was performed.
The review applies to the material and revision named here. A newer upstream release can change its behavior.
03 / PUT IT TO WORK
Add a skill to your workflow.
DOCUMENTED COMMAND
npx skills add huggingface/skills/skills/hf-cloud-serving-image-selection Run in your terminal to choose your AI tools, project or global scope, and installation method. This short command follows the current upstream version; the review covers the revision shown above.
Optional: install the reviewed revision
Keeps the original source and installer version where recorded, while leaving your AI tools and installation scope for you to choose.
npx skills add https://github.com/huggingface/skills/tree/c3ff942be3e07116b475253f0a8fc4f54682c1a8/skills/hf-cloud-serving-image-selection - Install the pinned skill directory interactively, retaining its bundled references and scripts.
- Read model architecture/task and the current AWS image/region catalogue before selecting a URI.
- If mirroring is needed, inspect AWS identity/region and source digest; validate the resulting image before deployment.
Before you start
- Target model configuration and AWS account/region
- Docker daemon and AWS CLI for mirroring
04 / FOLLOW THE EVIDENCE
The source trail.
Our notes are separate from the original resource.
Check upstream before adopting a new version.
- Pinned skill definition ↗Checked
https://github.com/huggingface/skills/blob/c3ff942be3e07116b475253f0a8fc4f54682c1a8/skills/hf-cloud-serving-image-selection/SKILL.md
Supports: summary, whySelected, bestFor, limitations, install, access, review
https://github.com/huggingface/skills/blob/c3ff942be3e07116b475253f0a8fc4f54682c1a8/README.md
Supports: install
- Apache-2.0 licence ↗Checked
https://github.com/huggingface/skills/blob/c3ff942be3e07116b475253f0a8fc4f54682c1a8/LICENSE
Supports: license
https://github.com/huggingface/skills/blob/c3ff942be3e07116b475253f0a8fc4f54682c1a8/skills/hf-cloud-serving-image-selection/scripts/mirror_image.py
Supports: review, limitations, access, install