Turn your HuggingFace vision model into a tested TAO pipeline.
What it does
Builds a native TAO training, evaluation, and TensorRT deployment pipeline for a HuggingFace computer vision model, with container-based validation.
When to use it
Use when a user wants a HuggingFace computer vision model integrated into TAO, including requests that supply a model URL or ID.
How to use it
Give it a HuggingFace model ID or URL, TAO repositories, and Toolkit image references; it implements and validates a local training and deployment pipeline.
What you provide
- HuggingFace model ID or URL
- TAO Toolkit image references
- An existing project
Uses
Good to know
- sends a credential to a server: NGC API key
What you need · 9
Requires Python 3.10+; Phase 0 verifies the version, while Python work runs inside Docker containers.
Requires an NVIDIA driver, with its host check delegated during Phase 0.
Access · 3
This skill
nvcr.io
Pulls TAO container images
HuggingFace
Inspects your HuggingFace model