tao-train-single-step
nvidiaTrain, evaluate, and export a TAO model in one run.
model training
What it does
Runs supervised TAO model training, optionally evaluates the result, and optionally exports it.
When to use it
Use for a single TAO training run without iterative augmentation, AutoML, or DEFT loops.
How to use it
Give it a TAO model, training dataset URI, and supported platform; it creates the confirmed training workflow and can include evaluation and export.
What you provide
- TAO model
- training dataset URI
- supported platform
Uses
Access · 4
This skill
TAO training container
TAO training
TAO evaluation
TAO evaluation
TAO export
Exports the TAO model
long-running monitoring
Monitors long-running jobs
What you need · 7
Docker must be available for the TAO workflow.
The NVIDIA container toolkit must be available for the TAO workflow.
A compatible TAO model such as clip, nvdinov2, or grounding_dino must be selected.
Cloud credentials or an external credential are required.
About this skill
- Visibility
- Public
- Repository
- nvidia/skills
- Created
- Oct 8, 2026
- Updated
- Oct 8, 2026
- Files
- 6