Train vision models by rebuilding hidden patches.
What it does
Produces pretrained and fine-tuned visual models by learning to reconstruct masked image patches.
When to use it
Use for training, evaluation, export, or inference with a TAO MAE backbone.
How to use it
Give it image data or model checkpoints and a requested MAE task; it writes training, evaluation, inference, or export results to files and folders.
What you provide
- Your files
What you need · 5
Requires Docker for the container-based workflow.
Requires NVIDIA Container Toolkit alongside Docker.
States a minimum of two GPUs with at least 24GB VRAM per GPU; recommends eight GPUs and A100 hardware.
Must run first for a standalone installation when the session was not initialized by the TAO skill bank plugin; handles host preflight, credentials, and cross-skill discovery.
Access · 0
No additional actions listed.