About NGC containers with xavier

Is it possible to use

nvcr.io/nvidia/tensorrt:xx.xx-py3

or

nvcr.io/nvidia/tensorflow:xx.xx-tfx-py3

NGC containers with xavier?

These are for amd64 architecture containers. You can search with “l4t” tag from here:
https://ngc.nvidia.com/catalog/containers?orderBy=scoreDESC&pageNumber=0&query=l4t&quickFilter=containers&filters=

I think, you can use

nvcr.io/nvidia/l4t-base:rxx.x.x

instead of nvcr.io/nvidia/tensorrt:xx.xx-py3

and

nvcr.io/nvidia/l4t-tensorflow:rxx.x.x-tfx.x-py3

instead of nvcr.io/nvidia/tensorflow:xx.xx-tfx-py3

I have never tried it before yet, but these should work.
Hope it helps.

That’s correct, you should use the l4t-* containers on Jetson. Also the TensorRT/CUDA/cuDNN are already mounted into l4t-base.

The l4t-pytorch, l4t-tensorflow, and l4t-ml containers are built from here if you wish to customize them: GitHub - dusty-nv/jetson-containers: Machine Learning Containers for NVIDIA Jetson and JetPack-L4T