you didn’t help me. You gave me this link : GitHub - dusty-nv/jetson-containers: Machine Learning Containers for NVIDIA Jetson and JetPack-L4T ; but here there are multiple docker files. I would like to get the link of the right one. I dont want to use docker commit. I want to add the packages I need internally to the dockerfile. It seems to be this : jetson-containers/Dockerfile.pytorch at master · dusty-nv/jetson-containers · GitHub ; but inside I read “ARG BASE_IMAGE=nvcr.io/nvidia/l4t-base:r32.4.4” ; so it is not the right one,since I need the version 32.5,not the 32.4.4.
@marietto2008
It appears that jetson/ARM infrastructure maintenance is to a large extent a self effort that couldn’t be helped externally .)
Nvidia is said to be more like a gaming company
However, x86_64 has more support available
Yes the PyTorch dockerfile is Dockerfile.pytorch.
The BASE_IMAGE gets set in scripts/docker_build_ml.sh to the correct l4t-base for whatever your detected JetPack version is. You don’t need to set it manually if you follow the build directions in the readme.