How to compile tensorflow for Jetson Orin Nano?

I googled around, and there is no guide on compiling tensorflow from source for Jetson Orin Nano.
And I’m trying to follow build guide on tensorflow website, but there is no luck.

When I installed binary from NVIDIA repo, it still has many warning/errors, which might be not good enough, especially for people like me to learn from the demo.

So is it possible for Jetson Orin Nano support team to give a guide for compiling tensorflow from source? Or any link that related to build tensorflow for Jetson platform is well appreciated.


TensorFlow is a third-party library so please check with them if their building guide is not working.

Is TensorFlow 2.16 essential for you? Any new feature?
If not, it’s more recommended to use our prebuilt so you don’t need to build it from the source.

The package is shared monthly.
So it’s also expected to have 2.16 shortly but no time is guaranteed.


Wowo… that’s good, as the binary is built by NVIDIA.
If NVIDIA can share how to build from source then it will be great!


We built it with similar steps listed in the TensorFlow document.

Yeah, right now.

I’m just don’t know how to get this inconsistency fixed. I think maybe those cross compile steps and runtime warnings might give a tip on the issue: Inconsistency of NVIDIA 2.15.0+nv24.03 v.s. Colab v.s. Tensorflow Documentation


Let’s close this topic.

Please see Tensorflow v2.16.1 GPU version local build on Jetson Orin Nano failed for the document issue and Inconsistency of NVIDIA 2.15.0+nv24.03 v.s. Colab v.s. Tensorflow Documentation for the functionality issue of v2-15-0-nv24-03.