Cannot install TensorFlow, Pytorch natively

According to what has been announced in this website: https://devblogs.nvidia.com/jetson-nano-ai-computing/, “The SDK also includes the ability to natively install popular open source Machine Learning (ML) frameworks such as TensorFlow, PyTorch, Caffe, Keras, and MXNet, along with frameworks for computer vision and robotics development like OpenCV and ROS.”

I have already flashed my TX2 with the new Jetpack 4.2 SDK. It’s cool but cannot install these frameworks natively. I’m wondering is this functionality only valid for Jetson Nano?

Regarding TensorFlow, we validated the 2019/03/19 release of TensorFlow for JetPack 4.2 with Jetson AGX Xavier and with Jetson Nano. Unfortunately there was not enough time to test this release with Jetson TX2. We have a monthly cadence though, and I expect it will be there for the April release.

The description for JP 4.2 version here:
https://developer.nvidia.com/embedded/downloads#?search=TensorFlow%20for%20JetPack
will be updated shortly to make this clear.

Thanks for the information! i will try it later;

So what about pytorch and Mxnet? Is there also a corresponding release for them?

Hi,

We have tested this official TensorFlow package on TX2:
https://devtalk.nvidia.com/default/topic/1048776/jetson-nano/official-tensorflow-for-jetson-nano-/

It seems to work without error, but hasn’t been officially validated for TX2.
Thanks.