Xavier Jetpack 4.2 TensorFlow - some issues in the installation - pip3 gets broken

First of all, I have TensorFlow running, and the TensorFlow example https://www.tensorflow.org/tutorials using the Keras MNIST example runs just fine. So far so good…

In the course of following the instructions in https://docs.nvidia.com/deeplearning/dgx/install-tf-xavier/index.html I ran into what is apparently not an uncommon problem: the pip3 installation got broken, and I got the ImportError: cannot import name ‘main’ error described here: https://github.com/pypa/pip/issues/5599

I was able to complete the installation using “python3 -m pip …” rather than “pip3 …”. The pip3 command is no longer usable and apparently cannot be repaired. But, that isn’t a biggie.

I’m curious about why this particular problem is left in the TensorFlow installation. I can’t be the only one smacking into this.

Hi,

This is known issue of pip3.

Please apply the change here:
https://devtalk.nvidia.com/default/topic/1051564/jetson-nano/how-to-install-the-sdk-and-tensorflow-/post/5337358/#5337358

Thanks.