Unable to install tensorflow 2.7 in Jetson Nano dev. B01

[wldnd32]

Hi there, We are currently planning to install tensorflow in jetsonano B01 kit
The jetpack installed on the jetson nano is 4.6.1 and after downloading Python 3.9.19 version to match Python support versions 3.7 to 3.9 in the site to install tensorflow 2.7, the problem occurred and the error changed with the following modification.

1. There was a problem with h5py.

2. After fixing it again by referring to other sites such as h5py, setuptools, etc., the installation was resumed, but the following error occurs.
   ERROR: could not find a version that satifties the requirement tensorflow==2.7.0+nv22.1 (from versions: 2.10.rc ~~~~~)

3. If you write the code without specifying a specific version of tensorflow, version 2.16.1 will be installed automatically and gpu will not run.

Q1. Please tell me how to solve this (point out what you expect to be a problem)

Q2. If you do not specify a specific version during the installation process, the packages are installed as the latest version. Tell me the version of the dependency packages that correspond to tensorflow 2.7 version!

Q2. The following link is found in your corresponding Nvidia (‘Official TensorFlow for Jetson Nano!’).
Currently, I would like to retry installing the tensorflow in the above way. But I have a question, as far as I know, the existing tensorflow 2.7 version does not support Python 3.6 version, is it possible to install it?

[AastaLLL]

Hi,

We don’t have a prebuilt for Python 3.7 or 3.9.
Only Python 3.6 is available.

For other versions, you will build it from the source or use a package without CUDA support.
Thanks.