Read tensorflow version failed

AI & Data Science Deep Learning (Training & Inference) Frameworks
1

Python 3.6 script failed to read tensorflow version，I installed tensorflow version 2.1.0 on Xavier platform.
log message is :

import tensorflow as tf
2020-07-03 09:43:24.986556: I tensorflow/stream_executor/platform/default/dso_loader.cc:48] Successfully opened dynamic library libcudart.so.10.2
2020-07-03 09:43:27.494229: I tensorflow/stream_executor/platform/default/dso_loader.cc:48] Successfully opened dynamic library libnvinfer.so.7
2020-07-03 09:43:27.496811: I tensorflow/stream_executor/platform/default/dso_loader.cc:48] Successfully opened dynamic library libnvinfer_plugin.so.7
tf.version
Traceback (most recent call last):
File “”, line 1, in
File “/usr/local/lib/python3.6/dist-packages/tensorflow_core/python/util/module_wrapper.py”, line 192, in getattr
attr = getattr(self._tfmw_wrapped_module, name)
AttributeError: module ‘tensorflow’ has no attribute ‘version’
tf.path
[‘/usr/local/lib/python3.6/dist-packages/tensorflow_core/python/keras/api/_v1’, ‘/usr/local/lib/python3.6/dist-packages/tensorflow_estimator/python/estimator/api/_v1’, ‘/usr/local/lib/python3.6/dist-packages/tensorflow_core’, ‘/usr/local/lib/python3.6/dist-packages/tensorflow_core/_api/v1’]

1 Like
2

Because there are some issues with the pre build version of tf. It seams the that the version is build with difrent parameters. Therefor you can access the version via: tf.version.VERSION.

You can do the following stuff to fix it:

  1. Build your own: Tensorflow 2.x on Jetson nano - #19 by mictiemix
  2. Wait that Nvidia fix the issue.

BR