0

Installation method:

I'm using the Anaconda distribution of Python instead of having multiple versions of python on my computer. I used the instructions under TensorFlow with Anaconda (link1)(link2) with the following commands:

C:> conda create -n tensorflow python=3.6

C:> activate tensorflow

(tensorflow)C:> pip install --ignore-installed --upgrade tensorflow

Error:

When running the test hello world code from within a tensorflow environment I received the following errors:

>>> import tensorflow as tf
>>> hello = tf.constant('Hello, TensorFlow!')
>>> sess = tf.Session()
2018-01-23 02:44:09.201798: I C:\tf_jenkins\home\workspace\rel-win\M\windows\PY\36\tensorflow\core\platform\cpu_feature_guard.cc:137] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX AVX2
>>> print(sess.run(hello))
b'Hello, TensorFlow!'

Questions:

  1. Does this mean my CPU does not support Tensorflow? (i7-6500U, 2.59GHz)

  2. Does the b' signify an environment output, or is this an error?

  3. I noticed the TensorFlow library doesn't appear in my CMD prompt version of python, nor in my Spyder executable. Should I use pip and install a second version of the library? Or does TensorFlow require an active environment to invoke the library?

Edit: I just noticed this line in a re-read:

In Anaconda, you may use conda to create a virtual environment. However, within Anaconda, we recommend installing TensorFlow with the pip install command, not with the conda install command.

Antares2018
  • 57
  • 1
  • 6
  • #2 (b') is answered here https://stackoverflow.com/questions/6269765/what-does-the-b-character-do-in-front-of-a-string-literal – Jeremy Voisey Jan 22 '18 at 18:27

2 Answers2

1

if you have installed tensforflow 1.8 for cpu correctly in win10 (python3.5.x) and you have an error, try to change version to 1.5 pip3 install tensorflow==1.5 I have spend one day to know it :)

tinyProgrammer
  • 41
  • 2
0

You should be able to run tensorflow just fine with that installation. However, you can install a specific version of tensorflow that was compiled to include instruction sets that will make the computation faster that your processor has access to.

Read this guide to find out how to build form source and improve your performance: https://www.tensorflow.org/install/install_sources

or feel free to continue using the installation you have now.

Grant Williams
  • 1,469
  • 1
  • 12
  • 24
  • Thanks Grant. I was trying to run a small python script to show the installed version history of the libraries but TensorFlow, Theano and Keras don't appear. It's as if I need a separate environment running for all three to see those libraries. I think my mistake is wanting to rely on one version of Python - the Anaconda distribution and wanting to run it through the default PATH environment. I'll reinstall the packages and start again tomorrow. – Antares2018 Jan 22 '18 at 16:27
  • I'd highly using a conda environment named tensorflow or something similar to maintain all of your libraries and dependencies. Edit: although it looks like you might already be doing that? – Grant Williams Jan 22 '18 at 16:33
  • Hi Grant, That's what I've done. I'm running everything through Anaconda now. – Antares2018 Jan 27 '18 at 12:26