Skip to content

Instantly share code, notes, and snippets.

Embed
What would you like to do?
take notes for studying tensorflow

Recommand: Cuda 8.0, cudnn 6.0

1. why cannot I use tensorboard?

  • before start tensorboard, make sure add following codes:

file_writer = tf.summary.FileWriter('log_path', sess.graph)

  • then start tensorboadrd:

tensorboard --logdir log_path

  • when you see a promote site, copy it and open in browser.

2. why cannot see the changes in tensorboard(using jupyter)?

  • you need use following codes first when run new changes

file_writer = tf.summary.FileWriter('log_path', sess.graph)

3. tensorflow not found in pip

  • Maybe you have a 32-bit Python ,you need a 64-bit version of Python. As for now Tensorflow only support Python 3.4, 3.5 64-bit version. refer to: stackoverflow
-- @rocket1037 @Thierry Lathuille

4. pip install jupyter or start jupyter meet srror

  • if you use windows, the reason maybe encoding, you can using following step: c:\program files\python36\lib\site-packages\pip\compat_init_.py:
    return s.decode('utf_8') -> return s.decode('cp936') or
    using other terminal liake Git Bash to install jupyter

5. Annoying log message !

  • tf.logging.set_verbosity(tf.logging.ERROR)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment