issues on running DIGITS 6.0 on Ubuntu 14.04; error message relates to numpy version

Hello,

currently I am trying to get DIGITS up and running. I installed it the way it is described here:
https://github.com/NVIDIA/DIGITS/blob/digits-6.0/docs/BuildDigits.md

Installation works fine, but when I start digits-devserver following error message appears:

RuntimeError: module compiled against API version 0xc but this version of numpy is 0xa
Traceback (most recent call last):
  File "/usr/lib/python2.7/runpy.py", line 162, in _run_module_as_main
    "__main__", fname, loader, pkg_name)
  File "/usr/lib/python2.7/runpy.py", line 72, in _run_code
    exec code in run_globals
  File "/home/si/digits/digits/__main__.py", line 70, in <module>
    main()
  File "/home/si/digits/digits/__main__.py", line 55, in main
    import digits.webapp
  File "digits/webapp.py", line 49, in <module>
    import digits.views  # noqa
  File "digits/views.py", line 18, in <module>
    from digits import dataset, extensions, model, utils, pretrained_model
  File "digits/extensions/__init__.py", line 5, in <module>
    from .view import *  # noqa
  File "digits/extensions/view/__init__.py", line 9, in <module>
    from . import imageSegmentation
  File "digits/extensions/view/imageSegmentation/__init__.py", line 4, in <module>
    from .view import Visualization
  File "digits/extensions/view/imageSegmentation/view.py", line 11, in <module>
    import skfmm
  File "/usr/local/lib/python2.7/dist-packages/skfmm/__init__.py", line 51, in <module>
    from .pfmm import distance, travel_time, extension_velocities
  File "/usr/local/lib/python2.7/dist-packages/skfmm/pfmm.py", line 4, in <module>
    from .cfmm import cFastMarcher
ImportError: numpy.core.multiarray failed to import
Exception KeyError: KeyError(139853259722416,) in <module 'threading' from '/usr/lib/python2.7/threading.pyc'> ignored

I understand, that the installation has a problem with the numpy version, therefore I searched, which version is installed on my system. I found different answers for this:

In directory /usr/local/lib/python2.7/dist-packages there are three folders naming:

  • numpy
  • numpy-1.11.0-py2.7.egg-info
  • numpy-1.14.0-py2.7.egg-info

In console,

pip show numpy

delivers

Version: 1.14.0

in the directory aforementioned.

Though, when I input in console:

python
>>> import numpy
>>> numpy.__version__
'1.11.0'
>>> numpy.__path__
['/usr/local/lib/python2.7/dist-packages/numpy']

How can I configure the setup, that it uses the numpy version needed? In the caffe requirements numpy version 1.11.0 is maximum requested. In digits requirements numpy version 1.14.0 is requested.

I also tried installing digits via docker (as explained here: https://github.com/NVIDIA/DIGITS/blob/digits-6.0/docs/DockerInstall.md). But then the files for the “Get Started”-Tutorial are nowhere to find.

I hope for your great support. For further information, don’t hesitate to ask.

Greets
Manfred

I have the same problem.

We created a new “Deep Learning Training and Inference” section in Devtalk to improve the experience for deep learning and accelerated computing, and HPC users:
https://devtalk.nvidia.com/default/board/301/deep-learning-training-and-inference-/

We are moving active deep learning threads to the new section.

URLs for topics will not change with the re-categorization. So your bookmarks and links will continue to work as earlier.

-Siddharth

A video walkthrough of natively installing NVIDIA DIGITS on Ubuntu 18.04 LTS is available here:

https://cudaeducation.com/nvidiadigits/

-Cuda Education
cudaeducation.com