TensorFlow.jl installation problem



When I install TensorFlow.jl under Julia 0.6.1 for 64 bit Windows I get this error:

Fetching package metadata …
Solving package specifications: .

UnsatisfiableError: The following specifications were found to be in conflict:
** - python 2.7***
** - tensorflow 1.2.0* -> python 3.5***
Use "conda info " to see the dependencies for each package.

=================================================[ ERROR: TensorFlow ]==================================================

LoadError: failed process:


====================================================[ BUILD ERRORS ]====================================================

WARNING: TensorFlow had build errors.

  • packages with build errors remain installed in C:\Users\arbenede.julia\v0.6
  • build the package(s) and all dependencies with Pkg.build("TensorFlow")
  • build a single package by running its deps/build.jl script


Any ideas?




TensorFlow.jl does not currently support windows: https://github.com/malmaud/TensorFlow.jl/issues/204


I tried Windows Subsystem for Linux and apparently Tensorflow does not support that either… time to get an Intel NUC and put Ubuntu on it!


I was really hopeful that TensorFlow.jl would support Linux Subsystem for Windows.

Can you please raise an issue with full details on the github repo?


Hi all,

I’m using JuliaPro- on Windows 10.

I’ve received “TensorFlow.jl doesn’t support Windows” using Julia>Pkg.add(“TensorFlow”).
Then I’ve used Julia>Pkg.build(“TensorFlow”). But I’ve received the same error.

Similar problem for installing MXNet.jl.

Any idea.




I encountered the same problem. I used flux.jl at last which is extremely easy to install. TensorFlow.jl should be a very useful package. It’s a pity that it does not support Windows.


As you might suspect the reason you get that error is because TensorFlow.jl doesn’t support windows.

I spent a day trying. You can find the PR.
It segfaults for basic operations.
I don’t have the time to try and debug weird issues with libtensorflow.DLL

Anyone is welcome to give it a shot and make a PR.
It should be trivial, in theory.
Just download the windows library binary.