4

I am deploying a django project using heroku cloud platform. I have added the dependencies in requirements.txt file. However when I push to heroku master, I get the following error:

Collecting tensorflow==1.0.0 (from -r /tmp/build_bc8be989466414998410d3ef4c97a115/requirements.txt (line 17))
remote:          Could not find a version that satisfies the requirement tensorflow==1.0.0 (from -r /tmp/build_bc8be989466414998410d3ef4c97a115/requirements.txt (line 17)) (from versions: )
remote:        No matching distribution found for tensorflow==1.0.0 (from -r /tmp/build_bc8be989466414998410d3ef4c97a115/requirements.txt (line 17))
remote:  !     Push rejected, failed to compile Python app.
remote: 
remote:  !     Push failed
remote: Verifying deploy....
remote: 
remote: !   Push rejected to what-the-image.
remote:

I am using Django v 1.10 and python 2.7. Where would I be going wrong?

Deesha
  • 538
  • 8
  • 27

1 Answers1

3

You will be able to install Tensorflow on Heroku by using a wheel.

Simply replace the tensorflow==1.0.0 line in your requirements.txt with https://storage.googleapis.com/tensorflow/linux/cpu/tensorflow-1.0.0-cp27-none-linux_x86_64.whl

Wheels and eggs are packaging formats for Python code. Wheels are intended to replace the older egg format and generally much more versatile as they don't require a compiler to be available (very useful when deploying to PaaS like Heroku, Microsoft's Azure).

One thing to note about wheels is the naming convention, which reflects the architecture and Python version they are meant to be used on. A quick way to find the type of wheel your system supports is via:

import pip
print(pip.pep425tags.get_supported())
Community
  • 1
  • 1
Adrian Ghiuta
  • 1,569
  • 16
  • 29