21

I'm using Tensorflow-0.8 on Ubuntu14.04. I first install Tensorflow from sources and then setup Tensorflow for development according to the official tutorial. When I want to uninstall tensorflow using the following command

sudo pip uninstall tensorflow

I encountered the following error:

Can't uninstall 'tensorflow'. No files were found to uninstall

Could anyone tell me where is wrong?

For your reference, the output of pip show tensorflow is

Name: tensorflow
Version: 0.8.0
Location: /home/AIJ/tensorflow/_python_build
Requires: numpy, six, protobuf, wheel

But I actually find another Tensorflow directory at 

/usr/local/lib/python2.7/dist-packages/tensorflow

Besides, I also have a question about the general usage of Python. I have seen two quite similar directories in my system, i.e.

/usr/lib/python2.7/dist-packages
/usr/local/lib/python2.7/dist-packages

Could any one tell me the differences between them? I noticed that everytime I use sudo pip install <package>, the package will be installed to /usr/local/lib/python2.7/dist-packages, could I instead install packages into /usr/lib/python2.7/dist-packages using pip install?

Thanks a lot for your help in advance!

Improve this question
0

2 Answers 2

Reset to default
10

It could be because you didn't install Tensorflow using pip, but using python setup.py develop instead as your link shows.

pip uninstall is likely to fail if the package is installed using python setup.py install as they do not leave behind metadata to determine what files were installed.

Therefore, you should be able to unistall Tensorflow with the option -u or --unistall of develop

cd /home/AIJ/tensorflow/_python_build
python setup.py develop --uninstall

To answer for the second (interestring) question about the two dist-package created under /usr/lib/python2.7 and /usr/local/lib/python2.7 it exists already a great Stack Overflow answer on the topic.

PS: Tensorflow is a good library, you should consider not uninstall it :)

Improve this answer
Sign up to request clarification or add additional context in comments.

6 Comments

Haven't tried but looks like this should be the correct answer. @OP - can you check and comment/upvote if this works
@Max That's exactly where the problem lays. I successfully uninstalled Tensorflow in your way (and then reinstall it in virtualenv:) ). Thanks a lot for offering the solution and giving a good reference for my second question!
@ROBOT AI, My pleasure! I learn a lot of things myself while writing the answer. You said you want to install Tensorflow invirtualenvand there is nothing wrong with that. imo, I prefer Docker. It could be more relevant for you as well.
None of these answers helped me in my Anaconda virtualenv. So what I did was rm -rf tensor* from the site-packages directory.
@GokulNC I don't think rm -rf is a good strategy: because you may still leave metadata somewhere else in you environment thus creating conflicts later on. I just read in package remove from Anaconda's doc that conda uninstall tensorflow should just work. If you confirm this (e.g. with another package) I will update my answer accordingly.
|
7

I believe pip isn't installed for python2.7

try :

pip -V

On my system for instance it says :

pip 8.1.2 from /usr/lib/python3.4/site-packages (python 3.4)

So basically using pip uninstall will only remove packages for python3.4 (and not python2.7).

So I don't use pip binary as such, and rather, call pip module from inside python.

In your case :

python2.7 -m pip uninstall tensorflow
Improve this answer

1 Comment

Thanks for your kind reply! The results of pip -V is pip 1.5.4 from /usr/lib/python2.7/dist-packages (python 2.7). Then I guess pip may have been installed for python2.7, isn't it? Then I also try python2.7 -m pip uninstall tensorflow, but the same error remains.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.