I do not have internet access on my linux computer therefore I installed TF from source by following TensorFlow Get Started.
I ran into a few trouble to build trainer_example due to the lack of internet connection hopefully someone from tensorflow helped me through it by creating local repositories for re2, gemmlowp, jpegsrc v9a, libpng and six and modifying WORKSPACE accordingly.
When I try to bazel build pip_package to create the wheel then I think I run into the same problem but :
-the list of repositories is insanely long (to manually install each of them) even if they seem to be mostly part of PolymerElements
Is there an easy workaround ?
If you are happy to create a PIP package without TensorBoard, you should be able to avoid rewriting the Polymer dependencies by removing this line ("//tensorflow/tensorbaord" in the build_pip_package dependencies) from tensorflow/tools/pip_package/BUILD.
Related
I have an action in a private repo that does pip install -e, where one of the setup.py requirements is tf-nightly. It seems that this causes all the past builds to be downloaded and then actions says No space left on device. What is the proper fix here?
The solution seems to have been that I had a numpy requirement of 1.18. I am not sure if it was downloading until it found a compatible version, or it just caused some other bug. Making the requirement simply numpy resolved the issue.
I'm currently working on a program to play a game similar to atari-games. I'm using keras (python 3). I finished writing the code and I want to test it, and I have few questions about the process:
first of all, I have trouble importing tesnorflow for some reason. I've installed it using pip. I've made sure to created new env. before the installation (which finished successfully), but when I try to run my program it says:
ModuleNotFoundError: No module named 'tensorflow'
I also, tried to install the package from within pycharm, but then I get this error:
Could not find a version that satisfies the requirement tensorflow (from versions: )
No matching distribution found for tensorflow
I've checked program requirements (such as pip, python, virtualenv and setuptools versions) and everything seems up to date. perhaps someone could point out what else might be the problem?
Is there any other way I can test the performance of my program?
Thank you very much for your time and attention.
Anaconda is a complete time-saver. I suggest create an enviornment using Anaconda and install the tensorflow by conda install tensorflow If you would like to use the gpu version, conda automatically installs the CUDA and cudnn for you too.
the redist(x86) is different from those answers provided before(x64).Is that the point?
Meanwhile it shows "Unless you are using bazel, you should try to import tensorflow from its source directory".But i didn't do that.
Ensure that you have Visual Studio Installed and that MSVCP140.DLL is on your computer AND in your PATH envVar. You shouldn't be building from source on Windows as it is not supported. I assume that you installed via pip?
Check for the DLL, VS, your PATH variables and, if that doesn't work, redownload the vc_redist_x64 version.
Failing that, uninstall Tensorflow and try again with pip install tensorflow or from a nightly. I had a similar issue and starting fresh seemed to help.
I am looking into using Tensorflow for my research soon, and looked at the online documentation for installing with Conda https://www.tensorflow.org/versions/r0.11/get_started/os_setup.html#anaconda-installation.
It suggested creating a new environment, and installing Tensorflow in it, and the installing other python packages afterwards.
But I already have an existing environment with lots of packages I need, and I'm wondering if its safe to add Tensorflow into that environment?
Also, I have a question about how this installation with conda works. I know that Conda will create a distinct set of folders containing the libraries needed for each environment, but if I install Tensorflow, what happens to all the base low level C++ and CUDA libraries that get compiled? Do they reside in my Conda environment's folder, or are they in some system wide libraries closer to my root?
PS: I'm using Ubuntu 16.04, and have a GPU that I want to run Tensorflow on.
Thank you.
But I already have an existing environment with lots of packages I need, and I'm wondering if its safe to add Tensorflow into that environment?
conda has this awesome feature called "revisions". You can show your current environment with
conda list --revisions
which allows you to revert changes to your conda environment. This allows you to install new packages with confidence that if something breaks you can always revert it later. See this page for more info: https://www.continuum.io/blog/developer/advanced-features-conda-part-2. tl;dr: conda install --revisions <revision_number>
what happens to all the base low level C++ and CUDA libraries that get compiled
Are you talking about the libraries that get compiled when you are trying to run your code? Or the C++/CUDA libraries? If you're talking about the C++/CUDA libs then conda is not compiling them, but merely installing a pre-compiled binary into a specific location that gets picked up. If you're talking about your code, then where exactly those files live would seem to depend on where you put them.
To install QGIS on my MacBook Pro I need to install the gdal framework; however, the 1.11 framework package needed for QGIS is an empty file at kyngchaos. I tried installing GDAL 1.10 complete, but QGIS is requiring 1.11. I have Mavericks. I've installed each individual package: UnixImageIO, PROJ, GEOS, SQLite2, numpy, rgdal, and on... Now when I run the GDAL 1.11 framework install, everything seems to run until the last page, where an "Install Fail" page shows up. When I try to view the error, I'm taken out of the installer.
After some googling, I tried installing using homebrew. (brew install gdal)... I got a little further, but it tells me to "brew link libpng libtiff" -- and when I run that command, I get:
Linking /usr/local/Cellar/libpng/1.5.14... Warning: Could not link libpng. Unlinking...
Error: Could not symlink file: /usr/local/Cellar/libpng/1.5.14/share/man/man3/libpngpf.3
/usr/local/share/man/man3 is not writable. You should change its permissions.
After some more searching, it seems I could change permissions from usr/local to me, but I'm not sure how to (exactly) and don't want to mess anything up. Any help would be greatly appreciated, and my apologies if I'm missing something painfully obvious! I'm a novice at the programming end of things, so am just kind of pushing through everything like a bull in a china shop.
UPDATE:
Okay, I found the answer, even though it didn't seem to be working initially--
I ran the following commands from the GDAL help documentation:
export PATH=/Library/Frameworks/GDAL.framework/Programs:$PATH
sudo ln -sfh [ver] /Library/Frameworks/GDAL.framework/Versions/Current
in Terminal, and the framework was updated. If I run the GDAL 1.11 installer it still shows up as a failed installation, BUT QGIS recognizes 1.11 as the installed, so that's great.
(I just needed to install matplotlib in addition, and QGIS was installed successfully.)
It seems that the GDAL Complete framework was just updated 3 days ago, so it should be a temporary error until they realize the package is empty.