在Windows上安装TensorFlow（Python 3.6.x）

Question

I'm trying to install TensorFlow on Windows .

I tried to install it with pip , but I always get the same error message:

... is not a supported wheel on this platform.

I first tried it with Python 3.5.1, now I upgraded to 3.6.0b4 , but it makes no difference.

---

Python:

Python 3.6.0b4 (default, Nov 22 2016, 05:30:12) [MSC v.1900 64 bit (AMD64)] on win32

pip:

pip 9.0.1 from ...\python\lib\site-packages (python 3.6)

---

To be exact, I tried the following two commands:

pip install --upgrade https://storage.googleapis.com/tensorflow/windows/cpu/tensorflow-0.12.0rc0-cp35-cp35m-win_amd64.whl
pip install --upgrade https://storage.googleapis.com/tensorflow/windows/gpu/tensorflow_gpu-0.12.0rc0-cp35-cp35m-win_amd64.whl

they output the following:

> tensorflow-0.12.0rc0-cp35-cp35m-win_amd64.whl is not a supported wheel on this platform.
> tensorflow_gpu-0.12.0rc0-cp35-cp35m-win_amd64.whl is not a supported wheel on this platform.

Does anyone know how to solve this problem? I'm not sure where I'm making a mistake.

Thanks!

---

Edit 1

Btw, I also tried pip install tensorflow and pip install tensorflow-gpu like suggested here . I got the following output:

 > Could not find a version that satisfies the requirement tensorflow (from versions: ) No matching distribution found for tensorflow > Could not find a version that satisfies the requirement tensorflow-gpu (from versions: ) No matching distribution found for tensorflow-gpu 

Answer 1

Update 15.11.2017

It seems that by now it is working like one would expect. Running the following commands using the following pip and python version should work.

---

Installing with Python 3.6.x

---

Version

Python : 3.6.3
pip : 9.0.1

---

Installation Commands

The following commands are based of the following installation guide here .

using cmd

C:> pip3 install --upgrade tensorflow // cpu
C:> pip3 install --upgrade tensorflow-gpu // gpu

using Anaconda

C:> conda create -n tensorflow python=3.5 
C:> activate tensorflow
(tensorflow)C:> pip install --ignore-installed --upgrade tensorflow
(tensorflow)C:> pip install --ignore-installed --upgrade tensorflow-gpu 

Additional Information

A list of common installation problems can be found here .

You can find an example console output of a successful tensorflow cpu installation here .

---

Old response :

Okay to conclude; use version 3.5.2 !
Neither 3.5.1 nor 3.6.x seem to work at the moment.

Versions:

Python 3.5.2 pip 8.1.1 .. (python 3.5)

Commands:

 // cpu C:> pip install --upgrade https://storage.googleapis.com/tensorflow/windows/cpu/tensorflow-0.12.0rc0-cp35-cp35m-win_amd64.whl // gpu C:> pip install --upgrade https://storage.googleapis.com/tensorflow/windows/gpu/tensorflow_gpu-0.12.0rc0-cp35-cp35m-win_amd64.whl 

Answer 2

If you are using anaconda distribution, you can do the following to use python 3.5 on the new environnement "tensorflow":

conda create --name tensorflow python=3.5
activate tensorflow
conda install jupyter
conda install scipy
pip install tensorflow
# or
# pip install tensorflow-gpu

It is important to add python=3.5 at the end of the first line, because it will install Python 3.5.

Source: https://github.com/tensorflow/tensorflow/issues/6999#issuecomment-278459224

Answer 3

After searching a lot and trying to install and reinstall Python, i found the solution was very simple

use the following for windows

python -m pip install --upgrade https://storage.googleapis.com/tensorflow/mac/cpu/tensorflow-0.12.0-py3-none-any.whl

change to following on mac

python3 -m pip install --upgrade https://storage.googleapis.com/tensorflow/mac/cpu/tensorflow-0.12.0-py3-none-any.whl

for Anaconda use corresponding conda

Answer 4

Tensorflow now works with python 3.6.4, don't use python builds that came after 3.6.4 and most importantly tensorflow doesn't work on Python 3.7.2. You can download 3.6.4 for windows from here python 3.6.4 .

Then just simply run pip install tensorflow-gpu

Answer 5

At the time of this writing, there is no official support for TensorFlow with Python 3.6 on Windows. The recommendation is to build TensorFlow yourself.

Some people have already done this and provide *.whl files that you can directly install with pip. These are unofficial, so use at your own risk:

• http://www.lfd.uci.edu/%7Egohlke/pythonlibs/#tensorflow

You can simply download them and install them with pip install <filename>.whl .

See also this GitHub comment .

Answer 6

I was having Python 3.6 and was facing the issue as " No module named tensorflow " on "pip install tensorflow". Turned out as my machine was of 64 bit while the Python 3.6 version installed was for 32 bit . Uninstalled it, reinstalled the Python 3.6 x64 version, pip installed tensorflow, problem solved.

Answer 7

Tensorflow is now supported on Python 3.6. Just make sure that the Python installation is 64 bit on a 64 bit machine and that pip is the latest ( pip install --upgrade pip ).

After that ( pip install --upgrade tensorflow ) works like a charm.

Answer 8

Tensorflow indeed supports Python 3.6.X version, but only for 64-bit architecture. Here is the link where you can download Python 3.6.X 64-bit version

Answer 9

For Pip installation on Windows and 64-bit Python 3.5:

CPU only version:

C:\> pip install --upgrade https://storage.googleapis.com/tensorflow/windows/cpu/tensorflow-0.12.0rc0-cp35-cp35m-win_amd64.whl

For the GPU version:

C:\> pip install --upgrade https://storage.googleapis.com/tensorflow/windows/gpu/tensorflow_gpu-0.12.0rc0-cp35-cp35m-win_amd64.whl

References: Tensorflow installation issue on windows "tensorflow_gpu-0.12.0rc0-cp35-cp35m-win_amd64.whl is not a supported wheel on th is platform." .

https://www.tensorflow.org/versions/r0.12/get_started/os_setup.html

Also see tensorflow not found in pip .

Answer 10

Tensorflow in Now supporting Python 3.6.0 .....I have successfully installed the Tensorflow for Python 3.6.0
Using this Simple Instruction // pip install -- tensorflow

[enter image description here][1]
[1]: https://i.stack.imgur.com/1Y3kf.png

Installing collected packages: protobuf, html5lib, bleach, markdown, tensorflow-tensorboard, tensorflow
Successfully installed bleach-1.5.0 html5lib-0.9999999 markdown-2.6.9 protobuf-3.4.0 tensorflow-1.3.0 tensorflow-tensorboard-0.1.5

Answer 11

Same issue with you .

py3.6x win10 pro x64 pip 9.0.1

Issue message
"Could not find a version that satisfies the requirement tensorflow (from versions: ) No matching distribution found for tensorflow"

Maybe you can try py3.5

Answer 12

Windows batch file for installing TensorFlow and Python 3.5 on Windows. The issue is that as of this date, TensorFlow is not updated to support Python 3.6+ and will not install. Additionally, many systems have an incompatible version of Python. This batch file should create a compatible environment without effecting other Python installs. See REM comments for assumptions.

REM download Anaconda3-4.2.0-Windows-x86_64.exe (contains python 3.5) from https://repo.continuum.io/archive/index.html
REM Assumes download is in %USERPROFILE%\Downloads
%USERPROFILE%\Downloads\Anaconda3-4.2.0-Windows-x86_64.exe

REM change path to use Anaconda3 (python 3.5).
PATH %USERPROFILE%\Anaconda3;%USERPROFILE%\Anaconda3\Scripts;%USERPROFILE%\Anaconda3\Library\bin;%PATH%

REM update pip to 9.0 or later (mandatory)
python -m pip install --upgrade pip

REM tell conda where to load tensorflow
conda config --add channels conda-forge

REM elevate command (mandatory) and install tensorflow - use explicit path to conda %USERPROFILE%\Anaconda3\scripts\conda
powershell.exe -Command start-process -verb runas cmd {/K "%USERPROFILE%\Anaconda3\scripts\conda install tensorflow"}

Be sure the above PATH is used when invoking TensorFlow.

Answer 13

Follow these steps to install ternsorflow :

(step 1)   conda create -n py35 python=3.5
(step 2)   activate py35
(step 3)   conda create -n tensorflow
(step 4,only for GPU)   pip install --ignore-installed --upgrade https://storage.googleapis.com/tensorflow/windows/gpu/tensorflow_gpu-1.1.0-cp35-cp35m-win_amd64.whl

Answer 14

Just found Tensorflow 1.1 for python 3.6 on windows x64 (including GPU version, but i tested only cpu): http://www.lfd.uci.edu/~gohlke/pythonlibs/#tensorflow . Unofficial apparently, but worked for me when I import tensorflow or tflearn in my code. They have scipy for windows there and bunch of other packages.

For some reason pip install URL returns code 404, so the installation would be the following:

1) Download protobuf whl package from here: http://www.lfd.uci.edu/~gohlke/pythonlibs/vu0h7y4r/protobuf-3.3.0-py3-none-any.whl

2) pip install {DownloadFolder}\\protobuf-3.3.0-py3-none-any.whl

3) Download TF whl file: http://www.lfd.uci.edu/~gohlke/pythonlibs/vu0h7y4r/tensorflow-1.1.0-cp36-cp36m-win_amd64.whl

4) pip install {DownloadFolder}\\tensorflow-1.1.0-cp36-cp36m-win_amd64.whl

It worked for me.

Answer 15

For someone w/ TF 1.3:

Current TensorFlow 1.3 support Python 3.6 , and then you need cuDNN 6 (cudnn64_6.dll)

Based on Tensorflow on windows - ImportError: DLL load failed: The specified module could not be found and this: https://github.com/tensorflow/tensorflow/issues/7705

Answer 16

On 2/22/18, when I tried the official recommendation:

pip3 install --upgrade tensorflow

I got this error

Could not find a version that satisfies the requirement tensorflow

But instead using

pip install --upgrade tensorflow

installed it ok. (I ran it from the ps command prompt.)

I have 64-bit windows 10, 64-bit python 3.6.3, and pip3 version 9.0.1.

Answer 17

Tensor flow on 32 bit machine.

There is no official build out for 32 bit, but still there is a workaround for that, follow the link http://cudamusing.blogspot.in/2015/11/building-tensorflow-for-jetson-tk1.html . I would not suggest doing this big reason is its not possible to follow the process every time there is a change in official tensor flow code.

Answer 18

I had the same issue, but I followed the following steps:-

1. I had Python 3.6.5 (32 bit) installed on my desktop, but from all the research I did, I could conclude that Tensorflow runs only on Python 3.5x or 3.6x 64 bit versions. So I uninstalled it and Installed Python 3.5.0 instead.
2. I ran Python 3.5.0 as an administrator. This step is necessary for Windows as, without it, the system doesn't get any privileges and can't install tensorflow.
3. Install Pip3 using command:- python -m pip install --upgrade pip
4. Once the newest version (10.0.1 in my case) is installed, you can install tensorflow usin command:- pip3 install --upgrade tensorflow
5. Your tensorflow will be downloaded and installed. For further help on how to run tensorflow programs, go to https://www.tensorflow.org/get_started/premade_estimators

Answer 19

https://i.stack.imgur.com/1Y3kf.png

pip install -- tensorflow This worked for me for this version of python Python 3.6.4 : : Anaconda, Inc.

Answer 20

Tensorflow is not compatible with python3.7 and spyder3.3.1

To work with stable tensorflow version

follow the procedure

windows-->search-->Anaconda prompt-->right click -->click Run as adminstrator

Below command create the virtual environment which does not disturb existing projects

conda create -n projectname 

Below command activates your virtual environment within this directory installed package will not disturb your existing project.

activate projectname

Below command installs python 3.6.7 and spyder 3.2.3 as well

conda install spyder=3.2.3

Below mentioned tensorflow version works without any error. As per your need, you can install tensorflow version specifically.

pip install tensorflow==1.3.0

To open spyder

spyder

To exit form Virtual environment

deactivate

Answer 21

For the GPU version, see my answer here .

In short, install Anaconda, then open an anaconda terminal and type conda create --name tf_gpu tensorflow-gpu

You are then free to use the conda environment tf_gpu

Answer 22

Tensor flow has only support for python 2.7 3.4 3.5 Other python versions are not supported So you please install the supported python version and try it again. The official link is https://www.tensorflow.org/install/install_linux#InstallingAnaconda It provides how to install it with anaconda . This will help you