TensorFlow installation: 'Native pip' on Windows

Question

I am desperately trying to install TensorFlow (and Keras) on a Windows 10 machine (64bit). I was able to have a working Keras installation on a Ubuntu Linux, but now need to have it working on Windows.

I am following the official instructions here: https://www.tensorflow.org/install/install_windows specifically those at "Installing with native pip".

First, I downloaded python 3.5 from python.org.

D:\>python --version
Python 3.5.3

Then I run pip3 install --upgrade tensorflow-gpu. CUDA 8, cuDNN 5.1 and all the other requirements are satisfied.

Now if I ask some details:

D:\>pip3 show tensorflow-gpu
Name: tensorflow-gpu
Version: 1.0.0
Summary: TensorFlow helps the tensors flow
Home-page: http://tensorflow.org/
Author: Google Inc.
Author-email: opensource@google.com
License: Apache 2.0
Location: d:\python\lib\site-packages
Requires: wheel, numpy, protobuf, six

If I test it:

D:\>python
Python 3.5.3 (v3.5.3:1880cb95a742, Jan 16 2017, 16:02:32) [MSC v.1900 64 bit (AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorflow as tf
>>> 

I have no error messages as a consequence of the import. But then:

>>> hello = tf.constant('Hello, TensorFlow!')
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
AttributeError: module 'tensorflow' has no attribute 'constant'

I have read the answers to other questions here on StackOverflow but could not find any pointer to a solution. In some cases, people had the same problem when using a file named tensorflow.py, but I checked this and could not find any script with that filename.

It was very easy under Linux and I thought it could not be this hard on Windows, but I was wrong.

I followed the instructions... what did I miss?

---

Answers to comments.

dir(tf)?

Note I am not running a script but using the prompt - nothing changes if I use a script.

d:\>python
Python 3.5.3 (v3.5.3:1880cb95a742, Jan 16 2017, 16:02:32) [MSC v.1900 64 bit (AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.

>>> import tensorflow as tf

>>> dir(tf)
['_LazyContribLoader', '__builtins__', '__cached__', '__doc__', '__file__',
'__loader__', '__name__', '__package__', '__path__', '__spec__', 
'absolute_import', 'contrib', 'division', 'print_function', 'python']

>>> tf.__file__
'D:\\python\\lib\\site-packages\\tensorflow\\__init__.py'

---

Update

I uninstalled tensorflow with the following command:

d:\pip3 uninstall tensorflow-gpu
[...]
d:\python\lib\site-packages\tensorflow_gpu-1.0.0.dist-info\wheel
d:\python\scripts\tensorboard.exe
Proceed (y/n)? y
Successfully uninstalled tensorflow-gpu-1.0.0

Now if I import tensorflow, I receive an error:

D:\>python
Python 3.5.3 (v3.5.3:1880cb95a742, Jan 16 2017, 16:02:32) [MSC v.1900 64 bit     
(AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorflow as tf
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  ImportError: No module named 'tensorflow'
>>>

So this behaviour should exclude the fact that another tensorflow file was causing the error I described. Am I right?

---

Next, I installed the module tensorflow-gpu again:

D:\>pip3 install --upgrade tensorflow-gpu
Collecting tensorflow-gpu
  Using cached tensorflow_gpu-1.0.0-cp35-cp35m-win_amd64.whl
Requirement already up-to-date: six>=1.10.0 in d:\python\lib\site-packages (from tensorflow-gpu)
Requirement already up-to-date: protobuf>=3.1.0 in d:\python\lib\site-packages (from tensorflow-gpu)
Requirement already up-to-date: numpy>=1.11.0 in d:\python\lib\site-packages (from tensorflow-gpu)
Requirement already up-to-date: wheel>=0.26 in d:\python\lib\site-packages (from tensorflow-gpu)
Requirement already up-to-date: setuptools in d:\python\lib\site-packages (from protobuf>=3.1.0->tensorflow-gpu)
Requirement already up-to-date: appdirs>=1.4.0 in d:\python\lib\site-packages (from setuptools->protobuf>=3.1.0->tensorflow-gpu)
Requirement already up-to-date: packaging>=16.8 in d:\python\lib\site-packages (from setuptools->protobuf>=3.1.0->tensorflow-gpu)
Requirement already up-to-date: pyparsing in d:\python\lib\site-packages (from packaging>=16.8->setuptools->protobuf>=3.1.0->tensorflow-gpu)
Installing collected packages: tensorflow-gpu
Successfully installed tensorflow-gpu-1.0.0

D:\>python
Python 3.5.3 (v3.5.3:1880cb95a742, Jan 16 2017, 16:02:32) [MSC v.1900 64 bit (AMD64)] on win32
Type "help", "copyright", "credits" or "license" for more information.

>>> import tensorflow as tf
I c:\tf_jenkins\home\workspace\release-win\device\gpu\os\windows\tensorflow\stream_executor\dso_loader.cc:135] successfully opened CUDA library cublas64_80.dll locally
I c:\tf_jenkins\home\workspace\release-win\device\gpu\os\windows\tensorflow\stream_executor\dso_loader.cc:126] Couldn't open CUDA library cudnn64_5.dll
I c:\tf_jenkins\home\workspace\release-win\device\gpu\os\windows\tensorflow\stream_executor\cuda\cuda_dnn.cc:3517] Unable to load cuDNN DSO
I c:\tf_jenkins\home\workspace\release-win\device\gpu\os\windows\tensorflow\stream_executor\dso_loader.cc:135] successfully opened CUDA library cufft64_80.dll locally
I c:\tf_jenkins\home\workspace\release-win\device\gpu\os\windows\tensorflow\stream_executor\dso_loader.cc:135] successfully opened CUDA library nvcuda.dll locally
I c:\tf_jenkins\home\workspace\release-win\device\gpu\os\windows\tensorflow\stream_executor\dso_loader.cc:135] successfully opened CUDA library curand64_80.dll locally
>>> hello = tf.constant('Hello, TensorFlow!')
>>> sess = tf.Session()
I c:\tf_jenkins\home\workspace\release-win\device\gpu\os\windows\tensorflow\core\common_runtime\gpu\gpu_device.cc:885] Found device 0 with properties:
name: GeForce GTX 970
major: 5 minor: 2 memoryClockRate (GHz) 1.228
pciBusID 0000:01:00.0
Total memory: 4.00GiB
Free memory: 3.31GiB
I c:\tf_jenkins\home\workspace\release-win\device\gpu\os\windows\tensorflow\core\common_runtime\gpu\gpu_device.cc:906] DMA: 0
I c:\tf_jenkins\home\workspace\release-win\device\gpu\os\windows\tensorflow\core\common_runtime\gpu\gpu_device.cc:916] 0:   Y
I c:\tf_jenkins\home\workspace\release-win\device\gpu\os\windows\tensorflow\core\common_runtime\gpu\gpu_device.cc:975] Creating TensorFlow device (/gpu:0) -> (device: 0, name: GeForce GTX 970, pci bus id: 0000:01:00.0)
>>> print(sess.run(hello))
E c:\tf_jenkins\home\workspace\release-win\device\gpu\os\windows\tensorflow\core\framework\op_kernel.cc:943] OpKernel ('op: "BestSplits" device_type: "CPU"') for unknown op: BestSplits
E c:\tf_jenkins\home\workspace\release-win\device\gpu\os\windows\tensorflow\core\framework\op_kernel.cc:943] OpKernel ('op: "CountExtremelyRandomStats" device_type: "CPU"') for unknown op: CountExtremelyRandomStats
E c:\tf_jenkins\home\workspace\release-win\device\gpu\os\windows\tensorflow\core\framework\op_kernel.cc:943] OpKernel ('op: "FinishedNodes" device_type: "CPU"') for unknown op: FinishedNodes
E c:\tf_jenkins\home\workspace\release-win\device\gpu\os\windows\tensorflow\core\framework\op_kernel.cc:943] OpKernel ('op: "GrowTree" device_type: "CPU"') for unknown op: GrowTree
E c:\tf_jenkins\home\workspace\release-win\device\gpu\os\windows\tensorflow\core\framework\op_kernel.cc:943] OpKernel ('op: "ReinterpretStringToFloat" device_type: "CPU"') for unknown op: ReinterpretStringToFloat
E c:\tf_jenkins\home\workspace\release-win\device\gpu\os\windows\tensorflow\core\framework\op_kernel.cc:943] OpKernel ('op: "SampleInputs" device_type: "CPU"') for unknown op: SampleInputs
E c:\tf_jenkins\home\workspace\release-win\device\gpu\os\windows\tensorflow\core\framework\op_kernel.cc:943] OpKernel ('op: "ScatterAddNdim" device_type: "CPU"') for unknown op: ScatterAddNdim
E c:\tf_jenkins\home\workspace\release-win\device\gpu\os\windows\tensorflow\core\framework\op_kernel.cc:943] OpKernel ('op: "TopNInsert" device_type: "CPU"') for unknown op: TopNInsert
E c:\tf_jenkins\home\workspace\release-win\device\gpu\os\windows\tensorflow\core\framework\op_kernel.cc:943] OpKernel ('op: "TopNRemove" device_type: "CPU"') for unknown op: TopNRemove
E c:\tf_jenkins\home\workspace\release-win\device\gpu\os\windows\tensorflow\core\framework\op_kernel.cc:943] OpKernel ('op: "TreePredictions" device_type: "CPU"') for unknown op: TreePredictions
E c:\tf_jenkins\home\workspace\release-win\device\gpu\os\windows\tensorflow\core\framework\op_kernel.cc:943] OpKernel ('op: "UpdateFertileSlots" device_type: "CPU"') for unknown op: UpdateFertileSlots
b'Hello, TensorFlow!'

So, now it seems to be working. The entire procedure was:

• install required software and libraries
• install python 3.5 from python.org
• install tensorflow-gpu natively (using pip3)
• uninstall tensorflow-gpu
• install tensorflow-gpu

May be a python expert can understand what happened...

---

PS. Yesterday I tried to install it using Anaconda, but had problems in subsequent steps when trying to install the Keras required packages. On Linux I did not have any problem with Keras either.

Answer 1

The solution was to uninstall and re-install tensorflow-gpu. Now it seems to be working.

It does not find cuDNN, but at least the test script runs.

Answer 2

I had the same problem. To solve it, I made a brand new installation of python 3.5.2 & tensorfow (pip3 install --upgrade tensorflow). I then write a helloMundo.py with the following content:

import tensorflow as tf
# Simple hello world using TensorFlow
# The value returned by the constructor represents the output
# of the Constant op.
hello = tf.constant('Hello, TensorFlow gr!')
# Start tf session
sess = tf.Session()
# Run the op
print(sess.run(hello))} 

I open the helloMundo.py from the python IDLE, then I open the menu Run -> Run Module (or F5) and it works.

Try to put your code in a .py file.

Regards, F. Kubha

Answer 3

This might sound odd, but are any of your folders or script named tensorflow?

tensorflow/script.py

or

dir/tensorflow.py

try changing those to not tensorflow and try again.
If that did not resolve things let me know

reasoning behind this and why I asked to see the output of dir(tf):

Here is the order in which a module will be searched when importing:

1. The directory containing the input script (or the current directory when no file is specified).

2. PYTHONPATH (a list of directory names, with the same syntax as the shell variable PATH).

3. The installation-dependent default.