How to install Tensorflow on Python 2.7 on Windows?

Question

I am trying to install tensorflow via pip but getting a notification

could not find a version satisfying the tensor flow requirement

Anyway, I could install it via pip

Answer 1

We are developing a bit on dirty_feri, because it is not completely updated.

Tensorflow for Windows is only supported with Python 3.5 and Python 3.6 (since version 1.2). As you download via pip, you will get the latest version 1.2 so you can work on 3.6.

No need to use the Anaconda version of Tensorflow, the distribution is not supported, and if you use anaconda, the pip version is just fine.

If you still require python 2.7 support for other projects, I can suggest using an environment manager like anaconda or virtualenv so you can use multiple versions of python at once.

After installing a supported version of python, you can run pip install tensorflow and install it in a few minutes.

Full installation instructions are available here: https://www.tensorflow.org/install/install_windows

Answer 2

If you use windows:

If you pick up the gang on the TensorFlow website in the Windows PIP window, specify the first line.

"Installing the gun in Windows

TensorFlow only supports 64-bit Python 3.5 on Windows. We tested pip packages with the following Python distributions: "

Now either install python 3.5 or use the unofficial version of Tensorflow from ANACONDA .

Another way is to download and install the docker toolbox for windows https://www.docker.com/docker-toolbox Open the cmd window and type: docker run -it b.gcr.io/tensorflow/tensorflow This should lead to a linux shell . Type python and I think everything will be fine!

Answer 3

If you only need TensorFlow because of Keras and you are on Python 2.7.x, you can avoid installing Tensorflow (Google) and replace it with CNTK (Microsoft). According to Jeong-Yoon Lee CNTK is much (about 2-4 times) faster than TensorFlow for LSTM (bidirectional LSTM for data and text IMDb Generation via LSTM), while speeds for other types of neural networks are close to each other. Your Keras code does not need to be changed (I tested it with two Keras examples using TensorFlow and successfully replaced TensorFlow with CNTK without changing anything Keras code.

So how do you install it?

-CCPU version of CNTK:

pip install https://cntk.ai/PythonWheel/CPU-Only/cntk-2.4-cp27-cp27m-win_amd64.whl

-GPU version of CNTK:

pip install https://cntk.ai/PythonWheel/GPU/cntk-2.4-cp27-cp27m-win_amd64.whl

-Test CNTK install:

python -c "import cntk; print (cntk. version )"

-Install Keras: Python Deep Learning Library

pip install keras

-Enable CNTK as Keras back end iso TensorFlow

Modify the keras.json file in the% USERPROFILE% / section. keras

 { "epsilon": 1e-07, "image_data_format": "channels_last", "backend": "cntk", "floatx": "float32" }