Install Tensorflow on Windows Docker installation
I am trying to install Tensorflow on my Windows PC using the Docker intallation (https://www.tensorflow.org/versions/r0.8/get_started/os_setup.html#docker–installation). After the installation of Docker, I have followed these instructions: http://www.netinstructions.com/how-to-install-and-run-tensorflow-on-a-windows-pc/ .
Everything works well ultil the last step: when I type the instruction
docker run -it b.gcr.io/tensorflow/tensorflow
I obtain the following output:
[I 10:10:57.073 NotebookApp] Writing notebook server al/share/jupyter/runtime/notebook_cookie_secret [W 10:10:57.104 NotebookApp] WARNING: The notebook s addresses and not using encryption. This is not rec [W 10:10:57.105 NotebookApp] WARNING: The notebook s addresses and not using authentication. This is hig nded. [I 10:10:57.110 NotebookApp] Serving notebooks from [I 10:10:57.110 NotebookApp] 0 active kernels [I 10:10:57.111 NotebookApp] The Jupyter Notebook is addresses on your system]:8888/ [I 10:10:57.111 NotebookApp] Use Control-C to stop t kernels (twice to skip confirmation).
I obtain the same message typing the command line both in the docker terminal and in the command window. Nothing else happens, and I am obliged to use Control-C to stop.
I have tryed to run also the IMAGE ID of
b.gcr.io/tensorflow/tensorflow and I obtain exactly the same thing.
Any idea of what is happening?
Thank you very much!
One Solution collect form web for “Install Tensorflow on Windows Docker installation”
From the log messages I can see, it looks like the Docker image has changed since this tutorial was written, so that it now starts a Jupyter (IPython) server in the container. To actually use this, you will need to configure port forwarding to your Windows host, using a command like:
$ docker run -it -p 8888:8888 gcr.io/tensorflow/tensorflow
From your comment, it looks like this command does not succeed because something is already listening on port 8888 on your computer. If you can’t stop whatever is already listening on port 8888, you could try a different port (e.g. 8080) as follows:
$ docker run -it -p 8080:8888 gcr.io/tensorflow/tensorflow
…then connect a web browser to