It's my first time, but I'll write it quickly.
I'm a windows10 user, but I want to run tensorflow, so I'm trying various things. It's not working yet. I am taking a Udacity Deep Learning course (https://www.udacity.com/course/deep-learning--ud730) and my goal is to be able to do that task. I will write with emphasis on that.
I proceeded according to the following site, and now it works with iPython for the time being. I use docker. (Additional note at the bottom of the page. The procedure has been changed.) https://medium.com/@Rapchik/running-google-s-deep-learning-course-material-under-windows-82d468b6d5be#.6pk8esr9i
However, the vagrant specification seems to have changed, and I couldn't find the insecure_private_key that appears on this site. This was resolved by adding the following description to the sentence that this site says to add to the Vagrantfile.
config.ssh.insert_key = false
As a result, it can be operated by the following procedure.
vagrant up
.cd tensorflow / tensorflow / examples / udacity /
on PuTTYdocker run -p 8888:8888 -it google_udacity
I managed to open it and tensorflow seems to work, so it was good for the time being. I can't save my python notebook. I looked at the following site and tried it, but it did not show an error, but it was not saved. .. .. Will it sync with some folder when it works? https://github.com/tensorflow/tensorflow/blob/master/tensorflow/examples/udacity/README.md I omitted the --pull option because it didn't respond. Is it useless? docker I'm not sure. .. ..
Also, if you execute the source code prepared in Exercise 1 of Udacity,
the kernel appears to have died. It will restart automatically.
I can't read the notMNIST dataset because I die. It doesn't work.
I would like to know if anyone has succeeded.
---- Addendum ---- Something went well when I struggled. I was surprised to see rumors that it can only be shared under / c / Users. I referred to the following site and changed the method in various ways. I don't know docker so I don't know what's going on. .. .. http://qiita.com/Algebra_nobu/items/6c2b05e2a9a4a5ee456d
docker-machine env default
, then the command ʻeval $ (path to docker-machine env default)` that appears on the screen.docker run -p 8888: 8888 -p 6006: 6006 --name tekitou -v // c / Users / Folder you want to share: / notebooks -it b.gcr.io/tensorflow/tensorflow
The folder you want to share is me Is ~ / tensorflow / examples / udacity
.docker start tekitou
docker exec -it tekitou bash
Also, now that files can be shared, files that were too large to be decompressed on iPython can now be decompressed and then sent to iPython. I wonder if this can be done. Expectations.
---- Addendum ---- (2016/12/29)
tensorflow is now compatible with windows!
I feel that I just did pip install tensorflow
on the spyder command line.
Recommended Posts