TensorFlow

Категория:Работа

Начало

https://www.tensorflow.org/

Текущая версия v1.5.0

Docker: https://hub.docker.com/r/tensorflow/tensorflow/

Using TensorFlow via Docker https://github.com/tensorflow/tensorflow/blob/master/tensorflow/tools/docker/README.md

Загрузка

Для CPU:

docker pull gcr.io/tensorflow/tensorflow

Запуск

Для CPU:

docker run -it -p 8888:8888 gcr.io/tensorflow/tensorflow

Для запуска с возможностью просмотра tensorboard добавить порт 6006:

docker run -it -з 6006:6006 -p 8888:8888 gcr.io/tensorflow/tensorflow

$ docker run -it -p 8888:8888 gcr.io/tensorflow/tensorflow 
[I 17:30:28.871 NotebookApp] Writing notebook server cookie secret to /root/.local/share/jupyter/runtime/notebook_cookie_secret
[W 17:30:28.898 NotebookApp] WARNING: The notebook server is listening on all IP addresses and not using encryption. This is not recommended.
[I 17:30:28.903 NotebookApp] Serving notebooks from local directory: /notebooks
[I 17:30:28.904 NotebookApp] 0 active kernels
[I 17:30:28.904 NotebookApp] The Jupyter Notebook is running at:
[I 17:30:28.904 NotebookApp] http://[all ip addresses on your system]:8888/?token=9fc3be91c196e8fe264d384f0a07aaba99d589c1288067a5
[I 17:30:28.904 NotebookApp] Use Control-C to stop this server and shut down all kernels (twice to skip confirmation).
[C 17:30:28.904 NotebookApp] 
    
    Copy/paste this URL into your browser when you connect for the first time,
    to login with a token:
        http://localhost:8888/?token=9fc3be91c196e8fe264d384f0a07aaba99d589c1288067a5

При запуске будут доступны 3 обширных урока на английском о том, как работать с этой системой

Запуск TensorBoard

docker exec -ti 6523620b7eb2 bash
# tensorboard --logdir=/tmp/tensorflow/mnist/logs/

где 6523620b7eb2 - id контейнера tensorflow

Интерфейс tensorboard будет доступен по адресу http://127.0.0.1:6006/

Материалы

https://habrahabr.ru/post/326650/

https://habrahabr.ru/post/305578/

https://habrahabr.ru/company/ods/blog/324898/