Skip to content

popfido/ud730_deep_learning

Repository files navigation

Deep Learning

This repository is used to store my progress in Udacity's Deep Learning course.

1. Download and run Docker container

docker run -dit -v /PATH_TO_CONTAINER_DIR:/notebooks -p port:port tensorflow/tensorflow

Then open the browser and go to [docker_container_ip]:port to verify it is running alright.

Noticed that port is a detailed number like "8888" or some other ports you want to use.

Or you can also install Tensorflow in your local machine without in virtual container.

2. Update jupyter module

Run bash in docker container:

docker exec -ti [docker_id] bash

Then update the module:

pip install -U jupyter

Then stop jupyter server by running:

kill $(pgrep jupyter)

This will actively stop docker container, Then re-start it again:

docker start [docker_id]

The same procedure can be used to update other modules in python package.

If you are using Jupyter Notebook on localhost but not virtual machine, to update packages, you just need to running:

pip install -U package_name

3. Change the indentation to use 2 spaces

Tensorflow uses 2-space indentation instead of the default 4-space. It is annoying to run the code on this default setting so lets change our ipython notebook's config for indentation:

  • Open javascript console.
  • Run the following code:
var cell = Jupyter.notebook.get_selected_cell();
var config = cell.config;
var patch = {
          CodeCell:{
                      cm_config:{indentUnit:2}
                            }
                                }
                                config.update(patch)
                                ```

                                If it returned error `Uncaught TypeError: config.update is not a
                                function(…)` then ipython notebook needs to be updated (see step 2
                                above), then clear browser's cache.

About

UDacity UD730 Deep Learning by Google

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published