The Author Online Book Forums are Moving

The Author Online Book Forums will soon redirect to Manning's liveBook and liveVideo. All book forum content will migrate to liveBook's discussion forum and all video forum content will migrate to liveVideo. Log in to liveBook or liveVideo with your Manning credentials to join the discussion!

Thank you for your engagement in the AoF over the years! We look forward to offering you a more enhanced forum experience.

Hmm, it would seem like there's a version mismatch between Keras and NumPy, or possibly Theano and NumPy. Given that Theano has stopped development since the release of this course, I would recommend switching to TensorFlow as the backend.
Oh man, my editor workflow setup is my favorite thing. I run `tmux`, with ViM on the right and IPython on the left, each in its own pane. Then, using a ViM plugin called `slimux`, I can issue a single keyboard command to send the current ViM line to another tmux pane, namely the one running IPython. It's so incredibly fluid. All the interactive power of IPython with all the features of my favorite text editor.

If you want to learn more about this setup, I have both an ebook (free, at, as well as a tutorial on Youtube ( I highly recommend it for exploratory data science and machine learning.

If you want this kind of setup yourself, I have a script for ubuntu that install all the necessary things from a clean distro. Check it out on github at Make sure to checkout branch `ubuntu`, and run ``. Happy computing!

While the videos are nominally in Keras 1.0, the code is available for both 1.0 and 2.0. Keras 2.0 is very friendly in that if you use a deprecated feature or name, it will usually warn you without crashing. This makes updating your code a lot less painful.

The decision to publish as 1.0 came down mostly to timing, but also to enterprise viewers who may not be able to upgrade to 2.0 yet. Machine Learning software, as a rule, moves very quickly, and Keras is no exception. Future editions of this course will almost certainly be Keras 2 (unless Keras 3 comes out first!).

If you prefer to use Python 2, you can probably get away with it. There may be some subtle differences, mostly in the `print` function and some of the list comprehensions. TensorFlow/Theano and all dependencies should all be available for Python 2 as well, so that shouldn't be a problem. But once you install, it should be mostly the same.

That said, I developed the course using Python 3, so that's the only "officially" supported version. But if you run into specific problems where code doesn't work with Python 2, feel free to post on this forum and I'll take a look.
Course Author here. All the files written or used in the course should be available with it or pointed to online. So thanks for pointing out if something seems to be missing. I know I get frustrated with algorithms that don't provide example data, so I appreciate where you're coming from. I think Greg has provided the link, so should address this concern. If it doesn't, please let us know and we'll rectify it.

-Dan Van Boxel
Course Author here. The videos use Keras 1 for 2 reasons. The biggest one is that I finished recording them about 1 week before Keras 2.0 came out. The secondary reason is that many users of Keras (especially from large institutions) are still using old Keras 1 code.

While the videos are for Keras 1, all provided code files are available in both Keras 1 and 2. If those are not in the course, I'll talk to the publisher to make sure they get listed.

As a bonus, there's an extra video that walks through the process of updating your Keras code from 1 to 2. It doesn't cover everything, but it tries to get the areas likely to come up.

-Dan Van Boxel