Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add links to Deepnote #75

Open
wants to merge 3 commits into
base: master
Choose a base branch
from
Open

Conversation

jirkalhotka
Copy link

I prefer to work with Deepnote over NbViewer, because it allows you to interactively edit the files online and save your work. Added links to launch the notebooks in Deepnote in case anyone else wants to play around with the code.

@rasbt
Copy link
Owner

rasbt commented Jan 20, 2019

Thanks a lot, this looks cool. While some notebooks execute fine there are a couple of issues.

I.e., the first cell (watermark) does cause an error because watermark is not installed in the containers. I think this is expected because it's not a standard package in Anaconda, but still it kind of looks bad or confusing to users if the notebook already crashes on the first cell. Could watermark be added to the container (or optionally check whether watermark is installed and if not pip install it?)

The same goes for Theano in the last notebook, for example. I realize that no one is really using Theano anymore, but still I don't want to change the notebooks too much because otherwise they won't match the book anymore (in the 2nd edition, everything is based on TensorFlow).

As an option, is it possible to initialize a container for these notebooks that have specific package versions? This way we can guarantee that users get the exact same results as in the book, which could avoid confusion.

@jirkalhotka
Copy link
Author

Thanks a lot for the response and for the feedback!:)

I added all the used libraries (and a few more common data science ones actually) to Deepnote so that people can run all the notebooks including the watermark magic etc.

I think this should address the issues?

@rasbt
Copy link
Owner

rasbt commented Jan 23, 2019

Thanks!

A little formatting related issue:

screen shot 2019-01-22 at 10 41 22 pm

Does this one button not render properly because the line is too long, or is there something that can be fixed?

Also, I just tried to open some notebooks via the links shown above, but I bump into the following issue:

screen shot 2019-01-22 at 10 44 23 pm

screen shot 2019-01-22 at 10 44 30 pm

Are these temporary server issues or do the links need to be updated?

@jirkalhotka
Copy link
Author

Thanks a lot and sorry for the delayed response! The error was to do with changing the image to include all the necessary libraries, it's all fixed now.

Considering chapter 6, yes you're exactly right, the line is too long, so I didn't put the button but just a textual link instead. I tried to make the button work, but it would have to be so small it would be barely legible so I thought the textual link made more sense... now on a second thought, you could also drop the "learning" from the title and just call it "Best practices for...", then the button would fit and the formatting would be uniform. These are the two options that come to my mind. What do you think?

@rasbt
Copy link
Owner

rasbt commented Jan 26, 2019

Nice, thanks, seems to work now!

the "learning" from the title and just call it "Best practices for...",

I would maybe leave the "Learning" because that corresponds to the chapter title in the book. Maybe it would be best to change it to

Learning Best Practices for Model Eval. and Hyperparameter Optimization

or if necessary

Learning Best Practices for Model Eval. and Hyperparameter Optim.

@jirkalhotka
Copy link
Author

Better?:)

@jirkalhotka
Copy link
Author

@rasbt Hey Sebastian, how's it going? A lot of new work done on Deepnote, have a look, I think you might like it.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

2 participants