Having Fun with Recurrent Neural Networks (RNNs)

(This repo contains all the files you need to replicate the Tutorial I've given on the PyData 2017 conference =D)

Description: https://pydata.org/sanluis2017/schedule/presentation/3/

Static Slides: http://tworld.io/extra/pydata/recurrently-happy-rnn.slides.html

Make sure you have installed the following before opening the notebook.

SciPy Stack
TensorFlow >= 1.0: (Tested on version 1.3.0. - GPU version recommended)
RISE Slideshow Extension: This notebook is meant to be shown as an Slideshow, install RISE using these commands:

~$ pip install RISE
~$ jupyter-nbextension install rise --py --sys-prefix
~$ jupyter-nbextension enable rise --py --sys-prefix

ABC player: to be able to play ABC songs, we'll use abcmidi and timidity commands. Try to install them on ubuntu by:

~$ sudo apt install abcmidi
~$ sudo apt-get install timidity timidity-interfaces-extra

Download the dataset from here: dataset.zip
Download the pretrained models from here: trained_models.zip
Extract those .zip files inside the notebook directory, so that you end up with something like this:

recurrently-happy-rnn
    ├──dataset/
    ├──trained_models/
    ├──images/
    ├──recurrently-happy-rnn.ipynb
    ...

Run the Jupyter notebook inside this directory (otherwise, images won't be displayed).

And that's it. Enjoy, have fun and happy hacking! :D

Provide feedback