pkmital / pycadl

Python package with source code from the course "Creative Applications of Deep Learning w/ TensorFlow"

Introduction

This package is part of the Kadenze Academy program Creative Applications of Deep Learning w/ TensorFlow.

COURSE 1: Creative Applications of Deep Learning with TensorFlow I (Free to Audit)
Session 1: Introduction to TensorFlow
Session 2: Training A Network W/ TensorFlow
Session 3: Unsupervised And Supervised Learning
Session 4: Visualizing And Hallucinating Representations
Session 5: Generative Models

COURSE 2: Creative Applications of Deep Learning with TensorFlow II (Program exclusive)
Session 1: Cloud Computing, GPUs, Deploying
Session 2: Mixture Density Networks
Session 3: Modeling Attention with RNNs, DRAW
Session 4: Image-to-Image Translation with GANs

COURSE 3: Creative Applications of Deep Learning with TensorFlow III (Program exclusive)
Session 1: Modeling Music and Art: Google Brain’s Magenta Lab
Session 2: Modeling Language: Natural Language Processing
Session 3: Autoregressive Image Modeling w/ PixelCNN
Session 4: Modeling Audio w/ Wavenet and NSynth

Requirements

Python 3.5+

Installation

pip install cadl

Then in python, you can import any module like so:

from cadl import vaegan

Or see a list of possible modules in an interactive console by typing:

from cadl import and then pressing tab to see the list of available modules.

Documentation

cadl.readthedocs.io

Contents

This package contains various models, architectures, and building blocks covered in the Kadenze Academy program including:

• Autoencoders
• Character Level Recurrent Neural Network (CharRNN)
• Conditional Pixel CNN
• CycleGAN
• Deep Convolutional Generative Adversarial Networks (DCGAN)
• Deep Dream
• Deep Recurrent Attentive Writer (DRAW)
• Gated Convolution
• Generative Adversarial Networks (GAN)
• Global Vector Embeddings (GloVe)
• Illustration2Vec
• Inception
• Mixture Density Networks (MDN)
• PixelCNN
• NSynth
• Residual Networks
• Sequence2Seqeuence (Seq2Seq) w/ Attention (both bucketed and dynamic rnn variants available)
• Style Net
• Variational Autoencoders (VAE)
• Variational Autoencoding Generative Adversarial Networks (VAEGAN)
• Video Style Net
• VGG16
• WaveNet / Fast WaveNet Generation w/ Queues / WaveNet Autoencoder (NSynth)
• Word2Vec

and more. It also includes various datasets, preprocessing, batch generators, input pipelines, and plenty more for datasets such as:

• CELEB
• CIFAR
• Cornell
• MNIST
• TedLium
• LibriSpeech
• VCTK

and plenty of utilities for working with images, GIFs, sound (wave) files, MIDI, video, text, TensorFlow, TensorBoard, and their graphs.

Examples of each module's use can be found in the tests folder.

Contributing

Contributions, such as other model architectures, bug fixes, dataset handling, etc... are welcome and should be filed on the GitHub.

Troubleshooting

Error: alsa/asoundlib.h: No such file or directory

  src/RtMidi.cpp:1101:28: fatal error: alsa/asoundlib.h: No such file or directory
  compilation terminated.
  error: command 'gcc' failed with exit status 1

This is a dependency of magenta (python-rtmidi) which requires libasound.

Solution: Install ALSA

CentOS:

sudo yum install alsa-lib-devel alsa-utils

Ubuntu:

sudo apt-get install libasound2-dev

More info:

https://python-rtmidi.readthedocs.io/en/latest/installation.html magenta/magenta#781

Error: jack/jack.h: No such file or directory

src/RtMidi.cpp:2448:23: fatal error: jack/jack.h: No such file or directory
compilation terminated.

Solution: Install Jack

Ubuntu:

sudo apt install libjack-dev

More info:

https://python-rtmidi.readthedocs.io/en/latest/installation.html magenta/magenta#781

pycadl's Issues

cadl affects GPU usage in TensorFlow

After running "pip install cadl" in my TensorFlow-1.12-gpu docker container, TF can't find any GPU device to perform training. Before installing cadl, TF can use GPU. Pretty sure it's caused by cadl. Any solution for that?