Python notebooks on deep generative models for image and text generations for PyData Eindhoven November 2019.
The Image Generation directory contains the notebooks for image generation using VAE and GAN trained on Fashion MNIST dataset by ZalandoResearch.
References for the Image Generation Part can be found in the slides and the notebooks.
The Text Generation directory contains the notebooks for text generation based on RNN and LSTM trained on Shakespeare and Sherlock Holmes text dataset. The text data can be found in the datasets directory. Several checkpoints are included to skip the training step. Load the appropriate checkpoint and generate conditional or unconditional texts in Shakespeare style.
- The Unreasonable Effectiveness of Recurrent Neural Networks
- Understanding LSTM Networks
- Illustrated Guide to LSTM’s and GRU’s: A step by step explanation