/VAE

Simple implementation of variational auto-encoder using tensorflow.

Primary LanguageJupyter NotebookApache License 2.0Apache-2.0

VAE

  • run vae_jaan.ipynb and see results in ./log/
  • The version of tensorflow is 1.1.0 (under which I run it).
  • To run the model, just focus on ``vae_jaan.py''. Set the directory parameters and run it.
  • ``vae.py'' is an uncomplished version but vae_jaan.py is enough and simple.
  • vae_jaan.py is a simplified version of JAAN ALTOSAAR's.
  • I removed some fancy part of the code and made it more straightforward. The code only includes very basic, easy-understood tf's api. There are not complex ones (which I don't understand and are roughly documented).