/WGAN-Project

Primary LanguageJupyter Notebook

WGAN-Project

Code for NYU CDS Intro to Machine Learning, Fall 2018, final project.

Investigating performance of GANs using various loss functions.

To run: Obtain edges2handbags dataset or facades dataset and place in "handbags" or "facades" directory. Update DRIVE_BASE_DIR as appropriate for your file system. (As currently configured, code expects inputs to be in "handbags/500_bags" containing a 500-item subsample or "facades/train" with the full data set.)

Set desired parameters per command flags (or, if running in Colaboratory, through the ARGSTRING variable) and run. Output will be placed in a directory with a name describing the parameters used to produce it.