Currently only support celebsa dataset.
Able to run on
python => 3.7.7
tensorflow => 2.2.0
tensorflow-examples => latest version
Require arguments:
-d : path to the dataset
Optional arguments(default):
--gpu : used to enable gpu mode (no set)
--lmbda : lambda constant (10)
--nsamples : number of samples to be taken from dataset (1000)
--niters : number of iterations (200)
--width : width of images to be trained on (256)
Step 1: python train.py -d ~/dataset
play around with network size