search

val-iisc / deligan

This project is an implementation of the Generative Adversarial Network proposed in our CVPR 2017 paper - DeLiGAN : Generative Adversarial Networks for Diverse and Limited Data. DeLiGAN is a simple but effective modification of the GAN framework and aims to improve performance on datasets which are diverse yet small in size.
MIT License
110 stars 28 forks source link

Results in dg_mnist.py #4

Closed LinkWoong closed 6 years ago

LinkWoong commented 6 years ago

I've tried the mnist model. However, I only got some noises.....any comments?

swami1995 commented 6 years ago

Hi, I'm sorry for taking this long to reply. I'm frankly speaking not sure what's causing this in your case. I just ran these experiments multiple times and it is converging to produce meaningful generations. One way to test if everything at your end is working properly is to check if increasing the dataset size gradually and training produces better results. Hope that helps.