simplestGAN
Simplest GAN implementation from scratch (using NumPy) demonstrated on the following problem:
We are given a matrix of dimension 2 x 2. Values of matrix are in range [0, 1], where 0 is white, and 1 is a black color.
We start from a noisy distribution, and the goal is to generate an image where on the main diagonal we have black color, on side diagonal white color.
Important files
- training.py - for the process of training
- main.py - to get the results from random vector z
Results
From noisy images (random distribution):
To the desired results (wanted distribution):
