This is a Generative Adversarial Network I made using Tensorflow, the neural network was trained on the CIFAR100 Dataset so that it can generate images that are similar to the dataset, the images generated during training are found in the images folder, however the images that looked best in my opinion are located in the final_images folder
The neural networks that are tested here modified versions of the state-of-the-art models and techniques as the method I used to improve the GAN models is by research papers and implementing the different techniques
Feel free to clone this repository and run or modify the jupyter notebook to play around with the Generative Adversarial Networks
The training of Generative Adversarial Networks are very unstable as GANs by nature of the training process contains a lot of failure modes and therefore, when re-running or playing around with the notebook, the generated pictures may be different than that of the original set of generated images