README.md

vanilla-gan

Have fun playing with a vanilla gan (tensorflow based).

Startup

You can run main.py directly to view some results, or look into the vanilla_gan.py to observe more details. Either CNN or DNN is supported, use model='cnn' or model='dnn' to shift from each other. When use 'cnn', running on GPU is strongly recommended.

Result Images

You can refer to vanilla-gan\result_images to observe the performances of this vanilla gan.

Reference

• NIPS 2016 Tutorial: Generative Adversarial Networks
• assignment 3, cs231n, Stanford
• Least Squares Generative Adversarial Networks