TF-GAN, A lightweight GAN library for TensorFlow 2.0

A newer version of TF-GAN (a lightweight library for training and evaluating Generative Adversarial Networks) was announced recently. TF-GAN has been used in a number of projects and papers 


Updated Features:

Cloud TPU support: TF-GANs can be used to train GAN’s Google Cloud TPU’s.

Self-study GAN course: Open source self study GAN courses based on internal Google study materials.

PyPi package: TF-GAN can be installed with ‘pip install tensorflow-gan’ and used with ‘import tensorflow_gan as tfgan’.

GAN metrics: TF-GAN has easier metrics to compare results from papers. TF-GAN metrics are computationally-efficient and syntactically easy.

GitHub Repository: It has its own Github repository and can be accessed easily.

Colaboratory tutorials: TF-GAN can now used be used with Google GPU’s and TPU’s.

TensorFlow 2.0: TF-GAN is now compatible with TensorFlow 2.

Image source:

Projects using TF-GAN

In terms of projects, here are some below:

Self-Attention GAN on Cloud TPUs [Paper]

Image Extension [Paper]

BigGAN [Paper]

GANSynth [Source]


Previous articleCerebras reveals AI Super computer on a chip for DeepLearning
Next articleIs The Gig Economy Over?
Asif Razzaq is an AI Journalist and Cofounder of Marktechpost, LLC. He is a visionary, entrepreneur and engineer who aspires to use the power of Artificial Intelligence for good. Asif's latest venture is the development of an Artificial Intelligence Media Platform (Marktechpost) that will revolutionize how people can find relevant news related to Artificial Intelligence, Data Science and Machine Learning. Asif was featured by Onalytica in it’s ‘Who’s Who in AI? (Influential Voices & Brands)’ as one of the 'Influential Journalists in AI' ( His interview was also featured by Onalytica (