Generative Adversarial Nets [8] were recently introduced as a novel way totrain generative models. In this work we introduce the conditional version ofgenerative adversarial nets, which can be constructed by simply feeding thedata, y, we wish to condition on to both the generator and discriminator. Weshow that this model can generate MNIST digits conditioned on class labels. Wealso illustrate how this model could be used to learn a multi-modal model, andprovide preliminary examples of an application to image tagging in which wedemonstrate how this approach can generate descriptive tags which are not partof training labels.