On the "steerability" of generative adversarial networks

Ali Jahanian, Lucy Chai, Phillip Isola

An open secret in contemporary machine learning is that many models work beautifully on standard benchmarks but fail to generalize outside the lab. This has been attributed to bias…