An Adversarial Review of “Adversarial Generation of Natural Language”
Yoav Goldberg
1.3K32

In general, I don’t want to participate in public holy wars, so I will make just two points. First, good job criticizing the Chinese poetry evaluation, I have no idea how the paper that proposed it made it to AAAI. Second, you might be wrong when you criticize the method. Wasserstein distance changes the original GAN idea quite a bit, and simply separating one-hot from not-one-hot will not make the discriminator from W-GAN entirely happy and saturated.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.