How (not) to Train your Generative Model: Scheduled Sampling, Likelihood, Adversary?

Devs

How (not) to Train your Generative Model: Scheduled Sampling, Likelihood, Adversary? | Read Paper on Bytez