Vector Quantised Variational Autoencoder

AbstractGenerating inferential texts about an event in different perspectives requires reasoning over different contexts that the event occurs. Existing works usually ignore the context that is not explicitly provided, resulting …...