SeqDiffuSeq: Text Diffusion with Encoder-Decoder Transformers
Diffusion model, a new generative modelling paradigm, has achieved great success in image, audio, and video generation. However, considering the discrete categorical nature of text, it is not...