DiffuSeq
DiffuSeq copied to clipboard
rounding issue
感谢作者的工作,请问代码中rounding部分的操作的解释在论文中有体现吗,怎么理解这个rounding操作后的结果就是word的词序号?
Rounding operation maps the word embedding vectors back to discrete tokens. and we still map these tokens into vectors as the input of next-step generation. This operation makes sure that at each step, we use the vectors standing for word tokens instead of vectors without corresponding tokens. We discuss this in Eq.7-Eq.8.