kanyun-inc/fairseq-gec

question about the sentences copy task

wangwang110 opened this issue · 4 comments

First,thank you for your good work,I want to know, in sentence-copy task,how to construct one batch, half of the correct sentence pairs and half of the edited sentence pairs? or one batch only containing the correct sentence pairs and the next batch only containing the edited sentence pairs

and if don‘t use attention, which output will be considered as the input of decoder part?

One batch only containing the correct sentence pairs and the next batch only containing the edited sentence pairs.

There are two attentions in the model. One is the encoder-decoder attention and the other is the copying attention. We only disable the encoder-decoder attention.

There are two attentions in the model. One is the encoder-decoder attention and the other is the copying attention. We only disable the encoder-decoder attention.

ok,I got,thank you very much