Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

why encoder input don't have start token <S> #64

Open
Satan012 opened this issue Dec 17, 2018 · 3 comments
Open

why encoder input don't have start token <S> #64

Satan012 opened this issue Dec 17, 2018 · 3 comments

Comments

@Satan012
Copy link

No description provided.

@Satan012 Satan012 changed the title why encoder input don't have start <S> why encoder input don't have start token <S> Dec 17, 2018
@TSLNIHAOGIT
Copy link

I think encoder shouldn't have start token and end token ,only decoder need start token and end token

@DaoD
Copy link

DaoD commented Mar 11, 2019

The start token in decoder is used to generate the first word in response.
There is no need to use a start token in encoder, since the encoder always receive input in both training and prediction process.

@MingLunHan
Copy link

MingLunHan commented May 16, 2019

Because encoder just extracts information from source language sentence, but when decoder wants to generate a target language sentence, it needs an signal to tell itself to start decoding with some compact representation of collected information, this signal is start token .

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

4 participants