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

Expose alignment history after decoding #2

Closed
guillaumekln opened this Issue Nov 2, 2017 · 1 comment

Comments

Projects
None yet
1 participant
@guillaumekln
Member

guillaumekln commented Nov 2, 2017

The dynamic_decode methods should return the alignment history from the attention model.

This feature is currently blocked by tensorflow/tensorflow#13154.

@guillaumekln

This comment has been minimized.

Show comment
Hide comment
@guillaumekln

guillaumekln Mar 31, 2018

Member

Update on this:

  • For RNN decoding, the BeamSearchDecoder in TensorFlow has finally be updated to support storing the alignment history during decoding. We can check this in when it is released in the next version.
  • For self-attention decoding, it is still unclear how to return a single alignment head. Suggestions for that are welcome.
Member

guillaumekln commented Mar 31, 2018

Update on this:

  • For RNN decoding, the BeamSearchDecoder in TensorFlow has finally be updated to support storing the alignment history during decoding. We can check this in when it is released in the next version.
  • For self-attention decoding, it is still unclear how to return a single alignment head. Suggestions for that are welcome.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment