This repository was archived by the owner on Jul 7, 2023. It is now read-only.
Storing encoder-decoder attention history at Transformer's cache during fast decoding. #1602
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
Thanks for correcting this! I had to tweak a few things, but hopefully it all works now. Thanks again! |
tensorflow-copybara
pushed a commit
that referenced
this pull request
Jun 20, 2019
PiperOrigin-RevId: 254119136
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Sometimes, attention weights may be useful for explaining model's results, visualization, etc. At speech recognition, encoder-decoder attention may be used for time alignment of waveform with recognized phonemes. In old releases, or without fast decoding, the attention is stored at
attention_weights, but it has been broken at fast decoding mode: #898 , #917 .This PR returns weights in
cache["attention_history"], works in both Eager and Graph modes.