Modified beam_search() to work with multi-character tokens #53
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
Beam search was reversing multi-character tokens when reconstructing the decoded sequence from the beam search path. This happened because the decoded string was built by reversing the output of the suffix tree traversal, which assumes each label corresponds to a single character. When using vocabularies with multi-character tokens (e.g. "TH", "XY"), this leads to incorrect outputs like "HT" instead of "TH". This is a minimal patch to ensure that multi-letter labels remain in their original order.
Changes:
search.rsandduplex.rs,removed the.chars().rev()step when building the final decoded sequence from the beam search path. The sequence is now reconstructed directly from the ordered list of labels.