Tokenize charlm #1511
Merged
Tokenize charlm #1511
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
AngledLuffa
force-pushed
the
tokenize_charlm
branch
from
fdd4bd1 to
15d191c
Compare
…rms of improving sentence splitting scores for certain languages. Also helps with MWT scores for Hebrew, not significantly tested on other languages yet Save & load the tokenizers without putting a charlm (if relevant) into the model file Pass the charlm_forward_file from the pipeline to the model When training a tokenizer, the run_tokenizer script finds a charlm, if possible, and attaches it to the model Ignore extra charlm when passed in from the run_ script or as part of the Pipeline if the saved model didn't use charlm
AngledLuffa
force-pushed
the
tokenize_charlm
branch
from
13998e0 to
c276e20
Compare
…ebuilding resources.json, if such a thing exists
…ve the necessary endpoints
…ing of the tokenizer... need to account for that when breaking down MWT
… and treats them as separate words, teaching the tokenizer to not treat 'can not' as a single token with the space This wound up being a problem in Hebrew when using the new charlm-attached tokenizer, although on further reflection there's no reason it couldn't happen in any language or with non-pretrained-charlm tokenizers as well
AngledLuffa
force-pushed
the
tokenize_charlm
branch
from
c276e20 to
52cea78
Compare
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Outdated suggestions cannot be applied.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.
Attaches the charlm to the tokenizer. Seems to reduce the Hebrew MWT error rate (although weirdly it sometimes grabs spaces in tokens now)