Skip to content

Choose the augmentation to perform based on script information #1172

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

Draft
wants to merge 5 commits into
base: main
Choose a base branch
from

Conversation

gregtatum
Copy link
Member

This is a bit bigger work compared to some of the other PRs. The first commit is a bit off topic, but includes some fixes that were hard to rebase around, so I left them in there with this commit.

This work generalizes the CJK and Arabic script changes to use script information rather than specific language checks for what type of augmentations can be run.

The file pipeline/data/lang_script.py was built with a combination of CLDR information and Wikipedia sleuthing to build up a big list of potential scripts. It's a bit overkill as we'll most likely never use a lot of the scripts, but I just got a list from CLDR and worked through it.

The opustrainer configs are now built dynamically based on the script features.

@gregtatum gregtatum requested review from a team as code owners July 2, 2025 14:58
@gregtatum gregtatum requested a review from hneiva July 2, 2025 14:58
@gregtatum gregtatum marked this pull request as draft July 2, 2025 15:06
@gregtatum gregtatum removed request for a team and hneiva July 2, 2025 15:06
@gregtatum
Copy link
Member Author

I'm moving this to draft because I forgot about the libicu errors I needed to sort out. I hacked around this in my release branch.

@evgenyrp
Copy link
Collaborator

evgenyrp commented Jul 7, 2025

I suggest we merge this first, then I can redistill Arabic.

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

Successfully merging this pull request may close these issues.

2 participants