Skip to content
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

Can handle non-English text with bonito ? #44

Open
zhanglt opened this issue Jan 10, 2025 · 5 comments
Open

Can handle non-English text with bonito ? #44

zhanglt opened this issue Jan 10, 2025 · 5 comments

Comments

@zhanglt
Copy link

zhanglt commented Jan 10, 2025

I found bontio only chandle English context。
thanks!

@nanyyyyyy
Copy link
Collaborator

Thank you for your interest in our work. But, Bonito doesn't support languages besides English. thanks

@zhanglt
Copy link
Author

zhanglt commented Feb 14, 2025

I tried to train bointo to make it support Chinese and acceptable.
kitsdk/bonito-chinese-v1

@nihalnayak nihalnayak reopened this Feb 14, 2025
@nihalnayak
Copy link
Member

Hi @zhanglt

This is awesome! Could you briefly link the dataset/recipe you used to fine-tune this new model on the huggingface repo card?

We'd be happy to list this model on the repo. If we see more interest, we'd even fine-tune a chinese-variant of Bonito.

@zhanglt
Copy link
Author

zhanglt commented Feb 16, 2025

Sorry, I'm just a rookie.
At first, I wanted to fine-tune bonito-v1 with the Chinese data collection to make bonito-vi support Chinese, but the effect was not very ideal.
Later, I simply switched to the Chinese basic model (Qwen2.5), using the original training code of bonito, and using the Chinese data collection in ctga format for rebuilding bonito. It initially seems that the effect is better than the first way。

@nihalnayak
Copy link
Member

That's great! The second approach makes more sense. Could you upload the Chinese CTGA dataset on Hugging Face? It would be really helpful for the community in general.

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

No branches or pull requests

3 participants