Training with custom arabic data

[alifaqi]

Hello, thank you for this great model.
I have a question, suppose I have a custom annotated Arabic dataset, do you think GLINER can be trained with it?
If yes, should i train it from scratch or fine-tune the existing model?

Thanks

[urchade]

Sure! Some people have done version for language such as japanese or Korean, on hugging face

[alifaqi]

Thank you for your answer.
Do you think it is better if we train it from scratch or it will be okay to just fine-tune the pre-trained model?

[urchade]

It depends on the size of your data,
If it is > 20k you can train from scratch