Try with Hindi langauge

[Nikhi001]

Hello,

Your model is good with the English language; however, I tried your repository with the Hindi dataset. The dataset I used is available at the link given below.

Link :- https://huggingface.co/datasets/c3rl/IIIT-INDIC-HW-WORDS-Hindi/viewer/default/train

I need your assistance on fine-tuning your model for the Hindi language, or can you please share with me the documentation? If you know any other researcher or team that is doing the same project, kindly share it with me

[ankanbhunia]

The method can be extended to any script with the right dataset.

But It seems that the above dataset does not contain the writer ids associated with the handwritings. Without those you can not train the style mimicry component of the architecture.