You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We are following the Fine_tune_UDOP_on_a_custom_dataset_(toy_RVL_CDIP_dataset).ipynb notebook example.
We used OCR text and coordinates based on CJK (Chinese, Japanese, Korean).
However, it seems that UDOPTokenizer does not support CJK.
Can you provide a guide or notebook code to change to the LayoutXLMTokenizer instead of the UDOPTokenizer?
The text was updated successfully, but these errors were encountered:
Hi @NielsRogge
We are following the Fine_tune_UDOP_on_a_custom_dataset_(toy_RVL_CDIP_dataset).ipynb notebook example.
We used OCR text and coordinates based on CJK (Chinese, Japanese, Korean).
However, it seems that UDOPTokenizer does not support CJK.
Can you provide a guide or notebook code to change to the LayoutXLMTokenizer instead of the UDOPTokenizer?
The text was updated successfully, but these errors were encountered: