Was BAAI/bge-large-zh-v1.5 trained on pure simplified Chinese?

#3
by phamvantoan - opened

Hi,

I'd like to know if BAAI/bge-large-zh-v1.5 can work with Traditional Chinese?

If yes, can I know how many percentages of traditional chinese, bge-large-zh-v1.5 was trained on? Also, how many percentages for simplified chinese?

Beijing Academy of Artificial Intelligence org

Hi, since the training data are all in simplified Chinese, BAAI/bge-large-zh-v1.5 cannot performers well on tasks in traditional chinese.

Hi,

Thank you for your clear answer!

so did you guys meet same question of it decoded out with some transitional Chinese characters when using bge-zh-large? more over the bge-large-zh-v1.5 has removed the traditional Chinese parts ? how much with the percentage of traditional Chinese parts takes?

Sign up or log in to comment