Is there a model version that is not explicitly trained for Chinese text and performs better on en,fr,sp etc.?

[Anaudia]

I've noticed that the model often misinterprets letters with apostrophes, such as "à", as Chinese characters like '口'.

Upon closer examination, I found that the default model used for the server configuration is svtrv2_ch. Meanwhile, the model weights stored on Google Drive are for svtrv2_smtr_gtc_rctc. This makes me wonder whether this model is also specifically trained for Chinese text or if it’s a more general version.