Take a language that is equidistant from english and chinese, like Malay, and translate that dataset into both languages. Then train a model and benchmark them.

Take a language that is equidistant from english and chinese, like Malay, and translate that dataset into both languages. Then train a model and benchmark them. I suspect the differences would be irrelevant.