Every time I ask gpt4o, it gives the wrong answer (M).
But when I change the prompt to Chinese, it always gives the correct answer (M+1). (Just translate to Chinese, ask, and then translate the result back to English).
I'm wondering if there are any papers discussing this kind of inconsistency between languages.
No comments yet.