That's like testing a human by asking him or her to calculate large numbers. The LLMs are known to have issues with individual letters due to the underlying to

That's like testing a human by asking him or her to calculate large numbers. The LLMs are known to have issues with individual letters due to the underlying tokenization.