It can do that. Verifying an answer is just another algorithm it can learn. LLMs mostly can't do math but that, like most of their other flaws, is because of t

It can do that. Verifying an answer is just another algorithm it can learn. LLMs mostly can't do math but that, like most of their other flaws, is because of the tokenizer.