I believe this presents a good challenge, as the article mentioned: the problems are unique :) None of them are found in ChatGPT's training dataset.