By the way, this type of approach is specifically important in ChatGPT because by forcing the model to plan, validate, and think better, you basically have more chances to get the best version of it.

→ View original post on X — @skirano,