3 ms·
How well do these papers replicate? In some of experiments with GPT-4 I've seen chain-of-thought style prompting make answers noticeably worse than plainly aski
by jstx1 3y ago
How well do these papers replicate? In some of experiments with GPT-4 I've seen chain-of-thought style prompting make answers noticeably worse than plainly asking a question once.
- rmbyrro 3y agoReally? Do you have any examples to share? I'd be surprised to see that in action.