3 ms·
Tip: Don't point out contradictions to a LLM AI. Just tell it to redo the wrong part. They can not figure out that they are wrong (yet.) They only spit out bes
by RecycledEle 4y ago
Tip: Don't point out contradictions to a LLM AI. Just tell it to redo the wrong part. They can not figure out that they are wrong (yet.) They only spit out best effort attempts.
- tempaccount420 4y agoThis advice doesn't apply to ChatGPT in my experience. It does great either way, probably thanks to great RLHF.