Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

This is the default setting. The true test would be if LLMs CAN'T produce distinct outputs. I think this problem can be solved by prompt engineering. Has anyone tried this with Kimi K2?


Consider applying for YC's Winter 2026 batch! Applications are open till Nov 10

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: