Hacker News new | ask | show | jobs
by cantor_S_drug 265 days ago
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?