Hacker News new | ask | show | jobs
by sunir 252 days ago
Maybe. I use QWAN frequently when working with the coding agents. That requires an llm equivalent of interoception to recognize when the model understanding is scrambled or “aligned with itself” which is what qwan is.