Skip to content

Comment on Smaller, faster, safer: running Kimi and GLM at scaleparent

Comments

"What's scary, to me, as a dev using AI, is that those LLMs do the same thing with code: it looks like proper code, but it really ain't so once you dig a bit."

Of course it always looks better, than it is, but it still works for me as in the code does what I asked the agent to.

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.