News for hackers. Stuff that matters.
Re:LLaMA Now Goes Faster on CPUs by dragonwriter on Monday April 01, 2024 @01:44PM (#39896840)
Language models are an inefficient way to store knowledge; if you want to have a “pseudo-backup of a large chunk of human knowledge,” download a wikipedia dump, not an LLM.

If you want a friendly but fallible UI to that dump, download an LLM and build a simple ReAct framework around it with prompting to use the wikipedia dump for reference.

There are no comments on this comment yet.