MHard

@MHard@lemmy.world

• ⁨Comment⁩ on ⁨How Much Do LLMs Hallucinate in Document Q&A Scenarios? A 172-Billion-Token Study Across Temperatures, Context Lengths, and Hardware Platforms [TLDR: 25%]⁩ ⁨⁨1⁩ ⁨week⁩ ago⁩:

  The task described in this article is asking questions about a document that was provided to the llm in the context.

  I would hope that if you give a human a text and ask them to cite facts from it they would do better than 99% correct.

  Also, when the tokens exceeded 200k, the llm error rate was higher than 10%

• ⁨Comment⁩ on ⁨‘Americans just work harder’ than Europeans, says CEO of Norway’s $1.6 trillion oil fund, because they have a higher ‘general level of ambition’⁩ ⁨⁨1⁩ ⁨year⁩ ago⁩:

  Unlike other posters, I think this is not wrong. You see a lot of “hustle culture” over there as well as the general belief that if they work hard they can get rich. Whether that’s a good thing or not is another question.