• XiELEd@piefed.social
    link
    fedilink
    English
    arrow-up
    4
    arrow-down
    1
    ·
    2 days ago

    Man, I’m pretty sure Local LLMs getting better and more efficient would be one of the reasons why the bubble’s gonna burst. It’s still very technical, though.

    • mic_check_one_two@lemmy.dbzer0.com
      link
      fedilink
      English
      arrow-up
      1
      ·
      20 hours ago

      That’s part of why DeepSeek is so worrying to these American companies. It’s much more efficient and doesn’t rely on Nvidia cards, so American companies see it as a massive needle inching towards the over-inflated bubble.

    • greyscale@lemmy.grey.ooo
      link
      fedilink
      English
      arrow-up
      6
      ·
      2 days ago

      I mean seriously, just try ollama with lfm2 on a macbook with the current NPU… Why the fuck would I pay these clowns to run that process elsewhere? Why would I pay for model access when they’ve plateaued and open/free models are everywhere?

      • mic_check_one_two@lemmy.dbzer0.com
        link
        fedilink
        English
        arrow-up
        1
        ·
        20 hours ago

        That’s part of why DeepSeek is so worrying to these American companies. It’s much more efficient and doesn’t rely on Nvidia cards, so American companies see it as a massive needle inching towards the over-inflated bubble.

      • mic_check_one_two@lemmy.dbzer0.com
        link
        fedilink
        English
        arrow-up
        1
        ·
        20 hours ago

        That’s part of why DeepSeek is so worrying to these American companies. It’s much more efficient and doesn’t rely on Nvidia cards, so American companies see it as a massive needle inching towards the over-inflated bubble.