Glad they set a time rather than leaving it open

  • Domi@lemmy.secnd.me
    link
    fedilink
    English
    arrow-up
    3
    ·
    3 days ago

    27b is much better though…

    And it still gets 15-20 t/s with MTP which is plenty fast.

    • e0qdk@reddthat.com
      link
      fedilink
      English
      arrow-up
      2
      ·
      2 days ago

      27b is much better though…

      I haven’t really found that to be the case for my uses, but getting much faster decode is really useful for me… Have you tried running the MoE model at a higher quant? I usually use Q6_K.

      • Domi@lemmy.secnd.me
        link
        fedilink
        English
        arrow-up
        2
        ·
        2 days ago

        I run both of them at Q8, unquantized KV cache and same settings.

        35b does really well at most tasks but once the dependencies get complicated and the tool calls reach into the hundreds, 35b struggles where 27b still excels.

        So generally I use 27b unless I need it fast, then I swap to 35b or 122b depending on what I need.