‹ BackHN Continuity

Thread

Shapelearn Qwen 3.8 27B (13.1 GB VRAM)

104 points · 39 comments · syntaxing

  1. _ache_ · · focus · HN ↗
    From my own test. It's not faster than the unsloth model.

    Disclarer: I'm unsing Vulkan on an AMD GC.

    1. DiabloD3 · · focus · HN ↗
      Surprised its not meaningfully slower.

      Vulkan and ROCm paths are missing a few optimized versions of the quants they're using.

      1. GrayShade · · focus · HN ↗
        I see around 1 tk/s after a while, so perhaps it is.
Open on Hacker News to reply ↗

Unofficial Hacker News client; not affiliated with Y Combinator.