4
Abrahamm3r/Z-Image-SDNQ-uint4-svd-r32 · A 4-bit quantized version of Tongyi-MAI/Z-Image using SDNQ (Structured Decomposable Neural Quantization) (huggingface.co)
submitted 9 months ago by to c/stable_diffusion@lemmy.dbzer0.com
4 comments fedilink hide all child comments
 

You need comfyui-sdnq to load SDNQ quantized models in ComfyUI: https://github.com/EnragedAntelope/comfyui-sdnq

you are viewing a single comment's thread
view the rest of the comments
[–] 3 points 9 months ago (3 children)

What are the tradeoffs? I can't believe a 50% to 75% RAM reduction is free. Is this like quantization?

  • permalink
  • fedilink
  • source
  • hideshow 3 child comments
    •
  • [–] [S] 2 points 9 months ago* (2 children)

    Yeah, it's a different form of quantization. I think it's supposed to look a little better than conventional quants, but it still has all the tradeoffs of a quant.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 2 child comments
    •