stable-diffusion.cpp icon indicating copy to clipboard operation
stable-diffusion.cpp copied to clipboard
Published 2 months ago • verified publisher icon leejet

Support ternary models

Open diimdeep opened this issue 1 year ago • 0 comments

ggml-quants : ternary packing for TriLMs and BitNet b1.58 #8151 is on the horizon

There are TerDiT: Ternary Diffusion Models

4.2B of ternary weights packed lossless:

  • 2,000 bpw - 1.05GB
  • 1,625 bpw - 914 MB

diimdeep avatar Aug 05 '24 09:08 diimdeep